doi
stringlengths 9
10
| chunk-id
stringlengths 1
4
| chunk
stringlengths 1
1.57k
| id
stringlengths 9
10
| title
stringlengths 10
127
| summary
stringlengths 581
1.92k
| source
stringlengths 30
31
| authors
sequence | categories
sequence | comment
stringclasses 28
values | journal_ref
stringclasses 1
value | primary_category
stringclasses 6
values | published
stringlengths 8
8
| updated
stringlengths 8
8
| references
list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1806.01261 | 368 | sha1_base64="gAQ7qdt3IKvK5oBqK3uN1PHYi1k=">AAAB6XicbVA9SwNBEJ2LXzF+RS1tFoNoFe4koIVFwMYyivmA5Ah7m7lkyd7esbsnhCP/wMZCEVv/kZ3/xk1yhSY+GHi8N8PMvCARXBvX/XYKa+sbm1vF7dLO7t7+QfnwqKXjVDFssljEqhNQjYJLbBpuBHYShTQKBLaD8e3Mbz+h0jyWj2aSoB/RoeQhZ9RY6aF13i9X3Ko7B1klXk4qkKPRL3/1BjFLI5SGCap113MT42dUGc4ETku9VGNC2ZgOs | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 369 | 3/1BjFLI5SGCap113MT42dUGc4ETku9VGNC2ZgOsWuppBFqP5tfOiVnVhmQMFa2pCFz9fdERiOtJ1FgOyNqRnrZm4n/ed3UhNd+xmWSGpRssShMBTExmb1NBlwhM2JiCWWK21sJG1FFmbHhlGwI3vLLq6R1WfXcqndfq9Rv8jiKcAKncAEeXEEd7qABTWAQwjO8wpszdl6cd+dj0Vpw8plj+APn8wcRSI0F</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 370 | sha1_base64="gAQ7qdt3IKvK5oBqK3uN1PHYi1k=">AAAB6XicbVA9SwNBEJ2LXzF+RS1tFoNoFe4koIVFwMYyivmA5Ah7m7lkyd7esbsnhCP/wMZCEVv/kZ3/xk1yhSY+GHi8N8PMvCARXBvX/XYKa+sbm1vF7dLO7t7+QfnwqKXjVDFssljEqhNQjYJLbBpuBHYShTQKBLaD8e3Mbz+h0jyWj2aSoB/RoeQhZ9RY6aF13i9X3Ko7B1klXk4qkKPRL3/1BjFLI5SGCap113MT42dUGc4ETku9VGNC2ZgOs | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 371 | 3/1BjFLI5SGCap113MT42dUGc4ETku9VGNC2ZgOsWuppBFqP5tfOiVnVhmQMFa2pCFz9fdERiOtJ1FgOyNqRnrZm4n/ed3UhNd+xmWSGpRssShMBTExmb1NBlwhM2JiCWWK21sJG1FFmbHhlGwI3vLLq6R1WfXcqndfq9Rv8jiKcAKncAEeXEEd7qABTWAQwjO8wpszdl6cd+dj0Vpw8plj+APn8wcRSI0F</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 372 | sha1_base64="gAQ7qdt3IKvK5oBqK3uN1PHYi1k=">AAAB6XicbVA9SwNBEJ2LXzF+RS1tFoNoFe4koIVFwMYyivmA5Ah7m7lkyd7esbsnhCP/wMZCEVv/kZ3/xk1yhSY+GHi8N8PMvCARXBvX/XYKa+sbm1vF7dLO7t7+QfnwqKXjVDFssljEqhNQjYJLbBpuBHYShTQKBLaD8e3Mbz+h0jyWj2aSoB/RoeQhZ9RY6aF13i9X3Ko7B1klXk4qkKPRL3/1BjFLI5SGCap113MT42dUGc4ETku9VGNC2ZgOs | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 373 | 3/1BjFLI5SGCap113MT42dUGc4ETku9VGNC2ZgOsWuppBFqP5tfOiVnVhmQMFa2pCFz9fdERiOtJ1FgOyNqRnrZm4n/ed3UhNd+xmWSGpRssShMBTExmb1NBlwhM2JiCWWK21sJG1FFmbHhlGwI3vLLq6R1WfXcqndfq9Rv8jiKcAKncAEeXEEd7qABTWAQwjO8wpszdl6cd+dj0Vpw8plj+APn8wcRSI0F</latexit> | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 374 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 375 | sha1_base64="m8MJ1M94ujO0d0COo5n2Dsol6rc=">AAAB53icbVC7SgNBFL0bXzG+opY2g0GwCrs2phAM2FhGMA9IFpmdvZsMmZ1dZmaFsKS0sbFQxNZP8Rfs/AZ/wsmj0MQDFw7nnMt9BKng2rjul1NYWV1b3yhulra2d3b3yvsHLZ1kimGTJSJRnYBqFFxi03AjsJMqpHEgsB0MryZ++x6V5om8NaMU/Zj2JY84o8ZKjbtyxa26U5Bl4s1J5fLj+4EAgM1/9sKEZTFKwwTVuuu5qfFzqgxnAselXqYxp | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 376 | 4EAgM1/9sKEZTFKwwTVuuu5qfFzqgxnAselXqYxpWxI+9i1VNIYtZ9P9xyTE6uEJEqULWnIVP3dkdNY61Ec2GRMzUAvehPxP6+bmajm51ymmUHJZoOiTBCTkMnRJOQKmREjSyhT3O5K2IAqyox9Tck+wVs8eZm0zqqeW/Vu3Er9AmYowhEcwyl4cA51uIYGNIFBCI/wDC8Od56cV+dtFi04855D+APn/Qd/sI8A</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 377 | sha1_base64="q1zM5ZsCNMJAtNUZI97B98ATlaA=">AAAB53icbVC7SgNBFL3rM8ZX1FKQwSBYhV0bLQQDNpYJmAckQWZn7yZDZmeXmVkhLCltbCwUsfUvbP0FO79BP8LJo9DEAxcO55zLffiJ4Nq47qezsLi0vLKaW8uvb2xubRd2dus6ThXDGotFrJo+1Si4xJrhRmAzUUgjX2DD71+O/MYtKs1jeW0GCXYi2pU85IwaK1VuCkW35I5B5ok3JcWL96+7g7fqt81/tIOYpRFKwwTVuuW5ielkVBnOBA7z7 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 378 | 6+7g7fqt81/tIOYpRFKwwTVuuW5ielkVBnOBA7z7VRjQlmfdrFlqaQR6k423nNIjqwSkDBWtqQhY/V3R0YjrQeRb5MRNT09643E/7xWasKzTsZlkhqUbDIoTAUxMRkdTQKukBkxsIQyxe2uhPWooszY1+TtE7zZk+dJ/aTkuSWv6hbL5zBBDvbhEI7Bg1MowxVUoAYMAriHR3hyuPPgPDsvk+iCM+3Zgz9wXn8AGkeQ8w==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 379 | sha1_base64="q1zM5ZsCNMJAtNUZI97B98ATlaA=">AAAB53icbVC7SgNBFL3rM8ZX1FKQwSBYhV0bLQQDNpYJmAckQWZn7yZDZmeXmVkhLCltbCwUsfUvbP0FO79BP8LJo9DEAxcO55zLffiJ4Nq47qezsLi0vLKaW8uvb2xubRd2dus6ThXDGotFrJo+1Si4xJrhRmAzUUgjX2DD71+O/MYtKs1jeW0GCXYi2pU85IwaK1VuCkW35I5B5ok3JcWL96+7g7fqt81/tIOYpRFKwwTVuuW5ielkVBnOBA7z7 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 380 | 