bibtex_url
null | proceedings
stringlengths 42
42
| bibtext
stringlengths 197
848
| abstract
stringlengths 303
3.45k
| title
stringlengths 10
159
| authors
sequencelengths 1
34
⌀ | id
stringclasses 44
values | arxiv_id
stringlengths 0
10
| GitHub
sequencelengths 1
1
| paper_page
stringclasses 899
values | n_linked_authors
int64 -1
13
| upvotes
int64 -1
109
| num_comments
int64 -1
13
| n_authors
int64 -1
92
| Models
sequencelengths 0
100
| Datasets
sequencelengths 0
19
| Spaces
sequencelengths 0
100
| old_Models
sequencelengths 0
100
| old_Datasets
sequencelengths 0
19
| old_Spaces
sequencelengths 0
100
| paper_page_exists_pre_conf
int64 0
1
| type
stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
null | https://openreview.net/forum?id=EGZ4XjZhLx | @inproceedings{
shulgin2023towards,
title={Towards a Better Theoretical Understanding of Independent Subnetwork Training},
author={Egor Shulgin and Peter Richt{\'a}rik},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=EGZ4XjZhLx}
} | Modern advancements in large-scale machine learning would be impossible without the paradigm of data-parallel distributed computing. Since distributed computing with large-scale models imparts excessive pressure on communication channels, significant recent research has been directed toward co-designing communication compression strategies and training algorithms with the goal of reducing communication costs. While pure data parallelism allows better data scaling, it suffers from poor model scaling properties. Indeed, compute nodes are severely limited by memory constraints, preventing further increases in model size. For this reason, the latest achievements in training giant neural network models also rely on some form of model parallelism. In this work, we take a closer theoretical look at Independent Subnetwork Training (IST), which is a recently proposed and highly effective technique for solving the aforementioned problems. We identify fundamental differences between IST and alternative approaches, such as distributed methods with compressed communication, and provide a precise analysis of its optimization performance on a quadratic model. | Towards a Better Theoretical Understanding of Independent Subnetwork Training | [
"Egor Shulgin",
"Peter Richtárik"
] | Workshop/OPT | 2306.16484 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=DqWAuvbSIb | @inproceedings{
scieur2023adaptive,
title={Adaptive Quasi-Newton and Anderson Acceleration Framework with Explicit Global (Accelerated) Convergence Rates},
author={Damien Scieur},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=DqWAuvbSIb}
} | Despite the impressive numerical performance of quasi-Newton and Anderson/nonlinear-acceleration methods, their global convergence rates have remained elusive for over 50 years. This paper addresses this long-standing question by introducing a framework that derives novel and adaptive quasi-Newton or nonlinear/Anderson acceleration schemes. Under mild assumptions, the proposed iterative methods exhibit explicit, non-asymptotic convergence rates that blend those of gradient descent and Cubic Regularized Newton's method. The proposed approach also includes an accelerated version for convex functions. Notably, these rates are achieved adaptively, without prior knowledge of the function's smoothness parameter. The framework presented in this paper is generic, and algorithms such as Newton's method with random subspaces, finite difference, or lazy Hessian can be seen as special cases of this paper's algorithm. Numerical experiments demonstrate the efficiency of the proposed framework, even compared to the L-BFGS algorithm with Wolfe line search. | Adaptive Quasi-Newton and Anderson Acceleration Framework with Explicit Global (Accelerated) Convergence Rates | [
"Damien Scieur"
] | Workshop/OPT | 2305.19179 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=D8WJ7gQEG1 | @inproceedings{
zhang2023sions,
title={Sion's Minimax Theorem in Geodesic Metric Spaces and a Riemannian Extragradient Algorithm},
author={Peiyuan Zhang and Jingzhao Zhang and Suvrit Sra},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=D8WJ7gQEG1}
} | Deciding whether saddle points exist or are approximable for nonconvex-nonconcave problems is usually intractable. We take a step toward understanding a broad class of nonconvex-nonconcave minimax problems that do remain tractable. Specifically, we study minimax problems in geodesic metric spaces. The first main result of the paper is a geodesic metric space version of Sion's minimax theorem; we believe our proof is novel and broadly accessible as it relies on the finite intersection property alone. The second main result is a specialization to geodesically complete Riemannian manifolds, for which we analyze first-order methods for smooth minimax problems. | Sion's Minimax Theorem in Geodesic Metric Spaces and a Riemannian Extragradient Algorithm | [
"Peiyuan Zhang",
"Jingzhao Zhang",
"Suvrit Sra"
] | Workshop/OPT | 2202.06950 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=CasSVavA1L | @inproceedings{
scharr2023cup,
title={Cup Curriculum: Curriculum Learning on Model Capacity},
author={Luca Scharr and Vanessa Toborek},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=CasSVavA1L}
} | Curriculum learning (CL) aims to increase the performance of a learner on a given task by applying a specialized learning strategy.
This strategy focuses on either the dataset, the task, or the model.
There is little to no work analysing the possibilities to apply CL on the model capacity in natural language processing.
To close this gap, we propose the cup curriculum.
In a first phase of training we use a variation of iterative magnitude pruning to reduce model capacity.
These weights are reintroduced in a second phase, resulting in the model capacity to show a cup-shaped curve over the training iterations.
We empirically evaluate different strategies of the cup curriculum and show that it outperforms early stopping reliably while exhibiting a high resilience to overfitting. | Cup Curriculum: Curriculum Learning on Model Capacity | [
"Luca Scharr",
"Vanessa Toborek"
] | Workshop/OPT | 2311.03956 | [
"https://github.com/luca-scharr/cupcurriculum"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=CSt6nJFlBn | @inproceedings{
kornowski2023an,
title={An Algorithm with Optimal Dimension-Dependence for Zero-Order Nonsmooth Nonconvex Stochastic Optimization},
author={Guy Kornowski and Ohad Shamir},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=CSt6nJFlBn}
} | We study the complexity of producing $(\delta,\epsilon)$-stationary points of Lipschitz objectives which are possibly neither smooth nor convex, using only noisy function evaluations.
Recent works proposed several stochastic zero-order algorithms that solve this task, all of which suffer from a dimension-dependence of $\Omega(d^{3/2})$ where $d$ is the dimension of the problem, which was conjectured to be optimal. We refute this conjecture by providing a faster algorithm that has complexity $O(d\delta^{-1}\epsilon^{-3})$, which is optimal (up to numerical constants) with respect to $d$ and also optimal with respect to the accuracy parameters $\delta,\epsilon$, thus solving an open question due to Lin et al. (NeurIPS'22). Moreover, the convergence rate achieved by our algorithm is also optimal for smooth objectives, proving that in the nonconvex stochastic zero-order setting, *nonsmooth optimization is as easy as smooth optimization*.
We provide algorithms that achieve the aforementioned convergence rate in expectation as well as with high probability.
Our analysis is based on a simple yet powerful geometric lemma regarding the Goldstein-subdifferential set, which allows utilizing recent advancements in first-order nonsmooth nonconvex optimization. | An Algorithm with Optimal Dimension-Dependence for Zero-Order Nonsmooth Nonconvex Stochastic Optimization | [
"Guy Kornowski",
"Ohad Shamir"
] | Workshop/OPT | 2307.04504 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | oral |
|
null | https://openreview.net/forum?id=C5PY6GMJyJ | @inproceedings{
lawless2023fair,
title={Fair Minimum Representation Clustering},
author={Connor Lawless and Oktay Gunluk},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=C5PY6GMJyJ}
} | Clustering is an unsupervised learning task that aims to partition data into a set of clusters. In many applications, these clusters correspond to real-world constructs (e.g., electoral districts) whose benefit can only be attained by groups when they reach a minimum level of representation (e.g., 50\% to elect their desired candidate). This paper considers the problem of performing k-means clustering while ensuring groups (e.g., demographic groups) have that minimum level of representation in a specified number of clusters. We formulate the problem through a mixed-integer optimization framework and present an alternating minimization algorithm, called MiniReL, that directly incorporates the fairness constraints. While incorporating the fairness criteria leads to an NP-Hard assignment problem within the algorithm, we provide computational approaches that make the algorithm practical even for large datasets. Numerical results show that the approach is able to create fairer clusters with practically no increase in the clustering cost across standard benchmark datasets. | Fair Minimum Representation Clustering | [
"Connor Lawless",
"Oktay Gunluk"
] | Workshop/OPT | 2302.03151 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=BDv9GkWG1E | @inproceedings{
fazzone2023fair,
title={Fair Representation in Submodular Subset Selection: A Pareto Optimization Approach},
author={Adriano Fazzone and Yanhao Wang and Francesco Bonchi},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=BDv9GkWG1E}
} | In this paper, we study a novel multi-objective combinatorial optimization problem called Submodular Maximization with Fair Representation (SMFR), which selects subsets of bounded costs from a ground set such that a submodular (utility) function $f$ is maximized while a set of $d$ submodular (representativeness) functions $g_1, \ldots, g_d$ are also maximized.
SMFR can find applications in machine learning problems where utility and representativeness objectives should be considered simultaneously, such as social advertising, recommendation, and feature selection.
We show that the maximization of $f$ and $g_1, \ldots, g_d$ might conflict with each other, so that no single solution can approximate all of them at the same time.
Therefore, we propose a Pareto optimization approach to SMFR, which finds a set of solutions to approximate all Pareto optimal solutions with different trade-offs between these objectives.
Specifically, it converts an instance of SMFR into several submodular cover instances by adjusting the weights of objective functions and provides approximate solutions by running the greedy algorithm on each submodular cover instance.
In future work, we will consider how to apply SMFR in real-world problems and extend it to more general cases. | Fair Representation in Submodular Subset Selection: A Pareto Optimization Approach | [
"Adriano Fazzone",
"Yanhao Wang",
"Francesco Bonchi"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=AyzhPfICBX | @inproceedings{
franke2023new,
title={New Horizons in Parameter Regularization: A Constraint Approach},
author={J{\"o}rg K.H. Franke and Michael Hefenbrock and Gregor Koehler and Frank Hutter},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=AyzhPfICBX}
} | This work presents constrained parameter regularization (CPR), an alternative to traditional weight decay. Instead of applying a constant penalty uniformly to all parameters, we enforce an upper bound on a statistical measure (e.g., the L2-norm) of individual parameter groups. This reformulates learning as a constrained optimization problem. To solve this, we utilize an adaptation of the augmented Lagrangian method. Our approach allows for varying regularization strengths across different parameter groups, removing the need for explicit penalty coefficients in the regularization terms. CPR only requires two hyperparameters and introduces no measurable runtime overhead. We offer empirical evidence of CPR's effectiveness through experiments in the "grokking" phenomenon, object detection, and language modeling. Our findings show that CPR can counteract the effects of grokking, and it consistently matches or surpasses the performance of traditional weight decay. | New Horizons in Parameter Regularization: A Constraint Approach | [
"Jörg K.H. Franke",
"Michael Hefenbrock",
"Gregor Koehler",
"Frank Hutter"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=Aw8GuIevIa | @inproceedings{
wang2023continually,
title={Continually Adapting Optimizers Improve Meta-Generalization},
author={Wenyi Wang and Louis Kirsch and Francesco Faccio and Mingchen Zhuge and J{\"u}rgen Schmidhuber},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=Aw8GuIevIa}
} | Meta-learned optimizers increasingly outperform analytical handcrafted optimizers such as SGD and Adam. On some tasks, however, they fail to generalize strongly, underperforming handcrafted methods. Then one can fall back on handcrafted methods through a guard, to combine the efficiency benefits of learned optimizers and the guarantees of analytical methods. At some point in the iterative optimization process, however, such guards may make the learned optimizer incompatible with the remaining optimization, and thus useless for further progress. Our novel method Meta Guard keeps adapting the learned optimizer to the target optimization problem. It experimentally outperforms other baselines, adapting to new tasks during training. | Continually Adapting Optimizers Improve Meta-Generalization | [
"Wenyi Wang",
"Louis Kirsch",
"Francesco Faccio",
"Mingchen Zhuge",
"Jürgen Schmidhuber"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=AtzAqJQpan | @inproceedings{
asad2023surrogate,
title={Surrogate Minimization: An Optimization Algorithm for Training Large Neural Networks with Model Parallelism},
author={Reza Asad and Reza Babanezhad Harikandeh and Issam H. Laradji and Nicolas Le Roux and Sharan Vaswani},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=AtzAqJQpan}
} | Optimizing large memory-intensive neural networks requires distributing its layers across multiple GPUs (referred to as model parallelism). We develop a framework that allows decomposing a neural network layer-wise and train it by optimizing layer-wise local losses in parallel. By using the resulting framework with GPipe [11] (an effective pipelining strategy for model parallelism), we propose the Surrogate Minimization (SM) algorithm. SM allows for multiple parallel updates to the layer-wise parameters of a distributed neural network and consequently improves the GPU utilization of GPipe. Our framework ensures that the sum of local losses is a global upper-bound on the
neural network loss, and can be minimized efficiently. Under mild technical assumptions, we prove that SM requires O(1/ε) iterations in order to guarantee convergence to an ε-neighbourhood of a stationary point of the neural network loss. Finally, our experimental results on MLPs demonstrate that SM leads to faster convergence compared to competitive baselines. | Surrogate Minimization: An Optimization Algorithm for Training Large Neural Networks with Model Parallelism | [
"Reza Asad",
"Reza Babanezhad Harikandeh",
"Issam H. Laradji",
"Nicolas Le Roux",
"Sharan Vaswani"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=AA0kBBerxw | @inproceedings{
ishikawa2023on,
title={On the Parallel Complexity of Multilevel Monte Carlo in Stochastic Gradient Descent},
author={Kei Ishikawa},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=AA0kBBerxw}
} | In the stochastic gradient descent (SGD) for sequential simulations such as the neural stochastic differential equations, the Multilevel Monte Carlo (MLMC) method is known to offer better theoretical computational complexity compared to the naive Monte Carlo approach.