6+7g7fqt81/tIOYpRFKwwTVuuW5ielkVBnOBA7z7VRjQlmfdrFlqaQR6k423nNIjqwSkDBWtqQhY/V3R0YjrQeRb5MRNT09643E/7xWasKzTsZlkhqUbDIoTAUxMRkdTQKukBkxsIQyxe2uhPWooszY1+TtE7zZk+dJ/aTkuSWv6hbL5zBBDvbhEI7Bg1MowxVUoAYMAriHR3hyuPPgPDsvk+iCM+3Zgz9wXn8AGkeQ8w==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 381 | sha1_base64="ioxb3woZF1oAlTScqds23PrgiMY=">AAAB53icbVC7SgNBFL3rM8ZX1NJmMAhWYdZGC4uAjWUE84BkkdnZ2WTI7Owyc1cIS37AxkIRW3/Jzr9xkmyhiQcGDuecy9x7wkxJi5R+e2vrG5tb25Wd6u7e/sFh7ei4Y9PccNHmqUpNL2RWKKlFGyUq0cuMYEmoRDcc38787pMwVqb6ASeZCBI21DKWnKGTWo+1Om3QOcgq8UtShxIu/zWIUp4nQiNXzNq+TzMMCmZQciWm1UFuRcb4mA1F31HNE | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 382 | p4nQiNXzNq+TzMMCmZQciWm1UFuRcb4mA1F31HNEmGDYr7nlJw7JSJxatzTSObq74mCJdZOktAlE4Yju+zNxP+8fo7xdVBIneUoNF98FOeKYEpmR5NIGsFRTRxh3Ei3K+EjZhhHV03VleAvn7xKOpcNnzb8e1pv3pR1VOAUzuACfLiCJtxBC9rAIYJneIU3T3ov3rv3sYiueeXMCfyB9/kDCGmMcA==</latexit> v | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 383 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 384 | sha1_base64="HCiXjOq04H3f4Ed7vqyiRfd+2dI=">AAAB7XicbVA9SwNBEJ2LXzF+nQo2NotBsAp3NlpYBGwsI5hLIDnj3mYvWbO3e+zuRcKR/2BjoYit/8fOf+Pmo9DEBwOP92aYmRelnGnjed9OYWV1bX2juFna2t7Z3XP3DwItM0VonUguVTPCmnImaN0ww2kzVRQnEaeNaHA98RtDqjST4s6MUhomuCdYzAg2VgraaZ/dDztu2at4U6Bl4s9JuXoUPD0AQK3jfrW7kmQJFYZwrHXL91IT5lgZRjgdl | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 385 | XoUPD0AQK3jfrW7kmQJFYZwrHXL91IT5lgZRjgdl9qZpikmA9yjLUsFTqgO8+m1Y3RqlS6KpbIlDJqqvydynGg9SiLbmWDT14veRPzPa2UmvgxzJtLMUEFmi+KMIyPR5HXUZYoSw0eWYKKYvRWRPlaYGBtQyYbgL768TILziu9V/FubxhXMUIRjOIEz8OECqnADNagDgUd4hld4c6Tz4rw7H7PWgjOfOYQ/cD5/AKnSkKA=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 386 | sha1_base64="/voSHBGyFE5xYPVKPYM/GTIarLQ=">AAAB7XicbVC7SgNBFL3rM8ZXVLCxWQyCVdi10cIiYGMZwWwCyRpmJ7PJmNmZYWY2siz5BxsLRbT0f+z8AP/DyaPQxAMXDufcy733RJJRbTzvy1laXlldWy9sFDe3tnd2S3v7gRapwqSOBROqGSFNGOWkbqhhpCkVQUnESCMaXI39xpAoTQW/NZkkYYJ6nMYUI2OloC379G7YKZW9ijeBu0j8GSlXD4MHmX2/1zqlz3ZX4DQh3GCGtG75njRhjpShm | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 387 | SlXD4MHmX2/1zqlz3ZX4DQh3GCGtG75njRhjpShmJFRsZ1qIhEeoB5pWcpRQnSYT64duSdW6bqxULa4cSfq74kcJVpnSWQ7E2T6et4bi/95rdTEF2FOuUwN4Xi6KE6Za4Q7ft3tUkWwYZklCCtqb3VxHymEjQ2oaEPw519eJMFZxfcq/o1N4xKmKMARHMMp+HAOVbiGGtQBwz08wjO8OMJ5cl6dt2nrkjObOYA/cD5+AGdXkq0=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 388 | sha1_base64="/voSHBGyFE5xYPVKPYM/GTIarLQ=">AAAB7XicbVC7SgNBFL3rM8ZXVLCxWQyCVdi10cIiYGMZwWwCyRpmJ7PJmNmZYWY2siz5BxsLRbT0f+z8AP/DyaPQxAMXDufcy733RJJRbTzvy1laXlldWy9sFDe3tnd2S3v7gRapwqSOBROqGSFNGOWkbqhhpCkVQUnESCMaXI39xpAoTQW/NZkkYYJ6nMYUI2OloC379G7YKZW9ijeBu0j8GSlXD4MHmX2/1zqlz3ZX4DQh3GCGtG75njRhjpShm | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 389 | SlXD4MHmX2/1zqlz3ZX4DQh3GCGtG75njRhjpShmJFRsZ1qIhEeoB5pWcpRQnSYT64duSdW6bqxULa4cSfq74kcJVpnSWQ7E2T6et4bi/95rdTEF2FOuUwN4Xi6KE6Za4Q7ft3tUkWwYZklCCtqb3VxHymEjQ2oaEPw519eJMFZxfcq/o1N4xKmKMARHMMp+HAOVbiGGtQBwz08wjO8OMJ5cl6dt2nrkjObOYA/cD5+AGdXkq0=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 390 | sha1_base64="Fc8T4ygtia14k1z/CDji4ezWDqY=">AAAB7XicbVA9SwNBEJ3zM8avqKXNYhCswp2NFhYBG8sI5gOSM+xtNsmavd1jdy4QjvwHGwtFbP0/dv4bN8kVmvhg4PHeDDPzokQKi77/7a2tb2xubRd2irt7+weHpaPjhtWpYbzOtNSmFVHLpVC8jgIlbyWG0ziSvBmNbmd+c8yNFVo94CThYUwHSvQFo+ikRicZisdxt1T2K/4cZJUEOSlDjlq39NXpaZbGXCGT1Np24CcYZtSgYJJPi53U8oSyE | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 391 | lq39NXpaZbGXCGT1Np24CcYZtSgYJJPi53U8oSyER3wtqOKxtyG2fzaKTl3So/0tXGlkMzV3xMZja2dxJHrjCkO7bI3E//z2in2r8NMqCRFrthiUT+VBDWZvU56wnCGcuIIZUa4WwkbUkMZuoCKLoRg+eVV0risBH4luPfL1Zs8jgKcwhlcQABXUIU7qEEdGDzBM7zCm6e9F+/d+1i0rnn5zAn8gff5A59Hjx0=</latexit>⇢e!v | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 392 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 393 | sha1_base64="s3/Cw/iD/Ic9TAit26LWmPV1hK0=">AAAB/nicbVBNS8NAEN3Ur1q/ouLJy2IRPJVEBD0WvXisYD+giWWznTRLN9mwu6mUUPCvePGgiFd/hzf/jds2B219MPB4b4aZeUHKmdKO822VVlbX1jfKm5Wt7Z3dPXv/oKVEJik0qeBCdgKigLMEmpppDp1UAokDDu1geDP12yOQionkXo9T8GMySFjIKNFG6tlHnozEQw6eZINIEynFIx5NenbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqP | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 394 | nbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqPydSM8phUvEyBSmhQzKArqEJiUH5+ez8CT41Sh+HQppKNJ6pvydyEis1jgPTGRMdqUVvKv7ndTMdXvk5S9JMQ0Lni8KMYy3wNAvcZxKo5mNDCJXM3IppRCSh2iRWMSG4iy8vk9Z5zXVq7t1FtX5dxFFGx+gEnSEXXaI6ukUN1EQU5egZvaI368l6sd6tj3lrySpmDtEfWJ8/7hmWGA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 