However, in practice, MLMC scales poorly on massively parallel computing platforms such as modern GPUs,
because of its large parallel complexity which is equivalent to that of the naive Monte Carlo method.
To cope with this issue, we propose the delayed MLMC gradient estimator that drastically reduces the parallel complexity of MLMC by recycling previously computed gradient components from earlier steps.
The proposed estimator provably reduces the average parallel complexity per iteration at the cost of a slightly worse per-iteration convergence rate.
In our numerical experiments, we employ an example of deep hedging to demonstrate the superior parallel complexity of our method compared to the standard MLMC in SGD. | On the Parallel Complexity of Multilevel Monte Carlo in Stochastic Gradient Descent | [
"Kei Ishikawa"
] | Workshop/OPT | 2310.02402 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=A05nWKTZAL | @inproceedings{
oh2023pruning,
title={Pruning Neural Networks with Velocity-Constrained Optimization},
author={Donghyun Oh and Jinseok Chung and Namhoon Lee},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=A05nWKTZAL}
} | Pruning has gained prominence as a way to compress over-parameterized neural networks. While
pruning can be understood as solving a sparsity-constrained optimization problem, pruning by di-
rectly solving this problem has been relatively underexplored. In this paper, we propose a method to
prune neural networks using the MJ algorithm, which interprets constrained optimization using the
framework of velocity-constrained optimization. The experimental results show that our method
can prune VGG19 and ResNet32 networks by more than 90% while preserving the high accuracy
of the dense network. | Pruning Neural Networks with Velocity-Constrained Optimization | [
"Donghyun Oh",
"Jinseok Chung",
"Namhoon Lee"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=9lm3FzJPpY | @inproceedings{
mathur2023feature,
title={Feature Selection in Generalized Linear models via the Lasso: To Scale or Not to Scale?},
author={Anant Mathur and Sarat Babu Moka and Zdravko Botev},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=9lm3FzJPpY}
} | The Lasso regression is a popular regularization method for feature selection in statistics.
Prior to computing the Lasso estimator in both linear and generalized linear models, it is common to conduct a preliminary rescaling of the feature matrix to ensure that all the features are standardized. Without this standardization, it is argued, the Lasso estimate will unfortunately depend on the units used to measure the features. We propose a new type of iterative rescaling of the features in the context of generalized linear models. Whilst existing Lasso algorithms perform a single scaling as a preprocessing step, the proposed rescaling is applied iteratively throughout the Lasso computation until convergence. We provide numerical examples, with both real and simulated data, illustrating that the proposed iterative rescaling can significantly improve the statistical performance of the Lasso estimator without incurring any significant additional computational cost. | Feature Selection in Generalized Linear models via the Lasso: To Scale or Not to Scale? | [
"Anant Mathur",
"Sarat Babu Moka",
"Zdravko Botev"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=9LSbN0sHwv | @inproceedings{
wang2023direct,
title={{DIRECT} Optimisation with Bayesian Insights: Assessing Reliability Under Fixed Computational Budgets},
author={Fu Wang and Zeyu Fu and Xiaowei Huang and Wenjie Ruan},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=9LSbN0sHwv}
} | We introduce a method for probabilistically evaluating the reliability of Lipschitzian global optimisation under a constrained computational budget, a context frequently encountered in various applications. By interpreting the slope data gathered during the optimisation process as samples from the objective function's derivative, we utilise Bayesian posterior prediction to derive a confidence score for the optimisation outcomes. We validate our approach using numerical experiments on four multi-dimensional test functions, and the results highlight the practicality and efficacy of our approach. | DIRECT Optimisation with Bayesian Insights: Assessing Reliability Under Fixed Computational Budgets | [
"Fu Wang",
"Zeyu Fu",
"Xiaowei Huang",
"Wenjie Ruan"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=8vggUWDeLF | @inproceedings{
liu2023understanding,
title={Understanding the Role of Optimization in Double Descent},
author={Chris Yuhao Liu and Jeffrey Flanigan},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=8vggUWDeLF}
} | The phenomenon of model-wise double descent, where the test error peaks and then reduces as the model size increases, is an interesting topic that has attracted the attention of researchers due to the striking observed gap between theory and practice \citep{Belkin2018ReconcilingMM}. Additionally, while double descent has been observed in various tasks and architectures, the peak of double descent can sometimes be noticeably absent or diminished, even without explicit regularization, such as weight decay and early stopping. In this paper, we investigate this intriguing phenomenon from the optimization perspective and propose a simple optimization-based explanation for why double descent sometimes occurs weakly or not at all. To the best of our knowledge, we are the first to demonstrate that many disparate factors contributing to model-wise double descent (initialization, normalization, batch size, learning rate, optimization algorithm) are unified from the viewpoint of optimization: model-wise double descent is observed if and only if the optimizer can find a sufficiently low-loss minimum. These factors directly affect the condition number of the optimization problem or the optimizer and thus affect the final minimum found by the optimizer, reducing or increasing the height of the double descent peak. We conduct a series of controlled experiments on random feature models and two-layer neural networks under various optimization settings, demonstrating this optimization-based unified view. Our results suggest the following implication: Double descent is unlikely to be a problem for real-world machine learning setups. Additionally, our results help explain the gap between weak double descent peaks in practice and strong peaks observable in carefully designed setups. | Understanding the Role of Optimization in Double Descent | [
"Chris Yuhao Liu",
"Jeffrey Flanigan"
] | Workshop/OPT | 2312.03951 | [
""
] | https://huggingface.co/papers/2312.03951 | 1 | 0 | 0 | 2 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=7Ie3aoqJg1 | @inproceedings{
gower2023variance,
title={Variance Reduced Model Based Methods: New rates and adaptive step sizes},
author={Robert M. Gower and Frederik Kunstner and Mark Schmidt},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=7Ie3aoqJg1}
} | Variance reduced gradients methods were introduced to control the variance of SGD (Stochastic Gradient Descent). Model-based methods are able to make use of a known lower bound on the loss, for instance, most loss functions are positive. We show how these two classes of methods can be seamlessly combined. As an example we present a Model-based Stochastic Average Gradient method MSAG, which results from using a truncated model together with the SAG method. At each iteration MSAG computes an adaptive learning rate based on a given known lower bound. When given access to the optimal objective as the lower bound, MSAG has several favorable convergence properties, including monotonic iterates, and convergence in the non-smooth, smooth and strongly convex setting. Our convergence theorems show that we can essentially trade-off knowing the smoothness constant $L_{\max}$ for knowing the optimal objective to achieve the favourable convergence of variance reduced gradient methods. Moreover our convergence proofs for MSAG are very simple, which is in contrast to complexity of the original convergence proofs of SAG. | Variance Reduced Model Based Methods: New rates and adaptive step sizes | [
"Robert M. Gower",
"Frederik Kunstner",
"Mark Schmidt"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=6tFGOXyykI | @inproceedings{
papadimitriou2023on,
title={On the convergence of warped proximal iterations for solving nonmonotone inclusions and applications},
author={Dimitri Papadimitriou and Bang C{\^o}ng Vu},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=6tFGOXyykI}
} | In machine learning, tackling fairness, robustness, and safeness requires to solve nonconvex optimization problems with various constraints. In this paper, we investigate the warped proximal iterations for solving the nonmonotone inclusions and its application to nonconvex QP with equality constraints. | On the convergence of warped proximal iterations for solving nonmonotone inclusions and applications | [
"Dimitri Papadimitriou",
"Bang Công Vu"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=6l7UCftiMx | @inproceedings{
wei2023on,
title={On the Synergy Between Label Noise and Learning Rate Annealing in Neural Network Training},
author={Stanley Wei and Tongzheng Ren and Simon Shaolei Du},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=6l7UCftiMx}
} | In the past decade, stochastic gradient descent (SGD) has emerged as one of the most dominant algorithms in neural network training, with enormous success in different application scenarios. However, the implicit bias of SGD with different training techniques is still under-explored. Some of the common heuristics in practice include 1) using large initial learning rates and decaying it as the training progresses, and 2) using mini-batch SGD instead of full-batch gradient descent. In this work, we show that under certain data distributions, these two techniques are both necessary to obtain good generalization on neural networks. We consider mini-batch SGD with label noise, and at the heart of our analysis lies the concept of feature learning order, which has previously been characterized theoretically by Li et al. (2019) and Abbe et al. (2021). Notably, we use this to give the first concrete separations in generalization guarantees, between training neural networks with both label noise SGD and learning rate annealing and training with one of these elements removed. | On the Synergy Between Label Noise and Learning Rate Annealing in Neural Network Training | [
"Stanley Wei",
"Tongzheng Ren",
"Simon Shaolei Du"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=6QSCcosB3a | @inproceedings{
gorantla2023optimizing,
title={Optimizing Group-Fair Plackett-Luce Ranking Models for Relevance and Ex-Post Fairness},
author={Sruthi Gorantla and Eshaan Bhansali and Amit Deshpande and Anand Louis},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=6QSCcosB3a}
} | In learning-to-rank (LTR), optimizing only the relevance (or the expected ranking utility) can cause representational harm to certain categories of items. We propose a novel objective that maximizes expected relevance only over those rankings that satisfy given representation constraints to ensure ex-post fairness.
Building upon recent work on an efficient sampler for ex-post group-fair rankings, we propose a group-fair Plackett-Luce model and show that it can be efficiently optimized for our objective in the LTR framework.
Experiments on three real-world datasets show that our algorithm guarantees fairness alongside usually having better relevance compared to the LTR baselines. In addition, our algorithm also achieves better relevance than post-processing baselines which also ensure ex-post fairness. Further, when implicit bias is injected into the training data, our algorithm typically outperforms existing LTR baselines in relevance. | Optimizing Group-Fair Plackett-Luce Ranking Models for Relevance and Ex-Post Fairness | [
"Sruthi Gorantla",
"Eshaan Bhansali",
"Amit Deshpande",
"Anand Louis"
] | Workshop/OPT | 2308.13242 | [
"https://github.com/sruthigorantla/group-fair-pl"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=5yGwzxRVcg | @inproceedings{
huang2023contrastive,
title={Contrastive Predict-and-Search for Mixed Integer Linear Programs},
author={Taoan Huang and Aaron M Ferber and Arman Zharmagambetov and Yuandong Tian and Bistra Dilkina},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=5yGwzxRVcg}
} | Mixed integer linear programs (MILP) are flexible and powerful tool for modeling and solving many difficult real-world combinatorial optimization problems. In this paper, we propose a novel machine learning-based framework ConPaS that learns to predict solutions to MILPs with contrastive learning. For training, we collect high-quality solutions as positive samples and low-quality or infeasible solutions as negative samples. We then learn to make discriminative predictions by contrasting the positive and negative samples. During test time, we predict assignments for a subset of integer variables of a MILP and then solve the resulting reduced MILP to construct high-quality solutions. Empirically, we show that ConPaS achieves state-of-the-art results compared to other ML-based approaches in terms of the quality of and the speed at which the solutions are found. | Contrastive Predict-and-Search for Mixed Integer Linear Programs | [
"Taoan Huang",
"Aaron M Ferber",
"Arman Zharmagambetov",
"Yuandong Tian",
"Bistra Dilkina"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=5wsQaTjEAa | @inproceedings{
r{\'a}cz2023optimization,
title={Optimization dependent generalization bound for Re{LU} networks based on sensitivity in the tangent bundle},
author={D{\'a}niel R{\'a}cz and Mihaly Petreczky and Balint Daroczy and Andr{\'a}s Csert{\'a}n},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=5wsQaTjEAa}
} | Recent advances in deep learning have given us some very
promising results on the generalization ability of deep neural networks, however
literature still lacks a comprehensive theory explaining why heavily
over-parametrized models are able to generalize well while fitting the training
data. In this paper we propose a PAC type bound on the generalization error of
feedforward ReLU networks via estimating the Rademacher complexity of the set of
networks available from an initial parameter vector via gradient descent. The
key idea is to bound the sensitivity of the network's gradient to perturbation
of the input data along the optimization trajectory. The obtained bound does
not explicitly depend on the depth of the network. Our results are
experimentally verified on the MNIST and CIFAR-10 datasets. | Optimization dependent generalization bound for ReLU networks based on sensitivity in the tangent bundle | [
"Dániel Rácz",
"Mihaly Petreczky",
"Balint Daroczy",
"András Csertán"
] | Workshop/OPT | 2310.17378 | [
"https://github.com/danielracz/tansens_public"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=5WNCD8fXUb | @inproceedings{
smith2023riemannian,
title={Riemannian Optimization for Euclidean Distance Geometry},
author={Chandler Mack Smith and Samuel P. Lichtenberg and HanQin Cai and Abiy Tasissa},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=5WNCD8fXUb}
} | The Euclidean distance geometry (EDG) problem is a crucial machine learning task that appears in many applications. Utilizing the pairwise Euclidean distance information of a given point set, EDG reconstructs the configuration of the point system. When only partial distance information is available, matrix completion techniques can be incorporated to fill in the missing pairwise distances. In this paper, we propose a novel dual basis Riemannian gradient descent algorithm, coined RieEDG, for the EDG completion problem. The numerical experiments verify the effectiveness of the proposed algorithm. In particular, we show that RieEDG can precisely reconstruct various datasets consisting of 2- and 3-dimensional points by accessing a small fraction of pairwise distance information. | Riemannian Optimization for Euclidean Distance Geometry | [
"Chandler Mack Smith",
"Samuel P. Lichtenberg",
"HanQin Cai",
"Abiy Tasissa"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=5J6SGZk69Y | @inproceedings{
solomou2023guc,
title={{GUC}: Unsupervised non-parametric Global Clustering and Anomaly Detection},
author={Chris Solomou},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=5J6SGZk69Y}
} | Anomaly Detection is a crucial task in the fields of optimization and Machine Learning,
with the ability of detecting global anomalies being of particular importance. In this paper,
we propose a novel non-parametric algorithm for automatically detecting global anomalies
in an unsupervised manner. Our algorithm is both effective and efficient, requiring no prior
assumptions or domain knowledge to be applied. It features two modes that utilize the
distance from the dataset’s center for grouping data points together. The first mode splits
the dataset into global clusters where each cluster signifies proximity from the center. The
second mode employs a threshold value for splitting the points into outliers and inliers. We
evaluate our proposal against other prominent methods using synthetic and real datasets.
Our experiments demonstrate that the proposed algorithm achieves state-of-the-art performance with minimum computational cost, and can successfully be applied to a wide range
of Machine Learning applications. | GUC: Unsupervised non-parametric Global Clustering and Anomaly Detection | [
"Chris Solomou"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=5HbXROFu3p | @inproceedings{
tian2023testing,
title={Testing Approximate Stationarity Concepts for Piecewise Affine Functions and Extensions},
author={Lai Tian and Anthony Man-Cho So},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=5HbXROFu3p}
} | We study various aspects of the fundamental computational problem of detecting approximate stationary points for piecewise affine (PA) functions, including computational complexity, regularity conditions, and robustness in implementation. Specifically, for a PA function, we show that testing first-order approximate stationarity concepts in terms of three commonly used subdifferential constructions is computationally intractable unless P=NP. To facilitate computability, we establish the first necessary and sufficient condition for the validity of an equality-type (Clarke) subdifferential sum rule for a certain representation of arbitrary PA functions. Our main tools are nonsmooth analysis and polytope theory. Moreover, to address an important implementation issue, we introduce the first oracle-polynomial-time algorithm to test near-approximate stationarity for PA functions. We complement our results with extensions to other subdifferentials and applications to a series of structured piecewise smooth functions, including $\rho$-margin-loss SVM, piecewise affine regression, and neural networks with nonsmooth activation functions. | Testing Approximate Stationarity Concepts for Piecewise Affine Functions and Extensions | [
"Lai Tian",
"Anthony Man-Cho So"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=4jEOdGqVC0 | @inproceedings{
zhou2023multihead,
title={Multi-head {CLIP}: Improving {CLIP} with Diverse Representations and Flat Minima},
author={Mo Zhou and Xiong Zhou and Li Erran Li and Stefano Ermon and Rong Ge},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=4jEOdGqVC0}
} | Contrastive Language-Image Pre-training (CLIP) has shown remarkable success in the field of multimodal learning by enabling joint understanding of text and images. In this paper, we introduce a novel method called Multi-head CLIP, inspired by Stein Variational Gradient Descent (SVGD) and Sharpness-aware Minimization (SAM). Our approach aims to enhance CLIP's learning capability by encouraging the model to acquire diverse features while also promoting convergence towards a flat loss region, resulting in improved generalization performance. We conduct extensive experiments on two benchmark datasets, YFCC15M and CC3M, to evaluate the effectiveness of our proposed method. The experimental results consistently demonstrate that multi-head CLIP outperforms both the original CLIP architecture and CLIP with the SAM optimizer. | Multi-head CLIP: Improving CLIP with Diverse Representations and Flat Minima | [
"Mo Zhou",
"Xiong Zhou",
"Li Erran Li",
"Stefano Ermon",
"Rong Ge"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=4TAdxCDadC | @inproceedings{
mathur2023dynalay,
title={DynaLay: An Introspective Approach to Dynamic Layer Selection for Deep Networks},
author={Mrinal Mathur and Sergey M. Plis},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=4TAdxCDadC}
} | Deep learning models have increasingly become computationally intensive, necessitating specialized hardware and significant runtime for both training and inference. In this work, we introduce DynaLay, a versatile and dynamic neural network architecture that employs a reinforcement learning agent to adaptively select which layers to execute for a given input. Our approach introduces an element of introspection into neural network architectures by enabling the model to recompute the results on more difficult inputs during inference, balancing the amount of expelled computation, optimizing for both performance and efficiency. The system comprises a main model constructed with Fixed-Point Iterative (FPI) layers, which can approximate complex functions with high fidelity, and an agent that chooses among these layers or a no-operation (NOP) action. Unique to our approach is a multi-faceted reward function that combines classification accuracy, computational time, and a penalty for redundant layer selection, thereby ensuring a harmonious trade-off between performance and cost. Experimental results demonstrate that DynaLay achieves comparable accuracy to conventional deep models while significantly reducing computational overhead. Our approach represents a significant step toward creating more efficient, adaptable, and universally applicable deep learning systems. | DynaLay: An Introspective Approach to Dynamic Layer Selection for Deep Networks | [
"Mrinal Mathur",
"Sergey M. Plis"
] | Workshop/OPT | 2312.12781 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=37Oh63lj1x | @inproceedings{
oh2023optimal,
title={Optimal Transport for Kernel Gaussian Mixture Models},
author={Jung Hun Oh and Rena Elkin and Anish Kumar Simhal and Jiening Zhu and Joseph O Deasy and Allen Tannenbaum},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=37Oh63lj1x}
} | The Wasserstein distance from optimal mass transport (OMT) is a powerful mathematical tool with numerous applications that provides a natural measure of the distance between two probability distributions. Several methods to incorporate OMT into widely used probabilistic models, such as Gaussian or Gaussian mixture, have been developed to enhance the capability of modeling complex multimodal densities of real datasets. However, very few studies have explored the OMT problems in a reproducing kernel Hilbert space (RKHS), wherein the kernel trick is utilized to avoid the need to explicitly map input data into a high-dimensional feature space. In the current study, we propose a Wasserstein-type metric to compute the distance between two Gaussian mixtures in a RKHS via the kernel trick, i.e., kernel Gaussian mixture models. | Optimal Transport for Kernel Gaussian Mixture Models | [
"Jung Hun Oh",
"Rena Elkin",
"Anish Kumar Simhal",
"Jiening Zhu",
"Joseph O Deasy",
"Allen Tannenbaum"
] | Workshop/OPT | 2310.18586 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=2X1L9ZLz5q | @inproceedings{
fatkhullin2023stochastic,
title={Stochastic Optimization under Hidden Convexity},
author={Ilyas Fatkhullin and Niao He and Yifan Hu},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=2X1L9ZLz5q}
} | In this work, we consider stochastic non-convex constrained optimization problems under hidden convexity, i.e., those that admit a convex reformulation via a black box (non-linear, but invertible) map $c: \mathcal{X} \rightarrow \mathcal{U}$. A number of non-convex problems ranging from optimal control, revenue and inventory management, to convex reinforcement learning all admit such a hidden convex structure. Unfortunately, in the majority of considered applications, the map $c(\cdot)$ is unavailable and therefore, the reduction to solving a convex optimization is not possible. On the other hand, the (stochastic) gradients with respect to the original variable $x\in \mathcal{X}$ are often easy to obtain. Motivated by these observations, we consider the projected stochastic (sub-) gradient methods under hidden convexity and provide the first sample complexity guarantees for global convergence in smooth and non-smooth settings. Additionally, we improve our results to the last iterate function value convergence in the smooth setting using the momentum variant of projected stochastic gradient descent. | Stochastic Optimization under Hidden Convexity | [
"Ilyas Fatkhullin",
"Niao He",
"Yifan Hu"
] | Workshop/OPT | 2401.00108 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=271xvdLOMV | @inproceedings{
dwaraknath2023on,
title={On Optimization Formulations of Finite Horizon {MDP}s},
author={Rajat Vadiraj Dwaraknath and Lexing Ying},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=271xvdLOMV}
} | In this paper, we extend the connection between linear programming formulations of MDPs and policy gradient methods for infinite horizon MDPs presented in (Ying, L., & Zhu, Y., 2020) to finite horizon MDPs. The main tool we use for this extension is a reduction from optimization formulations of finite horizon MDPs to infinite horizon MDPs. Additionally, we show using a reparameterization argument that the KKT conditions for the non-convex policy optimization problem for the finite horizon setting are sufficient for global optimality. Further, we use the reduction to extend the Quasi-Newton policy gradient algorithm of (Li et. al 2021) to the finite horizon case and achieve performance competitive with value iteration by exploiting backward induction for policy evaluation. To our knowledge, this serves as the first policy gradient-based method for finite horizon MDPs that is competitive with value iteration-based approaches. | On Optimization Formulations of Finite Horizon MDPs | [
"Rajat Vadiraj Dwaraknath",
"Lexing Ying"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=22cDNOtFGV | @inproceedings{
zelikman2023selftaught,
title={Self-Taught Optimizer ({STOP}): Recursively Self-Improving Code Generation},
author={Eric Zelikman and Eliana Lorch and Lester Mackey and Adam Tauman Kalai},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=22cDNOtFGV}
} | Several recent advances in AI systems (e.g., Tree-of-Thoughts and Program-Aided Language Models) solve problems by providing a "scaffolding" program that structures multiple calls to language models to generate better outputs. A scaffolding program is written in a programming language such as Python. In this work, we use a language-model-infused scaffolding program to improve itself. We start with a seed "improver" that improves an input program according to a given utility function by querying a language model several times and returning the best solution. We then run this seed improver to improve itself. Across a small set of downstream tasks, the resulting improved improver generates programs with significantly better performance than its seed improver. A variety of self-improvement strategies are proposed by the language model, including beam search, genetic algorithms, and simulated annealing. Since the language models themselves are not altered, this is not full recursive self-improvement. Nonetheless, it demonstrates that a modern language model, GPT-4 in our proof-of-concept experiments, is capable of writing code that can call itself to improve itself. We critically consider concerns around the development of self-improving technologies and evaluate the frequency with which the generated code bypasses a sandbox. | Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation | [
"Eric Zelikman",
"Eliana Lorch",
"Lester Mackey",
"Adam Tauman Kalai"
] | Workshop/OPT | 2310.02304 | [
"https://github.com/microsoft/stop"
] | https://huggingface.co/papers/2310.02304 | 1 | 1 | 0 | 4 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=0t1sHkPS06 | @inproceedings{
dong2023learning,
title={Learning Multiobjective Program Through Online Learning},
author={Chaosheng Dong and Yijia Wang and Bo Zeng},
booktitle={OPT 2023: Optimization for Machine Learning},
year={2023},
url={https://openreview.net/forum?id=0t1sHkPS06}
} | We investigate the problem of learning parameters (i.e., objective functions or constraints) of a multi-objective optimization problem, based on a set of sequentially arrived solutions. In particular, these solutions might not be exact and possibly carry measurement noise or are generated with the bounded rationality of decision makers. In this paper, we propose a general online learning framework to deal with this learning problem using inverse multi-objective optimization, and prove that this framework converges at a rate of under certain regularity conditions. More precisely, we develop two online learning algorithms with implicit update rules which can handle noisy data. Numerical results with both synthetic and real world datasets show that both algorithms can learn the parameters of a multi-objective program with great accuracy and are robust to noise. | Learning Multiobjective Program Through Online Learning | [
"Chaosheng Dong",
"Yijia Wang",
"Bo Zeng"
] | Workshop/OPT | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=yzKb3uFiir | @inproceedings{
song2023learning,
title={Learning Object Motion and Appearance Dynamics with Object-Centric Representations},
author={Yeon-Ji Song and Hyunseo Kim and Suhyung Choi and Jin-Hwa Kim and Byoung-Tak Zhang},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=yzKb3uFiir}
} | Human perception involves discerning objects based on attributes such as size, color, and texture, and making predictions about their movements using features such as weight and speed. This innate ability operates without the need for conscious learning, allowing individuals to perform actions like catching or avoiding objects when they are unaware. Accordingly, the fundamental key to achieving higher-level cognition lies in the capability to break down intricate multi-object scenes into meaningful appearances. Object-centric representations have emerged as a promising tool for scene decomposition by providing useful abstractions. In this paper, we propose a novel approach to unsupervised video prediction leveraging object-centric representations. Our methodology introduces a two-component model consisting of a slot encoder for object-centric disentanglement and a feature extraction module for masked patches. These components are integrated through a cross-attention mechanism, allowing for comprehensive spatio-temporal reasoning. Our model exhibits better performance when dealing with intricate scenes characterized by a wide range of object attributes and dynamic movements. Moreover, our approach demonstrates scalability across diverse synthetic environments, thereby showcasing its potential for widespread utilization in vision-related tasks. | Learning Object Motion and Appearance Dynamics with Object-Centric Representations | [
"Yeon-Ji Song",
"Hyunseo Kim",
"Suhyung Choi",
"Jin-Hwa Kim",
"Byoung-Tak Zhang"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=xzp8WYfIqL | @inproceedings{
lu2023attention,
title={Attention for Causal Relationship Discovery from Biological Neural Dynamics},
author={Ziyu Lu and Anika Tabassum and Shruti R. Kulkarn and Lu Mi and J. Nathan Kutz and Eric Todd SheaBrown and Seung-Hwan Lim},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=xzp8WYfIqL}
} | This paper explores the potential of the transformer models for learning Granger causality in networks with complex nonlinear dynamics at every node, as in neurobiological and biophysical networks. Our study primarily focuses on a proof-of-concept investigation based on simulated neural dynamics, for which the ground-truth causality is known through the underlying connectivity matrix. For transformer models trained to forecast neuronal population dynamics, we show that the cross attention module effectively captures the causal relationship among neurons, with an accuracy equal or superior to that for the most popular Granger causality analysis method. While we acknowledge that real-world neurobiology data will bring further challenges, including dynamic connectivity and unobserved variability, this research offers an encouraging preliminary glimpse into the utility of the transformer model for causal representation learning in neuroscience. | Attention for Causal Relationship Discovery from Biological Neural Dynamics | [
"Ziyu Lu",
"Anika Tabassum",
"Shruti R. Kulkarn",
"Lu Mi",
"J. Nathan Kutz",
"Eric Todd SheaBrown",
"Seung-Hwan Lim"
] | Workshop/CRL | 2311.06928 | [
"https://github.com/ziyulu-uw/causalformer"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=xRxks7FKXc | @inproceedings{
tumma2023leveraging,
title={Leveraging Low-Rank and Sparse Recurrent Connectivity for Robust Closed-Loop Control},
author={Neehal Tumma and Mathias Lechner and Noel Loo and Ramin Hasani and Daniela Rus},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=xRxks7FKXc}
} | Developing autonomous agents that can interact with changing environments is an open challenge in machine learning. Robustness is particularly important in these settings as agents are often fit offline on expert demonstrations but deployed online where they must generalize to the closed feedback loop within the environment. In this work, we explore the application of recurrent neural networks to tasks of this nature and understand how a parameterization of their recurrent connectivity influences robustness in closed-loop settings. Specifically, we represent the recurrent connectivity as a function of rank and sparsity and show both theoretically and empirically that modulating these two variables has desirable effects on network dynamics. The proposed low-rank, sparse connectivity induces an interpretable prior on the network that proves to be most amenable for a class of models known as closed-form continuous-time neural networks (CfCs). We find that CfCs with fewer parameters can outperform their full-rank, fully-connected counterparts in the online setting under distribution shift. This yields memory-efficient and robust agents while opening a new perspective on how we can modulate network dynamics through connectivity. | Leveraging Low-Rank and Sparse Recurrent Connectivity for Robust Closed-Loop Control | [
"Neehal Tumma",
"Mathias Lechner",
"Noel Loo",
"Ramin Hasani",
"Daniela Rus"
] | Workshop/CRL | 2310.03915 | [
""
] | https://huggingface.co/papers/2310.03915 | 0 | 0 | 0 | 5 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=xOgWq35OH1 | @inproceedings{
reber2023whats,
title={What's your Use Case? A Taxonomy of Causal Evaluations of Post-hoc Interpretability},
author={David Reber and Cristina Garbacea and Victor Veitch},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=xOgWq35OH1}
} | Post-hoc interpretability of neural network models, including Large Language Models (LLMs), often aims for mechanistic interpretations — detailed, causal descriptions of model behavior. However, human interpreters may lack the capacity or willingness to formulate intricate mechanistic models, let alone evaluate them. This paper addresses this challenge by introducing a taxonomy which dissects the overarching goal of mechanistic interpretability into constituent claims, each requiring distinct evaluation methods. By doing so, we transform these evaluation criteria into actionable learning objectives, providing a data-driven pathway to interpretability. This framework enables a methodologically rigorous yet pragmatic approach to evaluating the strengths and limitations of various interpretability tools. | What's your Use Case? A Taxonomy of Causal Evaluations of Post-hoc Interpretability | [
"David Reber",
"Cristina Garbacea",
"Victor Veitch"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=x9XRiwXNnm | @inproceedings{
yang2023learning,
title={Learning Unknown Intervention Targets in Structural Causal Models from Heterogeneous Data},
author={Yuqin Yang and Saber Salehkaleybar and Negar Kiyavash},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=x9XRiwXNnm}
} | We study the problem of identifying the unknown intervention targets in structural causal models where we have access to heterogeneous data collected from multiple environments. The unknown intervention targets are the set of endogenous variables whose corresponding exogenous noises change across the environments. We propose a two-phase approach which in the first phase recovers the exogenous noises corresponding to unknown intervention targets whose distributions have changed across environments. In the second phase, the recovered noises are matched with the corresponding endogenous variables. For the recovery phase, we provide sufficient conditions for learning these exogenous noises up to some
component-wise invertible transformation. For the matching phase, under the causal sufficiency assumption, we show that the proposed method uniquely identifies the intervention targets. In the presence of latent confounders, the intervention targets among the observed variables cannot be determined uniquely. We provide a candidate intervention target set which is a superset of the true intervention targets. Our approach improves upon the state of the art as the returned candidate set is always a subset of the target set returned by previous work. Moreover, we do not require restrictive assumptions such as linearity of the causal model or performing invariance tests to learn whether a distribution is changing across environments which could be highly sample inefficient.
Our experimental results show the effectiveness
of our proposed algorithm in practice. | Learning Unknown Intervention Targets in Structural Causal Models from Heterogeneous Data | [
"Yuqin Yang",
"Saber Salehkaleybar",
"Negar Kiyavash"
] | Workshop/CRL | 2312.06091 | [
"https://github.com/yuqin-yang/lit"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=u0K8gTasKX | @inproceedings{
komanduri2023learning,
title={Learning Causally Disentangled Representations via the Principle of Independent Causal Mechanisms},
author={Aneesh Komanduri and Yongkai Wu and Feng Chen and Xintao Wu},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=u0K8gTasKX}
} | Learning disentangled causal representations is a challenging problem that has gained significant attention recently due to its implications for extracting meaningful information for downstream tasks. In this work, we define a new notion of causal disentanglement from the perspective of independent causal mechanisms. We propose ICM-VAE, a framework for learning causally disentangled representations supervised by causally related observed labels. We model causal mechanisms using learnable flow-based diffeomorphic functions to map noise variables to latent causal variables. Further, to promote the disentanglement of causal factors, we propose a causal disentanglement prior that utilizes the known causal structure to encourage learning a causally factorized distribution in the latent space. Under relatively mild conditions, we provide theoretical results showing the identifiability of causal factors and mechanisms up to permutation and elementwise reparameterization. We empirically demonstrate that our framework induces highly disentangled causal factors, improves interventional robustness, and is compatible with counterfactual generation. | Learning Causally Disentangled Representations via the Principle of Independent Causal Mechanisms | [
"Aneesh Komanduri",
"Yongkai Wu",
"Feng Chen",
"Xintao Wu"
] | Workshop/CRL | 2306.01213 | [
"https://github.com/akomand/icm-vae"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=rCOdznPBf6 | @inproceedings{
kulinski2023towards,
title={Towards Characterizing Domain Counterfactuals for Invertible Latent Causal Models},
author={Sean Kulinski and Zeyu Zhou and Ruqi Bai and Murat Kocaoglu and David I. Inouye},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=rCOdznPBf6}
} | Answering counterfactual queries has many important applications such as knowledge discovery and explainability, but is challenging when causal variables are unobserved and we only see a projection onto an observation space, for instance, image pixels. One approach is to recover the latent Structural Causal Model (SCM), but this typically needs unrealistic assumptions, such as linearity of the causal mechanisms. Another approach is to use naïve ML approximations, such as generative models, to generate counterfactual samples; however, these lack guarantees of accuracy. In this work, we strive to strike a balance between practicality and theoretical guarantees by focusing on a specific type of causal query called *domain counterfactuals*, which hypothesizes what a sample would have looked like if it had been generated in a different domain (or environment). Concretely, by only assuming invertibility, sparse domain interventions and access to observational data from different domains, we aim to improve domain counterfactual estimation both theoretically and practically with less restrictive assumptions. We define *domain counterfactually equivalent* models and prove necessary and sufficient properties for equivalent models that provide a tight characterization of the domain counterfactual equivalence classes. Building upon this result, we prove that every equivalence class contains a model where all intervened variables are at the end when topologically sorted by the causal DAG. This surprising result suggests that a model design that only allows intervention in the last *k* latent variables may improve model estimation for counterfactuals. We then test this model design on extensive simulated and image-based experiments which show the sparse canonical model indeed improves counterfactual estimation over baseline non-sparse models. | Towards Characterizing Domain Counterfactuals for Invertible Latent Causal Models | [
"Sean Kulinski",
"Zeyu Zhou",
"Ruqi Bai",
"Murat Kocaoglu",
"David I. Inouye"
] | Workshop/CRL | 2306.11281 | [
"https://github.com/inouye-lab/ild-domain-counterfactuals"
] | https://huggingface.co/papers/2306.11281 | 1 | 0 | 0 | 5 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=q2uo4f3IRo | @inproceedings{
nam2023scadi,
title={{SCADI}: Self-supervised Causal Disentanglement in Latent Variable Models},
author={Heejeong Nam},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=q2uo4f3IRo}
} | Causal disentanglement has great potential for capturing complex situations. However, there is a lack of practical and efficient approaches. It is already known that most unsupervised disentangling methods are unable to produce identifiable results without additional information, often leading to randomly disentangled output. Therefore, most existing models for disentangling are weakly supervised, providing information about intrinsic factors, which incurs excessive costs. Therefore, we propose a novel model, SCADI(SElf-supervised CAusal DIsentanglement), that enables the model to discover semantic factors and learn their causal relationships without any supervision. This model combines a masked structural causal model (SCM) with a pseudo-label generator for causal disentanglement, aiming to provide a new direction for self-supervised causal disentanglement models. | SCADI: Self-supervised Causal Disentanglement in Latent Variable Models | [