395 | sha1_base64="s3/Cw/iD/Ic9TAit26LWmPV1hK0=">AAAB/nicbVBNS8NAEN3Ur1q/ouLJy2IRPJVEBD0WvXisYD+giWWznTRLN9mwu6mUUPCvePGgiFd/hzf/jds2B219MPB4b4aZeUHKmdKO822VVlbX1jfKm5Wt7Z3dPXv/oKVEJik0qeBCdgKigLMEmpppDp1UAokDDu1geDP12yOQionkXo9T8GMySFjIKNFG6tlHnozEQw6eZINIEynFIx5NenbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqP | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 396 | nbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqPydSM8phUvEyBSmhQzKArqEJiUH5+ez8CT41Sh+HQppKNJ6pvydyEis1jgPTGRMdqUVvKv7ndTMdXvk5S9JMQ0Lni8KMYy3wNAvcZxKo5mNDCJXM3IppRCSh2iRWMSG4iy8vk9Z5zXVq7t1FtX5dxFFGx+gEnSEXXaI6ukUN1EQU5egZvaI368l6sd6tj3lrySpmDtEfWJ8/7hmWGA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 397 | sha1_base64="s3/Cw/iD/Ic9TAit26LWmPV1hK0=">AAAB/nicbVBNS8NAEN3Ur1q/ouLJy2IRPJVEBD0WvXisYD+giWWznTRLN9mwu6mUUPCvePGgiFd/hzf/jds2B219MPB4b4aZeUHKmdKO822VVlbX1jfKm5Wt7Z3dPXv/oKVEJik0qeBCdgKigLMEmpppDp1UAokDDu1geDP12yOQionkXo9T8GMySFjIKNFG6tlHnozEQw6eZINIEynFIx5NenbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqP | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 398 | nbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqPydSM8phUvEyBSmhQzKArqEJiUH5+ez8CT41Sh+HQppKNJ6pvydyEis1jgPTGRMdqUVvKv7ndTMdXvk5S9JMQ0Lni8KMYy3wNAvcZxKo5mNDCJXM3IppRCSh2iRWMSG4iy8vk9Z5zXVq7t1FtX5dxFFGx+gEnSEXXaI6ukUN1EQU5egZvaI368l6sd6tj3lrySpmDtEfWJ8/7hmWGA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 399 | sha1_base64="s3/Cw/iD/Ic9TAit26LWmPV1hK0=">AAAB/nicbVBNS8NAEN3Ur1q/ouLJy2IRPJVEBD0WvXisYD+giWWznTRLN9mwu6mUUPCvePGgiFd/hzf/jds2B219MPB4b4aZeUHKmdKO822VVlbX1jfKm5Wt7Z3dPXv/oKVEJik0qeBCdgKigLMEmpppDp1UAokDDu1geDP12yOQionkXo9T8GMySFjIKNFG6tlHnozEQw6eZINIEynFIx5NenbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqP | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 400 | nbVqTkz4GXiFqSKCjR69pfXFzSLIdGUE6W6rpNqPydSM8phUvEyBSmhQzKArqEJiUH5+ez8CT41Sh+HQppKNJ6pvydyEis1jgPTGRMdqUVvKv7ndTMdXvk5S9JMQ0Lni8KMYy3wNAvcZxKo5mNDCJXM3IppRCSh2iRWMSG4iy8vk9Z5zXVq7t1FtX5dxFFGx+gEnSEXXaI6ukUN1EQU5egZvaI368l6sd6tj3lrySpmDtEfWJ8/7hmWGA==</latexit> e | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 401 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 402 | sha1_base64="gRKFy+QFytmwqWy0cvo5FmmPz8I=">AAAB7XicbVA9SwNBEJ3zM8avqGBjsxgEq3Bno4VFwMYygrkEkjPubeaSNXu3x+6eEo78BxsLRWz9P3b+GzcfhSY+GHi8N8PMvDAVXBvX/XaWlldW19YLG8XNre2d3dLevq9lphjWmRRSNUOqUfAE64Ybgc1UIY1DgY1wcDX2G4+oNJfJrRmmGMS0l/CIM2qs5LfTPr/DTqnsVtwJyCLxZqRcPfSf7gGg1il9tbuSZTEmhgmqdctzUxPkVBnOBI6K7 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 403 | fSf7gGg1il9tbuSZTEmhgmqdctzUxPkVBnOBI6K7UxjStmA9rBlaUJj1EE+uXZETqzSJZFUthJDJurviZzGWg/j0HbG1PT1vDcW//NamYkugpwnaWYwYdNFUSaIkWT8OulyhcyIoSWUKW5vJaxPFWXGBlS0IXjzLy8S/6ziuRXvxqZxCVMU4AiO4RQ8OIcqXEMN6sDgAZ7hFd4c6bw4787HtHXJmc0cwB84nz+QDpCP</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 404 | sha1_base64="74MJShuZzxGyM2nLY3EY87InuhI=">AAAB7XicbVC7SgNBFJ2NrxhfUcHGZjAIVmHXRguLgI1lBLMJJGuYndxNxszODjOzyrLkH2wsFNHS/7HzA/wPJ49CEw9cOJxzL/feE0rOtHHdL6ewtLyyulZcL21sbm3vlHf3fJ2kikKDJjxRrZBo4ExAwzDDoSUVkDjk0AyHl2O/eQ9Ks0TcmExCEJO+YBGjxFjJ78gBu4VuueJW3QnwIvFmpFI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRD | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 405 | FI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRDqNSJ9UgCR2SPrQtFSQGHeSTa0f42Co9HCXKljB4ov6eyEmsdRaHtjMmZqDnvbH4n9dOTXQe5EzI1ICg00VRyrFJ8Ph13GMKqOGZJYQqZm/FdEAUocYGVLIhePMvLxL/tOq5Ve/apnGBpiiiQ3SETpCHzlANXaE6aiCK7tAjekYvTuI8Oa/O27S14Mxm9tEfOB8/TZOSnA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 406 | sha1_base64="74MJShuZzxGyM2nLY3EY87InuhI=">AAAB7XicbVC7SgNBFJ2NrxhfUcHGZjAIVmHXRguLgI1lBLMJJGuYndxNxszODjOzyrLkH2wsFNHS/7HzA/wPJ49CEw9cOJxzL/feE0rOtHHdL6ewtLyyulZcL21sbm3vlHf3fJ2kikKDJjxRrZBo4ExAwzDDoSUVkDjk0AyHl2O/eQ9Ks0TcmExCEJO+YBGjxFjJ78gBu4VuueJW3QnwIvFmpFI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRD | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 407 | FI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRDqNSJ9UgCR2SPrQtFSQGHeSTa0f42Co9HCXKljB4ov6eyEmsdRaHtjMmZqDnvbH4n9dOTXQe5EzI1ICg00VRyrFJ8Ph13GMKqOGZJYQqZm/FdEAUocYGVLIhePMvLxL/tOq5Ve/apnGBpiiiQ3SETpCHzlANXaE6aiCK7tAjekYvTuI8Oa/O27S14Mxm9tEfOB8/TZOSnA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 