"Heejeong Nam"
] | Workshop/CRL | 2311.06567 | [
"https://github.com/hazel-heejeong-nam/self-supervised-causal-disentanglement"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=m9s6rnYWqm | @inproceedings{
wu2023invertedattention,
title={Inverted-Attention Transformers can Learn Object Representations: Insights from Slot Attention},
author={Yi-Fu Wu and Klaus Greff and Gamaleldin Fathy Elsayed and Michael Curtis Mozer and Thomas Kipf and Sjoerd van Steenkiste},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=m9s6rnYWqm}
} | Visual reasoning is supported by a causal understanding of the physical world, and theories of human cognition suppose that a necessary step to causal understanding is the discovery and representation of high-level entities like objects. Slot Attention is a popular method aimed at object-centric learning, and its popularity has resulted in dozens of variants and extensions. To help understand the core assumptions that lead to successful object-centric learning, we take a step back and identify the minimal set of changes to a standard Transformer architecture to obtain the same performance as the specialized Slot Attention models. We systematically evaluate the performance and scaling behaviour of several "intermediate" architectures on seven image and video datasets from prior work. Our analysis reveals that by simply inverting the attention mechanism of Transformers, we obtain performance competitive with state-of-the-art Slot Attention in several domains. | Inverted-Attention Transformers can Learn Object Representations: Insights from Slot Attention | [
"Yi-Fu Wu",
"Klaus Greff",
"Gamaleldin Fathy Elsayed",
"Michael Curtis Mozer",
"Thomas Kipf",
"Sjoerd van Steenkiste"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=kbXmvuk2Mc | @inproceedings{
xi2023triangular,
title={Triangular Monotonic Generative Models Can Perform Causal Discovery},
author={Quanhan Xi and Sebastian Gonzalez and Benjamin Bloem-Reddy},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=kbXmvuk2Mc}
} | Many causal discovery algorithms exploit conditional independence signatures in observational data, recovering a Markov equivalence class (MEC) of possible DAGs consistent with the data. In case the MEC is non-trivial, additional assumptions on the data generating process can be made, and generative models can be fit to further resolve the MEC. We show that triangular monotonic increasing (TMI) maps parametrize generative models that perform conditional independence-based causal discovery by searching over permutations, that additionally are flexible enough as generative models to fit a wide class of causal models. In this paper, we characterize the theoretical properties that make these models relevant as tools for causal discovery, make connections to existing methods, and highlight open challenges towards their deployment. | Triangular Monotonic Generative Models Can Perform Causal Discovery | [
"Quanhan Xi",
"Sebastian Gonzalez",
"Benjamin Bloem-Reddy"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=jWP5C8e8Db | @inproceedings{
maasch2023local,
title={Local Discovery by Partitioning: Polynomial-Time Causal Discovery Around Exposure-Outcome Pairs},
author={Jacqueline R. M. A. Maasch and Weishen Pan and Shantanu Gupta and Volodymyr Kuleshov and Kyra Gan and Fei Wang},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=jWP5C8e8Db}
} | This work addresses the problem of automated covariate selection under limited prior knowledge. Given an exposure-outcome pair {X,Y} and a variable set Z of unknown causal structure, the Local Discovery by Partitioning (LDP) algorithm partitions Z into subsets defined by their relation to {X,Y}. We enumerate eight exhaustive and mutually exclusive partitions of any arbitrary Z and leverage this taxonomy to differentiate confounders from other variable types. LDP is motivated by valid adjustment set identification, but avoids the pretreatment assumption commonly made by automated covariate selection methods. We provide theoretical guarantees that LDP returns a valid adjustment set for any Z that meets sufficient graphical conditions. Under stronger conditions, we prove that partition labels are asymptotically correct. Total independence tests is worst-case quadratic in |Z|, with sub-quadratic runtimes observed empirically. We numerically validate our theoretical guarantees on synthetic and semi-synthetic graphs. Adjustment sets from LDP yield less biased and more precise average treatment effect estimates than baselines, with LDP outperforming on confounder recall, test count, and runtime for valid adjustment set discovery. | Local Discovery by Partitioning: Polynomial-Time Causal Discovery Around Exposure-Outcome Pairs | [
"Jacqueline R. M. A. Maasch",
"Weishen Pan",
"Shantanu Gupta",
"Volodymyr Kuleshov",
"Kyra Gan",
"Fei Wang"
] | Workshop/CRL | 2310.17816 | [
"https://github.com/jmaasch/ldp"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=gNMANHvjO1 | @inproceedings{
clivio2023towards,
title={Towards representation learning for general weighting problems in causal inference},
author={Oscar Clivio and Avi Feller and Christopher C. Holmes},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=gNMANHvjO1}
} | Weighting problems in treatment effect estimation can be solved by minimising an appropriate probability distance. Choosing which distance to minimise, however, can be challenging as it depends on the unknown data generating process. An alternative is to instead choose a distance that depends on a suitable representation of covariates. In this work, we give errors that quantify the bias added to a weighting estimator when using a representation, giving clear objectives to minimise when learning the representation and generalising a large body of previous work on deconfounding, prognostic, balancing and propensity scores. We further outline a method minimising such objectives, and show promising numerical results on two semi-synthetic datasets. | Towards representation learning for general weighting problems in causal inference | [
"Oscar Clivio",
"Avi Feller",
"Christopher C. Holmes"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=fpSpF1Wjek | @inproceedings{
mutti2023exploiting,
title={Exploiting Causal Representations in Reinforcement Learning: A Posterior Sampling Approach},
author={Mirco Mutti and Riccardo De Santi and Marcello Restelli and Alexander Marx and Giorgia Ramponi},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=fpSpF1Wjek}
} | Posterior sampling allows the exploitation of prior knowledge of the environment's transition dynamics to improve the sample efficiency of reinforcement learning. The prior is typically specified as a class of parametric distributions, a task that can be cumbersome in practice, often resulting in the choice of uninformative priors. Instead, in this work we study how to exploit causal representations to build priors that are often more natural to design. Specifically, we propose a novel hierarchical posterior sampling approach, called C-PSRL, in which the prior is given as a (partial) causal graph over the environment's causal variables, such as listing known causal dependencies between biometric features in a medical treatment study. C-PSRL simultaneously learns a graph consistent with the true causal graph at the higher level and the parameters of the resulting factored dynamics at the lower level. For this procedure, we provide an analysis of its Bayesian regret, which explicitly connects the regret rate with the degree of causal knowledge, and we show how regret minimization leads to a weak notion of causal discovery. | Exploiting Causal Representations in Reinforcement Learning: A Posterior Sampling Approach | [
"Mirco Mutti",
"Riccardo De Santi",
"Marcello Restelli",
"Alexander Marx",
"Giorgia Ramponi"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=fkoqMdTlEg | @inproceedings{
moeed2023identifying,
title={Identifying Effects of Disease on Single-Cells with Domain-Invariant Generative Modeling},
author={Abdul Moeed and Martin Rohbeck and Kai Ueltzhoeffer and Pavlo Lutsik and Oliver Stegle and Marc Jan Bonder},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=fkoqMdTlEg}
} | A core challenge in computational biology is predicting the effects of disease on healthy tissue. From the machine learning perspective, effects of disease and other stimulations on gene expression of single cells can be modeled as a domain shift in a low-dimensional latent space applied to healthy cells. Guided by principles of domain-invariance and compositional models, we present "single-cell Domain Shift Autoencoder (scDSA)", a deep generative model for disentangling disease-invariant and disease-specific gene programs at single-cell resolution. scDSA uncovers latent factors that are conserved across healthy and disease cell states, and learns how these factors interact with disease. We show that our model i) predicts counterfactual healthy cell-types of diseased cells in unseen patients, ii) captures interpretable representations of disease(s), and iii) learns interaction of disease effects and cell-types. scDSA helps to further our understanding of how diseases perturb healthy tissue on a patient-specific basis therefore enabling advances in personalized healthcare. | Identifying Effects of Disease on Single-Cells with Domain-Invariant Generative Modeling | [
"Abdul Moeed",
"Martin Rohbeck",
"Kai Ueltzhoeffer",
"Pavlo Lutsik",
"Oliver Stegle",
"Marc Jan Bonder"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | oral |
||
null | https://openreview.net/forum?id=aCiFCj4rEA | @inproceedings{
pan2023learning,
title={Learning Endogenous Representation in Reinforcement Learning via Advantage Estimation},
author={Hsiao-Ru Pan and Bernhard Sch{\"o}lkopf},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=aCiFCj4rEA}
} | Recently, it was shown that the advantage function can be understood as quantifying the causal effect of an action on the cumulative reward. However, this connection remained largely analogical, with unclear implications. In the present work, we examine this analogy using the Exogenous Markov Decision Process (ExoMDP) framework, which factorizes an MDP into variables that are causally related to the agent's actions (endogenous) and variables that are beyond the agent's control (exogenous). We demonstrate that the advantage function can be expressed using only the endogenous variables, which is, in general, not possible for the (action-)value function. Through experiments in a toy ExoMDP, we found that estimating the advantage function directly can facilitate learning representations that are invariant to the exogenous variables. | Learning Endogenous Representation in Reinforcement Learning via Advantage Estimation | [
"Hsiao-Ru Pan",
"Bernhard Schölkopf"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=Zs3C7zytfp | @inproceedings{
singh2023causal,
title={Causal Regressions For Unstructured Data},
author={Amandeep Singh and Bolong Zheng},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=Zs3C7zytfp}
} | The focus of much recent research in economics and marketing has been (1) to allow for unstructured data in causal studies and (2) to flexibly address the issue of endogeneity with observational data and perform valid causal inference. Directly using machine learning algorithms to predict the outcome
variable can help deal with the issue of unstructured data; however, it is well known
that such an approach does not perform well in the presence of endogeneity in the
explanatory variables. On the other hand, extant methods catered towards addressing endogeneity issues make strong parametric assumptions and hence are incapable of “directly" incorporating high-dimensional unstructured data. In this paper, we propose an estimator, which we term “RieszIV" for carrying out estimation and inference with high-dimensional observational data
without resorting to parametric approximations. We demonstrate our estimator exhibits asymptotic consistency and normality under a mild set of conditions. We carry out extensive Monte Carlo simulations with both low-dimensional and high-dimensional
unstructured data to demonstrate the finite sample performance of our estimator. Finally, using app downloads and review data for apps on Google Play we demonstrate how our method can be used to conduct inference over counterfactual policies over rich text data. We show how large language models can be used as a viable counterfactual policy generation operator. This represents an important advance in expanding counterfactual inference to complex, real-world settings. | Causal Regressions For Unstructured Data | [
"Amandeep Singh",
"Bolong Zheng"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=YQt6emXCYR | @inproceedings{
corazza2023expediting,
title={Expediting Reinforcement Learning by Incorporating Temporal Causal Information},
author={Jan Corazza and Hadi Partovi Aria and Daniel Neider and Zhe Xu},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=YQt6emXCYR}
} | Reinforcement learning (RL) algorithms struggle with learning optimal policies for tasks where reward feedback is sparse and depends on a complex sequence of events in the environment. Probabilistic reward machines (PRMs) are finite-state formalisms that can capture temporal dependencies in the reward signal, along with nondeterministic task outcomes. While special RL algorithms can exploit this finite-state structure to expedite learning, PRMs remain difficult to modify and design by hand. This hinders the already difficult tasks of utilizing high-level causal knowledge about the environment, and transferring the reward formalism into a new domain with a different causal structure. This paper proposes a novel method to incorporate causal information in the form of Temporal Logic-based Causal Diagrams into the reward formalism, thereby expediting policy learning and aiding the transfer of task specifications to new environments. | Expediting Reinforcement Learning by Incorporating Temporal Causal Information | [
"Jan Corazza",
"Hadi Partovi Aria",
"Daniel Neider",
"Zhe Xu"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=XwAT4LSaXe | @inproceedings{
han2023disk,
title={{DISK}: Domain Inference for Discovering Spurious Correlation with {KL}-Divergence},
author={Yujin Han and Difan Zou},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=XwAT4LSaXe}
} | Existing methods utilize domain information to address the subpopulation shift issue and enhance model generalization. However, the availability of domain information is not always guaranteed. In response to this challenge, we introduce a novel end-to-end method called DISK. DISK discovers the spurious correlations present in the training and validation sets through KL-divergence and assigns spurious labels (which are also the domain labels) to classify instances based on spurious features. By combining spurious labels $y_s$ with true labels $y$, DISK effectively partitions the data into different groups with unique data distributions $\mathbb{P}(\mathbf{x}|y,y_s)$. The group partition inferred by DISK then can be seamlessly leveraged to design algorithms to further mitigate the subpopulation shift and improve generalization on test data. Unlike existing domain inference methods, such as ZIN \citep{lin2022zin} and DISC \citep{wu2023discover}, DISK reliably infers domains without requiring additional information. We extensively evaluated DISK on different datasets, considering scenarios where validation labels are either available or unavailable, demonstrating its effectiveness in domain inference and mitigating subpopulation shift. Furthermore, our results also suggest that for some complex data, the neural network-based DISK may have the potential to perform more reasonable domain inferences, which highlights the potential effective integration of DISK and human decisions when the (human-defined) domain information is available.
Codes of DISK are available at [https://anonymous.4open.science/r/DISK-E23A/](https://anonymous.4open.science/r/DISK-E23A/). | DISK: Domain Inference for Discovering Spurious Correlation with KL-Divergence | [
"Yujin Han",
"Difan Zou"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=Whr6uobelR | @inproceedings{
xu2023a,
title={A Sparsity Principle for Partially Observable Causal Representation Learning},
author={Danru Xu and Dingling Yao and Sebastien Lachapelle and Perouz Taslakian and Julius von K{\"u}gelgen and Francesco Locatello and Sara Magliacane},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=Whr6uobelR}
} | Causal representation learning (CRL) aims at identifying high-level causal variables from low-level data, e.g. images. Current methods usually assume that all causal variables are captured in the high-dimensional observations. In this work, we focus on learning causal representations from data under partial observability, i.e., when some of the causal variables are not observed in the measurements, and the set of masked variables changes across the different samples. We introduce some initial theoretical results for identifying causal variables under partial observability by exploiting a sparsity regularizer, focusing in particular on the linear and piecewise linear mixing function case. We provide a theorem that allows us to identify the causal variables up to permutation and element-wise linear transformations in the linear case and a lemma that allows us to identify causal variables up to linear transformation in the piecewise case. Finally, we provide a conjecture that would allow us to identify the causal variables up to permutation and element-wise linear transformations also in the piecewise linear case.
We test the theorem and conjecture on simulated data, showing the effectiveness of our method. | A Sparsity Principle for Partially Observable Causal Representation Learning | [
"Danru Xu",
"Dingling Yao",
"Sebastien Lachapelle",
"Perouz Taslakian",
"Julius von Kügelgen",
"Francesco Locatello",
"Sara Magliacane"
] | Workshop/CRL | 2403.08335 | [
"https://github.com/danrux/sparsity-crl"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=VJjTq7QmRv | @inproceedings{
baeg2023mixupbased,
title={Mixup-Based Knowledge Distillation with Causal Intervention for Multi-Task Speech Classification},
author={Kwangje Baeg and Hyeopwoo Lee and Yeomin Yoon and Jongmo Kim},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=VJjTq7QmRv}
} | Speech classification is an essential yet challenging subtask of multitask classification, which determines the gender and age groups of speakers. Existing methods face challenges while extracting the correct features indicative of some age groups that have several ambiguities of age perception in speech. Furthermore, the methods cannot fully understand the causal inferences between speech representation and multilabel spaces. In this study, the causes of ambiguous age group boundaries are attributed to the considerable variability in speech, even within the same age group. Additionally, features that indicate speech from the 20’s can be shared by some age groups in their 30’s. Therefore, a two-step approach to (1) mixup-based knowledge distillation to remove biased knowledge with causal intervention and (2) hierarchical multi-task learning with causal inference for the age group hierarchy to utilize the shared information of label dependencies is proposed. Empirical experiments on Korean open-set speech corpora demonstrate that the proposed methods yield a significant performance boost in multitask speech classification. | Mixup-Based Knowledge Distillation with Causal Intervention for Multi-Task Speech Classification | [
"Kwangje Baeg",
"Hyeopwoo Lee",
"Yeomin Yoon",
"Jongmo Kim"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=TIIufkRcYh | @inproceedings{
nalmpantis2023hierarchical,
title={Hierarchical Causal Representation Learning},
author={Angelos Nalmpantis and Phillip Lippe and Sara Magliacane},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=TIIufkRcYh}
} | Learning causal representations is a crucial step toward understanding and reasoning about an agent's actions in embodied AI and reinforcement learning. In many scenarios, an intelligent agent starts learning to interact with an environment by initially performing coarse actions with multiple simultaneous effects. During the learning process, the agent starts acquiring more fine-grained skills that can now affect only some of the factors in the environment. This setting is currently underexplored in current causal representation learning methods that typically learn a single causal representation and do not reuse or refine previously learned representations.
In this paper, we introduce the problem of hierarchical causal representation learning, which leverages causal representations learned with coarse interactions and progressively refines them, as more fine-grained interactions become available.
We propose HERCULES, a method that builds a hierarchical structure where at each level it gradually identifies more fine-grained causal variables by leveraging increasingly refined interventions. In experiments on two benchmarks of sequences of images with intervened causal factors, we demonstrate that HERCULES successfully recovers the causal factors of the underlying system and outperforms current state-of-the-art methods in scenarios with limited fine-grained data. At the same time, the acquired representations of HERCULES exhibit great adaptation capabilities under local transformations of the causal factors. | Hierarchical Causal Representation Learning | [
"Angelos Nalmpantis",
"Phillip Lippe",
"Sara Magliacane"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=T0PoOJg8cK | @inproceedings{
park2023the,
title={The Linear Representation Hypothesis and the Geometry of Large Language Models},
author={Kiho Park and Yo Joong Choe and Victor Veitch},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=T0PoOJg8cK}
} | Informally, the 'linear representation hypothesis' is the idea that high-level concepts are represented linearly as directions in some representation space.
In this paper, we address two closely related questions: What does "linear representation" actually mean? And, how do we make sense of geometric notions (e.g., cosine similarity or projection) in the representation space?
To answer these, we use the language of counterfactuals to give two formalizations of "linear representation", one in the output (word) representation space, and one in the input (sentence) space. We then prove these connect to linear probing and model steering, respectively.
To make sense of geometric notions, we use the formalization to identify a particular (non-Euclidean) inner product that respects language structure in a sense we make precise.
Using this causal inner product, we show how to unify all notions of linear representation.
In particular, this allows the construction of probes and steering vectors using counterfactual pairs.
Experiments with LLaMA-2 demonstrate the existence of linear representations of concepts, the connection to interpretation and control, and the fundamental role of the choice of inner product.
Code is available at github.com/KihoPark/linear\_rep\_geometry. | The Linear Representation Hypothesis and the Geometry of Large Language Models | [
"Kiho Park",
"Yo Joong Choe",
"Victor Veitch"
] | Workshop/CRL | 2311.03658 | [
"https://github.com/kihopark/linear_rep_geometry"
] | https://huggingface.co/papers/2311.03658 | 0 | 1 | 0 | 3 | [] | [] | [] | [] | [] | [] | 1 | oral |
null | https://openreview.net/forum?id=SvOXIxPWcM | @inproceedings{
patil2023debiasing,
title={Debiasing Multimodal Models via Causal Information Minimization},
author={Vaidehi Patil and Adyasha Maharana and Mohit Bansal},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=SvOXIxPWcM}
} | Most existing debiasing methods for multimodal models, including causal intervention and inference methods, utilize approximate heuristics to represent the biases, such as shallow features from early stages of training or unimodal features for multimodal tasks like VQA, etc., which may not be accurate. In this paper, we study bias arising from confounders in a causal graph for multimodal data, and examine a novel approach that leverages causally-motivated information minimization to learn the confounder representations. Robust predictive features contain diverse information that helps a model generalize to out-of-distribution data. Hence, minimizing the information content of features obtained from a pretrained biased model helps learn the simplest predictive features that capture the underlying data distribution. We treat these features as confounder representations and use them via methods motivated by causal theory to remove bias from models. We find that the learned confounder representations indeed capture dataset biases and the proposed debiasing methods improve out-of-distribution (OOD) performance on multiple multimodal datasets without sacrificing in-distribution performance. | Debiasing Multimodal Models via Causal Information Minimization | [
"Vaidehi Patil",
"Adyasha Maharana",
"Mohit Bansal"
] | Workshop/CRL | 2311.16941 | [
"https://github.com/vaidehi99/causalinfomin"
] | https://huggingface.co/papers/2311.16941 | 1 | 1 | 0 | 3 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=RS3ulfWvbo | @inproceedings{
wei2023unfairness,
title={Unfairness Detection within Power Systems through Transfer Counterfactual Learning},
author={Song Wei and Xiangrui Kong and Sarah Ann Huestis-Mitchell and Yao Xie and Shixiang Zhu and Alinson Santos Xavier and Feng Qiu},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=RS3ulfWvbo}
} | Energy justice is a growing area of interest in interdisciplinary energy research. However, identifying systematic biases in the energy sector remains challenging due to confounding variables, intricate heterogeneity in treatment effects, and limited data availability. To address these challenges, we introduce a novel approach for counterfactual causal analysis centered on energy justice. We use subgroup analysis to manage diverse factors and leverage the idea of transfer learning to mitigate data scarcity in each subgroup. In our numerical analysis, we apply our method to a large-scale customer-level power outage data set and investigate the counterfactual effect of demographic factors, such as income and age of the population, on power outage durations. Our results indicate that low-income and elderly-populated areas consistently experience longer power outages, regardless of weather conditions. This points to existing biases in the power system and highlights the need for focused improvements in areas with economic challenges. | Unfairness Detection within Power Systems through Transfer Counterfactual Learning | [
"Song Wei",
"Xiangrui Kong",
"Sarah Ann Huestis-Mitchell",
"Yao Xie",
"Shixiang Zhu",
"Alinson Santos Xavier",
"Feng Qiu"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=QBrGSFq0vz | @inproceedings{
singh2023choice,
title={Choice Models and Permutation Invariance: Demand Estimation in Differentiated Products Markets},
author={Amandeep Singh and Ye Liu and Hema Yoganarasimhan},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=QBrGSFq0vz}
} | Choice Modeling is at the core of many economics, operations, and marketing problems. In this paper, we propose a fundamental characterization of choice functions that encompasses a wide variety of extant choice models. We demonstrate how non-parametric estimators like neural nets can easily approximate such functionals and overcome the curse of dimensionality that is inherent in the non-parametric estimation of choice functions. We demonstrate through extensive simulations that our proposed functionals can flexibly capture underlying consumer behavior in a completely data-driven fashion and outperform traditional parametric models. As demand settings often exhibit endogenous features, we extend our framework to incorporate estimation under endogenous features. Further, we also describe a formal inference procedure to construct valid confidence intervals on objects of interest like price elasticity. Finally, to assess the practical applicability of our estimator, we utilize a real-world dataset from \cite{berry1995automobile}. Our empirical analysis confirms that the estimator generates realistic and comparable own- and cross-price elasticities that are consistent with the observations reported in the existing literature. | Choice Models and Permutation Invariance: Demand Estimation in Differentiated Products Markets | [
"Amandeep Singh",
"Ye Liu",
"Hema Yoganarasimhan"
] | Workshop/CRL | 2307.07090 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=PbPLk5OcAK | @inproceedings{
ghosh2023independent,
title={Independent Mechanism Analysis and the Manifold Hypothesis},