408 | sha1_base64="pLq6KB/1S9uyUeWp/G4byg43mK0=">AAAB7XicbVA9SwNBEJ2LXzF+RS1tFoNgFe5stLAI2FhGMB+QnGFvM5es2ds9dveEEPIfbCwUsfX/2Plv3CRXaOKDgcd7M8zMi1LBjfX9b6+wtr6xuVXcLu3s7u0flA+PmkZlmmGDKaF0O6IGBZfYsNwKbKcaaRIJbEWjm5nfekJtuJL3dpximNCB5DFn1Dqp2U2H/AF75Ypf9ecgqyTISQVy1Hvlr25fsSxBaZmgxnQCP7XhhGrLmcBpqZsZTCkb0 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 409 | Hvlr25fsSxBaZmgxnQCP7XhhGrLmcBpqZsZTCkb0QF2HJU0QRNO5tdOyZlT+iRW2pW0ZK7+npjQxJhxErnOhNqhWfZm4n9eJ7PxVTjhMs0sSrZYFGeCWEVmr5M+18isGDtCmebuVsKGVFNmXUAlF0Kw/PIqaV5UA78a3PmV2nUeRxFO4BTOIYBLqMEt1KEBDB7hGV7hzVPei/fufSxaC14+cwx/4H3+AIWDjww=</latexit>E<latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 410 | sha1_base64="iJ/x8cSgmmYNbMN8WtCvsNrlH/U=">AAAB6HicbVBNS8NAEJ3Ur1q/qh69LBbBU0lE0IOHgggeW7Af0Iay2U7atZtN2N0IJfQXePGgiFd/kjf/jds2B219MPB4b4aZeUEiuDau++0U1tY3NreK26Wd3b39g/LhUUvHqWLYZLGIVSegGgWX2DTcCOwkCmkUCGwH49uZ335CpXksH8wkQT+iQ8lDzqixUuOuX664VXcOskq8nFQgR71f/uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i1 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 411 | uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i11JJI9R+Nj90Ss6sMiBhrGxJQ+bq74mMRlpPosB2RtSM9LI3E//zuqkJr/2MyyQ1KNliUZgKYmIy+5oMuEJmxMQSyhS3txI2oooyY7Mp2RC85ZdXSeui6rlVr3FZqd3kcRThBE7hHDy4ghrcQx2awADhGV7hzXl0Xpx352PRWnDymWP4A+fzB5cfjMM=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 412 | sha1_base64="iJ/x8cSgmmYNbMN8WtCvsNrlH/U=">AAAB6HicbVBNS8NAEJ3Ur1q/qh69LBbBU0lE0IOHgggeW7Af0Iay2U7atZtN2N0IJfQXePGgiFd/kjf/jds2B219MPB4b4aZeUEiuDau++0U1tY3NreK26Wd3b39g/LhUUvHqWLYZLGIVSegGgWX2DTcCOwkCmkUCGwH49uZ335CpXksH8wkQT+iQ8lDzqixUuOuX664VXcOskq8nFQgR71f/uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i1 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 413 | uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i11JJI9R+Nj90Ss6sMiBhrGxJQ+bq74mMRlpPosB2RtSM9LI3E//zuqkJr/2MyyQ1KNliUZgKYmIy+5oMuEJmxMQSyhS3txI2oooyY7Mp2RC85ZdXSeui6rlVr3FZqd3kcRThBE7hHDy4ghrcQx2awADhGV7hzXl0Xpx352PRWnDymWP4A+fzB5cfjMM=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 414 | sha1_base64="iJ/x8cSgmmYNbMN8WtCvsNrlH/U=">AAAB6HicbVBNS8NAEJ3Ur1q/qh69LBbBU0lE0IOHgggeW7Af0Iay2U7atZtN2N0IJfQXePGgiFd/kjf/jds2B219MPB4b4aZeUEiuDau++0U1tY3NreK26Wd3b39g/LhUUvHqWLYZLGIVSegGgWX2DTcCOwkCmkUCGwH49uZ335CpXksH8wkQT+iQ8lDzqixUuOuX664VXcOskq8nFQgR71f/uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i1 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 415 | uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i11JJI9R+Nj90Ss6sMiBhrGxJQ+bq74mMRlpPosB2RtSM9LI3E//zuqkJr/2MyyQ1KNliUZgKYmIy+5oMuEJmxMQSyhS3txI2oooyY7Mp2RC85ZdXSeui6rlVr3FZqd3kcRThBE7hHDy4ghrcQx2awADhGV7hzXl0Xpx352PRWnDymWP4A+fzB5cfjMM=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 416 | sha1_base64="iJ/x8cSgmmYNbMN8WtCvsNrlH/U=">AAAB6HicbVBNS8NAEJ3Ur1q/qh69LBbBU0lE0IOHgggeW7Af0Iay2U7atZtN2N0IJfQXePGgiFd/kjf/jds2B219MPB4b4aZeUEiuDau++0U1tY3NreK26Wd3b39g/LhUUvHqWLYZLGIVSegGgWX2DTcCOwkCmkUCGwH49uZ335CpXksH8wkQT+iQ8lDzqixUuOuX664VXcOskq8nFQgR71f/uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i1 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 417 | uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i11JJI9R+Nj90Ss6sMiBhrGxJQ+bq74mMRlpPosB2RtSM9LI3E//zuqkJr/2MyyQ1KNliUZgKYmIy+5oMuEJmxMQSyhS3txI2oooyY7Mp2RC85ZdXSeui6rlVr3FZqd3kcRThBE7hHDy4ghrcQx2awADhGV7hzXl0Xpx352PRWnDymWP4A+fzB5cfjMM=</latexit>V<latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 418 | sha1_base64="xc4uzoZiBSxZUZkArgltxczS6nM=">AAAB6HicbVA9SwNBEJ2LXzF+RS1tFoNgFe5EiIVFwMYyAfMByRH2NnPJmr29Y3dPCEd+gY2FIrb+JDv/jZvkCk18MPB4b4aZeUEiuDau++0UNja3tneKu6W9/YPDo/LxSVvHqWLYYrGIVTegGgWX2DLcCOwmCmkUCOwEk7u533lCpXksH8w0QT+iI8lDzqixUrM9KFfcqrsAWSdeTiqQozEof/WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 419 | /WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ6mkEWo/Wxw6IxdWGZIwVrakIQv190RGI62nUWA7I2rGetWbi/95vdSEN37GZZIalGy5KEwFMTGZf02GXCEzYmoJZYrbWwkbU0WZsdmUbAje6svrpH1V9dyq17yu1G/zOIpwBudwCR7UoA730IAWMEB4hld4cx6dF+fd+Vi2Fpx85hT+wPn8AbDjjNQ=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 420 | sha1_base64="xc4uzoZiBSxZUZkArgltxczS6nM=">AAAB6HicbVA9SwNBEJ2LXzF+RS1tFoNgFe5EiIVFwMYyAfMByRH2NnPJmr29Y3dPCEd+gY2FIrb+JDv/jZvkCk18MPB4b4aZeUEiuDau++0UNja3tneKu6W9/YPDo/LxSVvHqWLYYrGIVTegGgWX2DLcCOwmCmkUCOwEk7u533lCpXksH8w0QT+iI8lDzqixUrM9KFfcqrsAWSdeTiqQozEof/WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 