author={Shubhangi Ghosh and Luigi Gresele and Julius von K{\"u}gelgen and Michel Besserve and Bernhard Sch{\"o}lkopf},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=PbPLk5OcAK}
} | Independent Mechanism Analysis (IMA) seeks to address non-identifiability in nonlinear Independent Component Analysis (ICA) by assuming that the Jacobian of the mixing function has orthogonal columns. As typical in ICA, previous work focused on the case with an equal number of latent components and observed mixtures. Here, we extend IMA to settings with a larger number of mixtures that reside on a manifold embedded in a higher-dimensional space—in line with the _manifold hypothesis_ in representation learning. For this setting, we show that IMA still circumvents several non-identifiability issues, suggesting that it can also be a beneficial principle for higher-dimensional observations when the manifold hypothesis holds. Further, we prove that the IMA principle is approximately satisfied with high probability (increasing with the number of observed mixtures) when the directions along which the latent components influence the observations are chosen independently at random. This provides a new and rigorous statistical interpretation of IMA. | Independent Mechanism Analysis and the Manifold Hypothesis | [
"Shubhangi Ghosh",
"Luigi Gresele",
"Julius von Kügelgen",
"Michel Besserve",
"Bernhard Schölkopf"
] | Workshop/CRL | 2312.13438 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=O9jfSs82XU | @inproceedings{
rajwade2023cellsvec,
title={Cells2Vec: Bridging the gap between experiments and simulations using causal representation learning},
author={Dhruva Rajwade and Atiyeh Ahmadi and Brian Paul Ingalls},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=O9jfSs82XU}
} | Calibration of computational simulations of biological dynamics against experimental observations is often a challenge. In particular, the selection of features that can be used to construct a goodness-of-fit function for agent-based models of spatiotemporal behaviour can be difficult (Yip et al. (2022)). In this study, we generate one-dimensional embeddings of high-dimensional simulation outputs using causal dilated convolutions for encoding and a triplet loss-based training strategy. We verify the robustness of the trained encoder using simulations generated by unseen input parameter sets. Furthermore, we use the generated embeddings to estimate the parameters of simulations using XGBoost Regression. We demonstrate the results of parameter estimation for corresponding time-series real-world experimental observations, identifying a causal relationship between simulation-specific input parameters and real-world experiments. Our regression approach is able to estimate simulation parameters with an average $R^2$ metric of 0.90 for model runs with embedding dimensions of 4,8,12 and 16. Model calibration led to simulations with an average cosine similarity agreement of 0.95 and an average normalized Euclidean similarity of 0.69 with real-world experiments over multiple model runs. | Cells2Vec: Bridging the gap between experiments and simulations using causal representation learning | [
"Dhruva Rajwade",
"Atiyeh Ahmadi",
"Brian Paul Ingalls"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=MytNJ6lXAV | @inproceedings{
varici2023scorebased,
title={Score-based Causal Representation Learning from Interventions: Nonparametric Identifiability},
author={Burak Varici and Emre Acart{\"u}rk and Karthikeyan Shanmugam and Ali Tajer},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=MytNJ6lXAV}
} | This paper focuses on causal representation learning (CRL) under a general nonparametric causal latent model and a general transformation model that maps the latent data to the observational data. It establishes **identifiability** and **achievability** results using two (stochastic) hard **uncoupled** interventions per node in the latent causal graph. Notably, one does not know which pair of intervention environments have the same node intervened (hence, uncoupled environments). For identifiability, the paper establishes that perfect recovery of the latent causal model and variables is guaranteed under uncoupled interventions. For achievability, an algorithm is designed that uses observational and interventional data and recovers the latent causal model and variables with provable guarantees for the algorithm. This algorithm leverages score variations across different environments to estimate the inverse of the transformer and, subsequently, the latent variables. The analysis, additionally, recovers the existing identifiability result for two hard **coupled** interventions, that is when metadata about the pair of environments that have the same node intervened is known. It is noteworthy that the existing results on non-parametric identifiability require assumptions on interventions and additional faithfulness assumptions. This paper shows that when observational data is available, additional faithfulness assumptions are unnecessary. | Score-based Causal Representation Learning from Interventions: Nonparametric Identifiability | [
"Burak Varici",
"Emre Acartürk",
"Karthikeyan Shanmugam",
"Ali Tajer"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | oral |
||
null | https://openreview.net/forum?id=LtkQY7PJVB | @inproceedings{
ahuja2023multidomain,
title={Multi-Domain Causal Representation Learning via Weak Distributional Invariances},
author={Kartik Ahuja and Amin Mansouri and Yixin Wang},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=LtkQY7PJVB}
} | Causal representation learning has emerged as the center of action in causal machine learning research. In particular, multi-domain datasets present a natural opportunity for showcasing the advantages of causal representation learning over standard unsupervised representation learning. While recent works have taken crucial steps towards learning causal representations, they often lack applicability to multi-domain datasets due to over-simplifying assumptions about the data; e.g. each domain comes from a different single-node perfect intervention. In this work, we relax these assumptions and capitalize on the following observation: there often exists a subset of latents whose certain distributional properties (e.g., support, variance) remain stable across domains (e.g., when each domain comes from a multi-node imperfect intervention). Leveraging this observation, we show that autoencoders that incorporate such invariances can provably identify the stable set of latents from the rest in a host of different settings. | Multi-Domain Causal Representation Learning via Weak Distributional Invariances | [
"Kartik Ahuja",
"Amin Mansouri",
"Yixin Wang"
] | Workshop/CRL | 2310.02854 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=KqMKDMw2Iv | @inproceedings{
wu2023counterfactual,
title={Counterfactual Generative Models for Time-Varying Treatments},
author={Shenghao Wu and Wenbin Zhou and Minshuo Chen and Shixiang Zhu},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=KqMKDMw2Iv}
} | Estimating the counterfactual outcome of treatment is essential for decision-making in public health and clinical science, among others. Often, treatments are administered in a sequential, time-varying manner, leading to an exponentially increased number of possible counterfactual outcomes. Furthermore, in modern applications, the outcomes are high-dimensional and conventional average treatment effect estimation fails to capture disparities in individuals. To tackle these challenges, we propose a novel conditional generative framework capable of producing counterfactual samples under time-varying treatment, without the need for explicit density estimation. Our method carefully addresses the distribution mismatch between the observed and counterfactual distributions via a loss function based on inverse probability weighting. We present a thorough evaluation of our method using both synthetic and real-world data. Our results demonstrate that our method is capable of generating high-quality counterfactual samples and outperforms the state-of-the-art baselines. | Counterfactual Generative Models for Time-Varying Treatments | [
"Shenghao Wu",
"Wenbin Zhou",
"Minshuo Chen",
"Shixiang Zhu"
] | Workshop/CRL | 2305.15742 | [
"https://github.com/shenghaowu/counterfactual-generative-models"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=JoISqbH8vl | @inproceedings{
eastwood2023selfsupervised,
title={Self-Supervised Disentanglement by Leveraging Structure in Data Augmentations},
author={Cian Eastwood and Julius von K{\"u}gelgen and Linus Ericsson and Diane Bouchacourt and Pascal Vincent and Mark Ibrahim and Bernhard Sch{\"o}lkopf},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=JoISqbH8vl}
} | Self-supervised representation learning often uses data augmentations to induce some invariance to "style" attributes of the data. However, with downstream tasks generally unknown at training time, it is difficult to deduce a priori which attributes of the data are indeed "style" and can be safely discarded. To address this, we introduce a more principled approach that seeks to disentangle style features rather than discard them. The key idea is to add multiple style embedding spaces where: (i) each is invariant to all-but-one augmentation; and (ii) joint entropy is maximized. We formalize our structured data-augmentation procedure from a causal latent-variable-model perspective, and prove identifiability of both content and (multiple blocks of) style variables. We empirically demonstrate the benefits our approach on synthetic datasets and then present promising but limited results on ImageNet. | Self-Supervised Disentanglement by Leveraging Structure in Data Augmentations | [
"Cian Eastwood",
"Julius von Kügelgen",
"Linus Ericsson",
"Diane Bouchacourt",
"Pascal Vincent",
"Mark Ibrahim",
"Bernhard Schölkopf"
] | Workshop/CRL | 2311.08815 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=HAymeESPKo | @inproceedings{
wu2023objectcentric,
title={Object-Centric Semantic Vector Quantization},
author={Yi-Fu Wu and Minseung Lee and Sungjin Ahn},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=HAymeESPKo}
} | Neural discrete representations are crucial components of modern neural networks. However, their main limitation is that the primary strategies such as VQ-VAE can only provide representations at the patch level. Therefore, one of the main goals of representation learning, acquiring conceptual, semantic, and compositional abstractions such as the color and shape of an object, remains elusive. In this paper, we present the first approach to semantic neural discrete representation learning. The proposed model, called Semantic Vector-Quantized Variational Autoencoder (SVQ), leverages recent advances in unsupervised object-centric learning to address this limitation. Specifically, we observe that a simple approach quantizing at the object level poses a significant challenge and propose constructing scene representations hierarchically, from low-level discrete concept schemas to object representations. Additionally, we suggest a novel method for training a prior over these semantic representations, enabling the ability to generate images following the underlying data distribution, which is lacking in most object-centric models. In experiments on various 2D and 3D object-centric datasets, we find that our model achieves superior generation performance compared to non-semantic vector quantization methods such as VQ-VAE and previous object-centric generative models. Furthermore, we find that the semantic discrete representations can solve downstream scene understanding tasks that require reasoning about the properties of different objects in the scene. | Object-Centric Semantic Vector Quantization | [
"Yi-Fu Wu",
"Minseung Lee",
"Sungjin Ahn"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=H3yd4W0vCM | @inproceedings{
talon2023towards,
title={Towards the Reusability and Compositionality of Causal Representations},
author={Davide Talon and Phillip Lippe and Stuart James and Alessio Del Bue and Sara Magliacane},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=H3yd4W0vCM}
} | Causal Representation Learning (CRL) aims at identifying high-level causal factors and their relationships from high-dimensional observations, e.g., images. While most CRL works focus on learning causal representations in a single environment, in this work we instead propose a first step towards learning causal representations from temporal sequences of images that can be adapted in a new environment, or composed across multiple related environments. In particular, we introduce DECAF, a framework that detects which causal factors can be reused and which need to be adapted from previously learned causal representations. Our approach is based on the availability of intervention targets, that indicate which variables are perturbed at each time step.