421 | /WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ6mkEWo/Wxw6IxdWGZIwVrakIQv190RGI62nUWA7I2rGetWbi/95vdSEN37GZZIalGy5KEwFMTGZf02GXCEzYmoJZYrbWwkbU0WZsdmUbAje6svrpH1V9dyq17yu1G/zOIpwBudwCR7UoA730IAWMEB4hld4cx6dF+fd+Vi2Fpx85hT+wPn8AbDjjNQ=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 422 | sha1_base64="xc4uzoZiBSxZUZkArgltxczS6nM=">AAAB6HicbVA9SwNBEJ2LXzF+RS1tFoNgFe5EiIVFwMYyAfMByRH2NnPJmr29Y3dPCEd+gY2FIrb+JDv/jZvkCk18MPB4b4aZeUEiuDau++0UNja3tneKu6W9/YPDo/LxSVvHqWLYYrGIVTegGgWX2DLcCOwmCmkUCOwEk7u533lCpXksH8w0QT+iI8lDzqixUrM9KFfcqrsAWSdeTiqQozEof/WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 423 | /WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ6mkEWo/Wxw6IxdWGZIwVrakIQv190RGI62nUWA7I2rGetWbi/95vdSEN37GZZIalGy5KEwFMTGZf02GXCEzYmoJZYrbWwkbU0WZsdmUbAje6svrpH1V9dyq17yu1G/zOIpwBudwCR7UoA730IAWMEB4hld4cx6dF+fd+Vi2Fpx85hT+wPn8AbDjjNQ=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 424 | sha1_base64="xc4uzoZiBSxZUZkArgltxczS6nM=">AAAB6HicbVA9SwNBEJ2LXzF+RS1tFoNgFe5EiIVFwMYyAfMByRH2NnPJmr29Y3dPCEd+gY2FIrb+JDv/jZvkCk18MPB4b4aZeUEiuDau++0UNja3tneKu6W9/YPDo/LxSVvHqWLYYrGIVTegGgWX2DLcCOwmCmkUCOwEk7u533lCpXksH8w0QT+iI8lDzqixUrM9KFfcqrsAWSdeTiqQozEof/WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 425 | /WHMUsjlIYJqnXPcxPjZ1QZzgTOSv1UY0LZhI6wZ6mkEWo/Wxw6IxdWGZIwVrakIQv190RGI62nUWA7I2rGetWbi/95vdSEN37GZZIalGy5KEwFMTGZf02GXCEzYmoJZYrbWwkbU0WZsdmUbAje6svrpH1V9dyq17yu1G/zOIpwBudwCR7UoA730IAWMEB4hld4cx6dF+fd+Vi2Fpx85hT+wPn8AbDjjNQ=</latexit> | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 426 | (d) Non-local neural network | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 427 | Edge blockNode blockGlobal blocku0 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 428 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 429 | sha1_base64="Z/n1gIms2/ONBt0R58c8NGdBbqU=">AAAB8nicbVDLSsNAFL2pr1pfVZduBovoqiQi6MJFwY3LCvYBbSiT6aQdOpmEmRuhhH6GGxeKuPVr3Pk3TtostPXAwOGce5lzT5BIYdB1v53S2vrG5lZ5u7Kzu7d/UD08aps41Yy3WCxj3Q2o4VIo3kKBkncTzWkUSN4JJne533ni2ohYPeI04X5ER0qEglG0Uq8fURwHYZbOzgfVmlt35yCrxCtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azST | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 430 | CtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azSTw1PKJvQEe9ZqmjEjZ/NI8/ImVWGJIy1fQrJXP29kdHImGkU2Mk8oln2cvE/r5dieONnQiUpcsUWH4WpJBiT/H4yFJozlFNLKNPCZiVsTDVlaFuq2BK85ZNXSfuy7rl17+Gq1rgt6ijDCZzCBXhwDQ24hya0gEEMz/AKbw46L86787EYLTnFzjH8gfP5A1s4kUQ=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 431 | sha1_base64="Z/n1gIms2/ONBt0R58c8NGdBbqU=">AAAB8nicbVDLSsNAFL2pr1pfVZduBovoqiQi6MJFwY3LCvYBbSiT6aQdOpmEmRuhhH6GGxeKuPVr3Pk3TtostPXAwOGce5lzT5BIYdB1v53S2vrG5lZ5u7Kzu7d/UD08aps41Yy3WCxj3Q2o4VIo3kKBkncTzWkUSN4JJne533ni2ohYPeI04X5ER0qEglG0Uq8fURwHYZbOzgfVmlt35yCrxCtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azST | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 432 | CtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azSTw1PKJvQEe9ZqmjEjZ/NI8/ImVWGJIy1fQrJXP29kdHImGkU2Mk8oln2cvE/r5dieONnQiUpcsUWH4WpJBiT/H4yFJozlFNLKNPCZiVsTDVlaFuq2BK85ZNXSfuy7rl17+Gq1rgt6ijDCZzCBXhwDQ24hya0gEEMz/AKbw46L86787EYLTnFzjH8gfP5A1s4kUQ=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 433 | sha1_base64="Z/n1gIms2/ONBt0R58c8NGdBbqU=">AAAB8nicbVDLSsNAFL2pr1pfVZduBovoqiQi6MJFwY3LCvYBbSiT6aQdOpmEmRuhhH6GGxeKuPVr3Pk3TtostPXAwOGce5lzT5BIYdB1v53S2vrG5lZ5u7Kzu7d/UD08aps41Yy3WCxj3Q2o4VIo3kKBkncTzWkUSN4JJne533ni2ohYPeI04X5ER0qEglG0Uq8fURwHYZbOzgfVmlt35yCrxCtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azST | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 434 | CtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azSTw1PKJvQEe9ZqmjEjZ/NI8/ImVWGJIy1fQrJXP29kdHImGkU2Mk8oln2cvE/r5dieONnQiUpcsUWH4WpJBiT/H4yFJozlFNLKNPCZiVsTDVlaFuq2BK85ZNXSfuy7rl17+Gq1rgt6ijDCZzCBXhwDQ24hya0gEEMz/AKbw46L86787EYLTnFzjH8gfP5A1s4kUQ=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 435 | sha1_base64="Z/n1gIms2/ONBt0R58c8NGdBbqU=">AAAB8nicbVDLSsNAFL2pr1pfVZduBovoqiQi6MJFwY3LCvYBbSiT6aQdOpmEmRuhhH6GGxeKuPVr3Pk3TtostPXAwOGce5lzT5BIYdB1v53S2vrG5lZ5u7Kzu7d/UD08aps41Yy3WCxj3Q2o4VIo3kKBkncTzWkUSN4JJne533ni2ohYPeI04X5ER0qEglG0Uq8fURwHYZbOzgfVmlt35yCrxCtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azST | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 436 | CtIDQo0B9Wv/jBmacQVMkmN6Xlugn5GNQom+azSTw1PKJvQEe9ZqmjEjZ/NI8/ImVWGJIy1fQrJXP29kdHImGkU2Mk8oln2cvE/r5dieONnQiUpcsUWH4WpJBiT/H4yFJozlFNLKNPCZiVsTDVlaFuq2BK85ZNXSfuy7rl17+Gq1rgt6ijDCZzCBXhwDQ24hya0gEEMz/AKbw46L86787EYLTnFzjH8gfP5A1s4kUQ=</latexit> u | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 437 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 438 | sha1_base64="znt8hwWv6wryqwCugrweUa+jkM8=">AAAB7XicbVA9SwNBEJ3zM8avqGBjsxgEq3Bno4VFwMYygrkEkjPubfaSNXu7y+6eEo78BxsLRWz9P3b+GzcfhSY+GHi8N8PMvFhxZqzvf3tLyyura+uFjeLm1vbObmlvPzQy04TWieRSN2NsKGeC1i2znDaVpjiNOW3Eg6ux33ik2jApbu1Q0SjFPcESRrB1UthWfXaXdUplv+JPgBZJMCPl6mH4dA8AtU7pq92VJEupsIRjY1qBr2yUY20Z4XRUb | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 439 | mH4dA8AtU7pq92VJEupsIRjY1qBr2yUY20Z4XRUbGeGKkwGuEdbjgqcUhPlk2tH6MQpXZRI7UpYNFF/T+Q4NWaYxq4zxbZv5r2x+J/XymxyEeVMqMxSQaaLkowjK9H4ddRlmhLLh45gopm7FZE+1phYF1DRhRDMv7xIwrNK4FeCG5fGJUxRgCM4hlMI4ByqcA01qAOBB3iGV3jzpPfivXsf09YlbzZzAH/gff4AqE6Qnw==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 440 | sha1_base64="Nc0DXje6uYB+/0fHlXL99yCq0no=">AAAB7XicbVC7SgNBFL0bXzG+ooKNzWAQrMKujRYWARvLCGYTSNYwO5lNxszODjOzyrLkH2wsFNHS/7HzA/wPJ49CEw9cOJxzL/feE0rOtHHdL6ewtLyyulZcL21sbm3vlHf3fJ2kitAGSXiiWiHWlDNBG4YZTltSURyHnDbD4eXYb95TpVkibkwmaRDjvmARI9hYye/IAbtNu+WKW3UnQIvEm5FK7cB/kNn3e71b/uz0EpLGVBjCsdZtz5UmyLEyj | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 441 | 5FK7cB/kNn3e71b/uz0EpLGVBjCsdZtz5UmyLEyjHA6KnVSTSUmQ9ynbUsFjqkO8sm1I3RslR6KEmVLGDRRf0/kONY6i0PbGWMz0PPeWPzPa6cmOg9yJmRqqCDTRVHKkUnQ+HXUY4oSwzNLMFHM3orIACtMjA2oZEPw5l9eJP5p1XOr3rVN4wKmKMIhHMEJeHAGNbiCOjSAwB08wjO8OInz5Lw6b9PWgjOb2Yc/cD5+AGXTkqw=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 442 | sha1_base64="Nc0DXje6uYB+/0fHlXL99yCq0no=">AAAB7XicbVC7SgNBFL0bXzG+ooKNzWAQrMKujRYWARvLCGYTSNYwO5lNxszODjOzyrLkH2wsFNHS/7HzA/wPJ49CEw9cOJxzL/feE0rOtHHdL6ewtLyyulZcL21sbm3vlHf3fJ2kitAGSXiiWiHWlDNBG4YZTltSURyHnDbD4eXYb95TpVkibkwmaRDjvmARI9hYye/IAbtNu+WKW3UnQIvEm5FK7cB/kNn3e71b/uz0EpLGVBjCsdZtz5UmyLEyj | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 443 | 5FK7cB/kNn3e71b/uz0EpLGVBjCsdZtz5UmyLEyjHA6KnVSTSUmQ9ynbUsFjqkO8sm1I3RslR6KEmVLGDRRf0/kONY6i0PbGWMz0PPeWPzPa6cmOg9yJmRqqCDTRVHKkUnQ+HXUY4oSwzNLMFHM3orIACtMjA2oZEPw5l9eJP5p1XOr3rVN4wKmKMIhHMEJeHAGNbiCOjSAwB08wjO8OInz5Lw6b9PWgjOb2Yc/cD5+AGXTkqw=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 444 | sha1_base64="S5XnA5iYIAgqxiI+i0ptSwAiKP4=">AAAB7XicbVA9SwNBEJ2LXzF+RS1tFoNgFe5stLAI2FhGMB+QnGFvM5es2ds9dveEEPIfbCwUsfX/2Plv3CRXaOKDgcd7M8zMi1LBjfX9b6+wtr6xuVXcLu3s7u0flA+PmkZlmmGDKaF0O6IGBZfYsNwKbKcaaRIJbEWjm5nfekJtuJL3dpximNCB5DFn1Dqp2U2H/CHrlSt+1Z+DrJIgJxXIUe+Vv7p9xbIEpWWCGtMJ/NSGE6otZwKnpW5mMKVsR | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 445 | e+Vv7p9xbIEpWWCGtMJ/NSGE6otZwKnpW5mMKVsRAfYcVTSBE04mV87JWdO6ZNYaVfSkrn6e2JCE2PGSeQ6E2qHZtmbif95nczGV+GEyzSzKNliUZwJYhWZvU76XCOzYuwIZZq7WwkbUk2ZdQGVXAjB8surpHlRDfxqcOdXatd5HEU4gVM4hwAuoQa3UIcGMHiEZ3iFN095L96797FoLXj5zDH8gff5A53Djxw=</latexit>⇢e!u | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 446 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 447 | sha1_base64="2suSYs2KtjHJeb1CIts1JrhYPII=">AAAB/nicbVBNS8NAEN34WetXVDx5WSyCp5KIoMeiF48V7Ac0sWy2m3bpZjfsTpQSCv4VLx4U8erv8Oa/cdvmoK0PBh7vzTAzL0oFN+B5387S8srq2nppo7y5tb2z6+7tN43KNGUNqoTS7YgYJrhkDeAgWDvVjCSRYK1oeD3xWw9MG67kHYxSFiakL3nMKQErdd3DQA/Ufc4CzfsDIFqrR5yNu27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shz | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 448 | 27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shzIkGTgUbl4PMsJTQIemzjqWSJMyE+fT8MT6xSg/HStuSgKfq74mcJMaMksh2JgQGZt6biP95nQziyzDnMs2ASTpbFGcCg8KTLHCPa0ZBjCwhVHN7K6YDogkFm1jZhuDPv7xImmdV36v6t+eV2lURRwkdoWN0inx0gWroBtVRA1GUo2f0it6cJ+fFeXc+Zq1LTjFzgP7A+fwB7JSWFw==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 449 | sha1_base64="2suSYs2KtjHJeb1CIts1JrhYPII=">AAAB/nicbVBNS8NAEN34WetXVDx5WSyCp5KIoMeiF48V7Ac0sWy2m3bpZjfsTpQSCv4VLx4U8erv8Oa/cdvmoK0PBh7vzTAzL0oFN+B5387S8srq2nppo7y5tb2z6+7tN43KNGUNqoTS7YgYJrhkDeAgWDvVjCSRYK1oeD3xWw9MG67kHYxSFiakL3nMKQErdd3DQA/Ufc4CzfsDIFqrR5yNu27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shz | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 450 | 27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shzIkGTgUbl4PMsJTQIemzjqWSJMyE+fT8MT6xSg/HStuSgKfq74mcJMaMksh2JgQGZt6biP95nQziyzDnMs2ASTpbFGcCg8KTLHCPa0ZBjCwhVHN7K6YDogkFm1jZhuDPv7xImmdV36v6t+eV2lURRwkdoWN0inx0gWroBtVRA1GUo2f0it6cJ+fFeXc+Zq1LTjFzgP7A+fwB7JSWFw==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 451 | sha1_base64="2suSYs2KtjHJeb1CIts1JrhYPII=">AAAB/nicbVBNS8NAEN34WetXVDx5WSyCp5KIoMeiF48V7Ac0sWy2m3bpZjfsTpQSCv4VLx4U8erv8Oa/cdvmoK0PBh7vzTAzL0oFN+B5387S8srq2nppo7y5tb2z6+7tN43KNGUNqoTS7YgYJrhkDeAgWDvVjCSRYK1oeD3xWw9MG67kHYxSFiakL3nMKQErdd3DQA/Ufc4CzfsDIFqrR5yNu27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shz | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 452 | 27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shzIkGTgUbl4PMsJTQIemzjqWSJMyE+fT8MT6xSg/HStuSgKfq74mcJMaMksh2JgQGZt6biP95nQziyzDnMs2ASTpbFGcCg8KTLHCPa0ZBjCwhVHN7K6YDogkFm1jZhuDPv7xImmdV36v6t+eV2lURRwkdoWN0inx0gWroBtVRA1GUo2f0it6cJ+fFeXc+Zq1LTjFzgP7A+fwB7JSWFw==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 453 | sha1_base64="2suSYs2KtjHJeb1CIts1JrhYPII=">AAAB/nicbVBNS8NAEN34WetXVDx5WSyCp5KIoMeiF48V7Ac0sWy2m3bpZjfsTpQSCv4VLx4U8erv8Oa/cdvmoK0PBh7vzTAzL0oFN+B5387S8srq2nppo7y5tb2z6+7tN43KNGUNqoTS7YgYJrhkDeAgWDvVjCSRYK1oeD3xWw9MG67kHYxSFiakL3nMKQErdd3DQA/Ufc4CzfsDIFqrR5yNu27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shz | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 454 | 27Fq3pT4EXiF6SCCtS77lfQUzRLmAQqiDEd30shzIkGTgUbl4PMsJTQIemzjqWSJMyE+fT8MT6xSg/HStuSgKfq74mcJMaMksh2JgQGZt6biP95nQziyzDnMs2ASTpbFGcCg8KTLHCPa0ZBjCwhVHN7K6YDogkFm1jZhuDPv7xImmdV36v6t+eV2lURRwkdoWN0inx0gWroBtVRA1GUo2f0it6cJ+fFeXc+Zq1LTjFzgP7A+fwB7JSWFw==</latexit> e | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 455 | <latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 456 | sha1_base64="gRKFy+QFytmwqWy0cvo5FmmPz8I=">AAAB7XicbVA9SwNBEJ3zM8avqGBjsxgEq3Bno4VFwMYygrkEkjPubeaSNXu3x+6eEo78BxsLRWz9P3b+GzcfhSY+GHi8N8PMvDAVXBvX/XaWlldW19YLG8XNre2d3dLevq9lphjWmRRSNUOqUfAE64Ybgc1UIY1DgY1wcDX2G4+oNJfJrRmmGMS0l/CIM2qs5LfTPr/DTqnsVtwJyCLxZqRcPfSf7gGg1il9tbuSZTEmhgmqdctzUxPkVBnOBI6K7 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 457 | fSf7gGg1il9tbuSZTEmhgmqdctzUxPkVBnOBI6K7UxjStmA9rBlaUJj1EE+uXZETqzSJZFUthJDJurviZzGWg/j0HbG1PT1vDcW//NamYkugpwnaWYwYdNFUSaIkWT8OulyhcyIoSWUKW5vJaxPFWXGBlS0IXjzLy8S/6ziuRXvxqZxCVMU4AiO4RQ8OIcqXEMN6sDgAZ7hFd4c6bw4787HtHXJmc0cwB84nz+QDpCP</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 458 | sha1_base64="74MJShuZzxGyM2nLY3EY87InuhI=">AAAB7XicbVC7SgNBFJ2NrxhfUcHGZjAIVmHXRguLgI1lBLMJJGuYndxNxszODjOzyrLkH2wsFNHS/7HzA/wPJ49CEw9cOJxzL/feE0rOtHHdL6ewtLyyulZcL21sbm3vlHf3fJ2kikKDJjxRrZBo4ExAwzDDoSUVkDjk0AyHl2O/eQ9Ks0TcmExCEJO+YBGjxFjJ78gBu4VuueJW3QnwIvFmpFI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRD | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 459 | FI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRDqNSJ9UgCR2SPrQtFSQGHeSTa0f42Co9HCXKljB4ov6eyEmsdRaHtjMmZqDnvbH4n9dOTXQe5EzI1ICg00VRyrFJ8Ph13GMKqOGZJYQqZm/FdEAUocYGVLIhePMvLxL/tOq5Ve/apnGBpiiiQ3SETpCHzlANXaE6aiCK7tAjekYvTuI8Oa/O27S14Mxm9tEfOB8/TZOSnA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 460 | sha1_base64="74MJShuZzxGyM2nLY3EY87InuhI=">AAAB7XicbVC7SgNBFJ2NrxhfUcHGZjAIVmHXRguLgI1lBLMJJGuYndxNxszODjOzyrLkH2wsFNHS/7HzA/wPJ49CEw9cOJxzL/feE0rOtHHdL6ewtLyyulZcL21sbm3vlHf3fJ2kikKDJjxRrZBo4ExAwzDDoSUVkDjk0AyHl2O/eQ9Ks0TcmExCEJO+YBGjxFjJ78gBu4VuueJW3QnwIvFmpFI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRD | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 461 | FI78B9k9v1e75Y/O72EpjEIQznRuu250gQ5UYZRDqNSJ9UgCR2SPrQtFSQGHeSTa0f42Co9HCXKljB4ov6eyEmsdRaHtjMmZqDnvbH4n9dOTXQe5EzI1ICg00VRyrFJ8Ph13GMKqOGZJYQqZm/FdEAUocYGVLIhePMvLxL/tOq5Ve/apnGBpiiiQ3SETpCHzlANXaE6aiCK7tAjekYvTuI8Oa/O27S14Mxm9tEfOB8/TZOSnA==</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 462 | sha1_base64="pLq6KB/1S9uyUeWp/G4byg43mK0=">AAAB7XicbVA9SwNBEJ2LXzF+RS1tFoNgFe5stLAI2FhGMB+QnGFvM5es2ds9dveEEPIfbCwUsfX/2Plv3CRXaOKDgcd7M8zMi1LBjfX9b6+wtr6xuVXcLu3s7u0flA+PmkZlmmGDKaF0O6IGBZfYsNwKbKcaaRIJbEWjm5nfekJtuJL3dpximNCB5DFn1Dqp2U2H/AF75Ypf9ecgqyTISQVy1Hvlr25fsSxBaZmgxnQCP7XhhGrLmcBpqZsZTCkb0 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 463 | Hvlr25fsSxBaZmgxnQCP7XhhGrLmcBpqZsZTCkb0QF2HJU0QRNO5tdOyZlT+iRW2pW0ZK7+npjQxJhxErnOhNqhWfZm4n9eJ7PxVTjhMs0sSrZYFGeCWEVmr5M+18isGDtCmebuVsKGVFNmXUAlF0Kw/PIqaV5UA78a3PmV2nUeRxFO4BTOIYBLqMEt1KEBDB7hGV7hzVPei/fufSxaC14+cwx/4H3+AIWDjww=</latexit>E<latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 464 | sha1_base64="iJ/x8cSgmmYNbMN8WtCvsNrlH/U=">AAAB6HicbVBNS8NAEJ3Ur1q/qh69LBbBU0lE0IOHgggeW7Af0Iay2U7atZtN2N0IJfQXePGgiFd/kjf/jds2B219MPB4b4aZeUEiuDau++0U1tY3NreK26Wd3b39g/LhUUvHqWLYZLGIVSegGgWX2DTcCOwkCmkUCGwH49uZ335CpXksH8wkQT+iQ8lDzqixUuOuX664VXcOskq8nFQgR71f/uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i1 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 465 | uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i11JJI9R+Nj90Ss6sMiBhrGxJQ+bq74mMRlpPosB2RtSM9LI3E//zuqkJr/2MyyQ1KNliUZgKYmIy+5oMuEJmxMQSyhS3txI2oooyY7Mp2RC85ZdXSeui6rlVr3FZqd3kcRThBE7hHDy4ghrcQx2awADhGV7hzXl0Xpx352PRWnDymWP4A+fzB5cfjMM=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 466 | sha1_base64="iJ/x8cSgmmYNbMN8WtCvsNrlH/U=">AAAB6HicbVBNS8NAEJ3Ur1q/qh69LBbBU0lE0IOHgggeW7Af0Iay2U7atZtN2N0IJfQXePGgiFd/kjf/jds2B219MPB4b4aZeUEiuDau++0U1tY3NreK26Wd3b39g/LhUUvHqWLYZLGIVSegGgWX2DTcCOwkCmkUCGwH49uZ335CpXksH8wkQT+iQ8lDzqixUuOuX664VXcOskq8nFQgR71f/uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i1 | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
1806.01261 | 467 | uoNYpZGKA0TVOuu5ybGz6gynAmclnqpxoSyMR1i11JJI9R+Nj90Ss6sMiBhrGxJQ+bq74mMRlpPosB2RtSM9LI3E//zuqkJr/2MyyQ1KNliUZgKYmIy+5oMuEJmxMQSyhS3txI2oooyY7Mp2RC85ZdXSeui6rlVr3FZqd3kcRThBE7hHDy4ghrcQx2awADhGV7hzXl0Xpx352PRWnDymWP4A+fzB5cfjMM=</latexit><latexit | 1806.01261 | Relational inductive biases, deep learning, and graph networks | Artificial intelligence (AI) has undergone a renaissance recently, making
major progress in key domains such as vision, language, control, and
decision-making. This has been due, in part, to cheap data and cheap compute
resources, which have fit the natural strengths of deep learning. However, many
defining characteristics of human intelligence, which developed under much
different pressures, remain out of reach for current approaches. In particular,
generalizing beyond one's experiences--a hallmark of human intelligence from
infancy--remains a formidable challenge for modern AI.
The following is part position paper, part review, and part unification. We
argue that combinatorial generalization must be a top priority for AI to
achieve human-like abilities, and that structured representations and
computations are key to realizing this objective. Just as biology uses nature
and nurture cooperatively, we reject the false choice between
"hand-engineering" and "end-to-end" learning, and instead advocate for an
approach which benefits from their complementary strengths. We explore how
using relational inductive biases within deep learning architectures can
facilitate learning about entities, relations, and rules for composing them. We
present a new building block for the AI toolkit with a strong relational
inductive bias--the graph network--which generalizes and extends various
approaches for neural networks that operate on graphs, and provides a
straightforward interface for manipulating structured knowledge and producing
structured behaviors. We discuss how graph networks can support relational
reasoning and combinatorial generalization, laying the foundation for more
sophisticated, interpretable, and flexible patterns of reasoning. As a
companion to this paper, we have released an open-source software library for
building graph networks, with demonstrations of how to use them in practice. | http://arxiv.org/pdf/1806.01261 | [
"Peter W. Battaglia",
"Jessica B. Hamrick",
"Victor Bapst",
"Alvaro Sanchez-Gonzalez",
"Vinicius Zambaldi",
"Mateusz Malinowski",
"Andrea Tacchetti",
"David Raposo",
"Adam Santoro",
"Ryan Faulkner",
"Caglar Gulcehre",
"Francis Song",
"Andrew Ballard",
"Justin Gilmer",
"George Dahl",
"Ashish Vaswani",
"Kelsey Allen",
"Charles Nash",
"Victoria Langston",
"Chris Dyer",
"Nicolas Heess",
"Daan Wierstra",
"Pushmeet Kohli",
"Matt Botvinick",
"Oriol Vinyals",
"Yujia Li",
"Razvan Pascanu"
] | [
"cs.LG",
"cs.AI",
"stat.ML"
] | null | null | cs.LG | 20180604 | 20181017 | [] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.