Experiments on three benchmark datasets show that integrating our framework with four state-of-the-art CRL approaches leads to accurate representations in a new environment with only a few samples. | Towards the Reusability and Compositionality of Causal Representations | [
"Davide Talon",
"Phillip Lippe",
"Stuart James",
"Alessio Del Bue",
"Sara Magliacane"
] | Workshop/CRL | 2403.09830 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=GaoLn9UI1G | @inproceedings{
mansouri2023objectcentric,
title={Object-centric architectures enable efficient causal representation learning},
author={Amin Mansouri and Jason Hartford and Yan Zhang and Yoshua Bengio},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=GaoLn9UI1G}
} | Causal representation learning has showed a variety of settings in which we can disentangle latent variables with identifiability guarantees (up to some reasonable equivalence class). Common to all of these approaches is the assumption that (1) the latent variables are represented as
$d-$dimensional vectors, and (2) that the observations are the output of some injective generative function of these latent variables. While these assumptions appear benign, we show that when the observations are of multiple objects, the generative function is no longer injective and disentanglement fails in practice. We can address this failure by combining recent developments in object-centric learning and causal representation learning. By modifying the Slot Attention architecture (Locatello et al., 2020), we develop an object-centric architecture that leverages weak supervision from sparse perturbations to disentangle each object's properties. This approach is more data-efficient in the sense that it requires significantly fewer perturbations than a comparable approach that encodes to a Euclidean space and we show that this approach successfully disentangles the properties of a set of objects in a series of simple image-based disentanglement experiments. | Object-centric architectures enable efficient causal representation learning | [
"Amin Mansouri",
"Jason Hartford",
"Yan Zhang",
"Yoshua Bengio"
] | Workshop/CRL | 2310.19054 | [
"https://github.com/amansouri3476/oc-crl"
] | https://huggingface.co/papers/2310.19054 | 1 | 1 | 0 | 4 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=FVq8SSco3q | @inproceedings{
zhou2023rewardrelevancefiltered,
title={Reward-Relevance-Filtered Linear Offline Reinforcement Learning},
author={Angela Zhou},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=FVq8SSco3q}
} | This paper studies causal variable selection in the setting of a Markov decision process, specifically offline reinforcement learning with linear function approximation. The structural restrictions of the data-generating process presume that the transitions factor into sparse dynamics that affect the reward, and additional exogenous dynamics that do not affect the reward. Although the minimally sufficient adjustment set for estimation of full-state transition properties depends on the whole state, the optimal policy and therefore state-action value function is sparse. This is a novel "causal sparsity" notion that does not occur in pure estimation settings. We develop methods for filtering the estimation of the state-action value function to the sparse component by a modification of thresholded lasso: we use thresholded lasso to recover the support of the rewards, and use this estimated support to estimate the state-action $Q$ function. Such a method has sample complexity depending only on the size of the sparse component. Although this problem differs from the typical statement of "causal representation learning", this notion of "causal sparsity" may be of interest, and our methods connect to a classical statistical literature with theoretical guarantees that can be a stepping stone for more complex representation learning. | Reward-Relevance-Filtered Linear Offline Reinforcement Learning | [
"Angela Zhou"
] | Workshop/CRL | 2401.12934 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=FNlfCBZ1tD | @inproceedings{
karlsson2023putting,
title={Putting Causal Identification to the Test: Falsification using Multi-Environment Data},
author={Rickard Karlsson and Ștefan Creast{\u{a}} and JH Krijthe},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=FNlfCBZ1tD}
} | We study the problem of falsifying the assumptions behind a set of broadly applied causal identification strategies: namely back-door adjustment, front-door adjustment, and instrumental variable estimation. While these assumptions are untestable from observational data in general, we show that with access to data coming from multiple heterogeneous environments, there exist novel independence constraints that can be used to falsify the validity of each strategy. Most interestingly, we make no parametric assumptions, instead relying on that changes between environments happen under the principle of independent causal mechanisms. | Putting Causal Identification to the Test: Falsification using Multi-Environment Data | [
"Rickard Karlsson",
"Ștefan Creastă",
"JH Krijthe"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=E8IhOxNREv | @inproceedings{
yao2023multiview,
title={Multi-View Causal Representation Learning with Partial Observability},
author={Dingling Yao and Danru Xu and Sebastien Lachapelle and Sara Magliacane and Perouz Taslakian and Georg Martius and Julius von K{\"u}gelgen and Francesco Locatello},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=E8IhOxNREv}
} | We present a unified framework for studying the identifiability of representations learned from simultaneously observed views, such as different data modalities. We allow a partially observed setting in which each view constitutes a nonlinear mixture of a subset of underlying latent variables, which can be causally related. We prove that the information shared across all subsets of any number of views can be learned up to a smooth bijection using contrastive learning and a single encoder per view. We also provide graphical criteria indicating which latent variables can be identified through a simple set of rules, which we refer to as identifiability algebra. Our general framework and theoretical results unify and extend several previous works on multi-view nonlinear ICA, disentanglement, and causal representation learning. We experimentally validate our claims on numerical, image, and multi-modal data sets. Further, we demonstrate that the performance of prior methods is recovered in different special cases of our setup. Overall, we find that access to multiple partial views enables identifying a more fine-grained representation, under the generally milder assumption of partial observability. | Multi-View Causal Representation Learning with Partial Observability | [
"Dingling Yao",
"Danru Xu",
"Sebastien Lachapelle",
"Sara Magliacane",
"Perouz Taslakian",
"Georg Martius",
"Julius von Kügelgen",
"Francesco Locatello"
] | Workshop/CRL | 2311.04056 | [
"https://github.com/causallearningai/multiview-crl"
] | https://huggingface.co/papers/2311.04056 | 1 | 0 | 0 | 8 | [] | [] | [] | [] | [] | [] | 1 | oral |
null | https://openreview.net/forum?id=DyitJM9vAE | @inproceedings{
bing2023invariance,
title={Invariance \& Causal Representation Learning: Prospects and Limitations},
author={Simon Bing and Jonas Wahl and Urmi Ninad and Jakob Runge},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=DyitJM9vAE}
} | In causal models, a given mechanism is assumed to be invariant to changes of other mechanisms. While this principle has been utilized for inference in settings where the causal variables are observed, theoretical insights when the variables of interest are latent are largely missing. We assay the connection between invariance and causal representation learning by establishing impossibility results which show that invariance alone is insufficient to identify latent causal variables. Together with practical considerations, we use these theoretical findings to highlight the need for additional constraints in order to identify representations by exploiting invariance. | Invariance Causal Representation Learning: Prospects and Limitations | [
"Simon Bing",
"Jonas Wahl",
"Urmi Ninad",
"Jakob Runge"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=BQpfwa0NNJ | @inproceedings{
farzam2023curvature,
title={Curvature and Causal Inference in Network Data},
author={Amirhossein Farzam and Allen Tannenbaum and Guillermo Sapiro},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=BQpfwa0NNJ}
} | Learning causal mechanisms involving networked units of data is a notoriously challenging task with various applications. Graph Neural Networks (GNNs) have proven to be effective for learning representations that capture complex dependencies between data units. This effectiveness is largely due to the conduciveness of GNNs to tools that characterize the geometry of graphs. The potential of geometric deep learning for GNN-based causal representation learning, however, remains underexplored. This work makes three key contributions to bridge this gap. First, we establish a theoretical connection between graph curvature and causal inference, showing that negative curvatures pose challenges to learning the causal mechanisms underlying network data. Second, based on this theoretical insight, we present empirical results using the Ricci curvature to gauge the error in treatment effect estimates made from representations learned by GNNs. This empirically demonstrates that positive curvature regions yield more accurate results. Lastly, as an example of the potentials unleashed by this newfound connection between geometry and causal inference, we propose a method using Ricci flow to improve the treatment effect estimation on networked data. Our experiments confirm that this method reduces the error in treatment effect estimates by flattening the network, showcasing the utility of geometric methods for enhancing causal representation learning. Our findings open new avenues for leveraging discrete geometry in causal representation learning, offering insights and tools that enhance the performance of GNNs in learning robust structural relationships. | Curvature and Causal Inference in Network Data | [
"Amirhossein Farzam",
"Allen Tannenbaum",
"Guillermo Sapiro"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=9cDEglKlj5 | @inproceedings{
lorch2023causal,
title={Causal Modeling with Stationary Diffusions},
author={Lars Lorch and Andreas Krause and Bernhard Sch{\"o}lkopf},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=9cDEglKlj5}
} | We develop a novel approach towards causal inference. Rather than structural equations over a causal graph, we learn stochastic differential equations (SDEs) whose stationary densities model a system's behavior under interventions. These stationary diffusion models do not require the formalism of causal graphs, let alone the common assumption of acyclicity. We show that in several cases, they generalize to unseen interventions on their variables, often better than classical approaches. Our inference method is based on a new theoretical result that expresses a stationarity condition on the diffusion's generator in a reproducing kernel Hilbert space. The resulting kernel deviation from stationarity (KDS) is an objective function of independent interest. | Causal Modeling with Stationary Diffusions | [
"Lars Lorch",
"Andreas Krause",
"Bernhard Schölkopf"
] | Workshop/CRL | 2310.17405 | [
"https://github.com/larslorch/stadion"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=8btzvHmSfU | @inproceedings{
cho2023learning,
title={Learning to ignore: Single Source Domain Generalization via Oracle Regularization},
author={Dong Kyu Cho and Sanghack Lee},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=8btzvHmSfU}
} | Machine learning frequently suffers from the discrepancy in data distribution, commonly known as domain shift. Single-source Domain Generalization (sDG) is a task designed to simulate domain shift artificially, in order to train a model that can generalize well to multiple unseen target domains from a single source domain. A popular approach is to learn robustness via the alignment of augmented samples. However, prior works frequently overlooked what is learned from such alignment. In this paper, we study the effectiveness of augmentation-based sDG methods via a causal interpretation of the data generating process. We highlight issues in using augmentation for generalization, namely, the distinction between domain invariance and augmentation invariance. To alleviate these issues, we introduce a novel regularization method that leverages pretrained models to guide the learning process via a feature-level regularization, which we name PROF (Progressive mutual information Regularization for Online distillation of Frozen oracles). PROF can be applied to conventional augmentation-based methods to moderate the impact of stochasticity in models repeatedly trained on augmented data, encouraging the model to learn domain-invariant representations. We empirically show that PROF stabilizes the learning process for sDG. | Learning to ignore: Single Source Domain Generalization via Oracle Regularization | [
"Dong Kyu Cho",
"Sanghack Lee"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=6vmxURfudf | @inproceedings{
wang2023instancedependent,
title={Instance-Dependent Partial Label Learning with Identifiable Causal Representations},
author={Yizhi Wang and Weijia Zhang and Min-Ling Zhang},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=6vmxURfudf}
} | Partial label learning (PLL) deals with the problem where each training example is annotated with a set of candidate labels, among which only one is true. In real-world scenarios, the candidate labels are generally dependent to the instance features. However, existing PLL methods focus solely on classification accuracy, whereas the possibility of exploiting the dependency for causal representation learning remains unexplored. In this paper, we investigate learning causal representations under the PLL paradigm and propose a novel framework which learns identifiable latent factors up to permutation, scaling and translation. Qualitative and quantitative experiments confirmed the effectiveness of this approach. | Instance-Dependent Partial Label Learning with Identifiable Causal Representations | [
"Yizhi Wang",
"Weijia Zhang",
"Min-Ling Zhang"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=6YIX9sc7Xt | @inproceedings{
yin2023causal,
title={Causal Markov Blanket Representation Learning for Out-of-distribution Generalization},
author={Naiyu Yin and Hanjing Wang and Tian Gao and Amit Dhurandhar and Qiang Ji},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=6YIX9sc7Xt}
} | The pursuit of generalizable representations in the realm of machine learning and computer vision is a dynamic field of research. Typically, current methods aim to secure invariant representations by either harnessing domain expertise or leveraging data from multiple domains. In this paper, we introduce a novel approach that involves acquiring Causal Markov Blanket (CMB) representations to improve prediction performance in the face of distribution shifts. Causal Markov Blanket representations comprise the direct causes and effects of the target variable. Theoretical analyses have demonstrated their capacity to harbor maximum information about the target, resulting in minimal Bayes error during prediction. To elaborate, our approach commences with the introduction of a novel structural causal model (SCM) equipped with latent representations, designed to capture the underlying causal mechanisms governing the data generation process. Subsequently, we propose a CMB representation learning framework that derives representations conforming to the proposed SCM. In comparison to state-of-the-art domain generalization methods, our approach exhibits robustness and adaptability under distribution shifts. | Causal Markov Blanket Representation Learning for Out-of-distribution Generalization | [
"Naiyu Yin",
"Hanjing Wang",
"Tian Gao",
"Amit Dhurandhar",
"Qiang Ji"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=5l1OdD81W4 | @inproceedings{
saengkyongam2023identifying,
title={Identifying Representations for Intervention Extrapolation},
author={Sorawit Saengkyongam and Elan Rosenfeld and Pradeep Kumar Ravikumar and Niklas Pfister and Jonas Peters},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=5l1OdD81W4}
} | The premise of identifiable and causal representation learning is to improve the current representation learning paradigm in terms of generalizability or robustness. Despite recent progress in questions of identifiability, more theoretical results demonstrating concrete advantages of these methods for downstream tasks are needed. In this paper, we consider the task of intervention extrapolation: predicting how interventions affect an outcome, even when those interventions are not observed at training time, and show that identifiable representations can provide an effective solution to this task even if the interventions affect the outcome non-linearly. Our setup includes an outcome variable $Y$, observed features $X$, which are generated as a non-linear transformation of latent features $Z$, and exogenous action variables $A$, which influence $Z$. The objective of intervention extrapolation is then to predict how interventions on $A$ that lie outside the training support of $A$ affect $Y$. Here, extrapolation becomes possible if the effect of $A$ on $Z$ is linear and the residual when regressing Z on A has full support. As $Z$ is latent, we combine the task of intervention extrapolation with identifiable representation learning, which we call $\texttt{Rep4Ex}$: we aim to map the observed features $X$ into a subspace that allows for non-linear extrapolation in $A$. We show that the hidden representation is identifiable up to an affine transformation in $Z$-space, which, we prove, is sufficient for intervention extrapolation. The identifiability is characterized by a novel constraint describing the linearity assumption of $A$ on $Z$. Based on this insight, we propose a flexible method that enforces the linear invariance constraint and can be combined with any type of autoencoder. We validate our theoretical findings through a series of synthetic experiments and show that our approach can indeed succeed in predicting the effects of unseen interventions. | Identifying Representations for Intervention Extrapolation | [
"Sorawit Saengkyongam",
"Elan Rosenfeld",
"Pradeep Kumar Ravikumar",
"Niklas Pfister",
"Jonas Peters"
] | Workshop/CRL | 2310.04295 | [
""
] | https://huggingface.co/papers/2310.04295 | 1 | 0 | 0 | 5 | [] | [] | [] | [] | [] | [] | 1 | oral |
null | https://openreview.net/forum?id=31aOhyoQRI | @inproceedings{
liu2023learning,
title={Learning Causally-Aware Representations of Multi-Agent Interactions},
author={Yuejiang Liu and Ahmad Rahimi and Po-Chien Luan and Frano Raji{\v{c}} and Alexandre Alahi},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=31aOhyoQRI}
} | Modeling spatial-temporal interactions between neighboring agents is at the heart of multi-agent problems such as motion forecasting and crowd navigation. Despite notable progress, it remains unclear to which extent modern representations can capture the causal relationships behind agent interactions. In this work, we take an in-depth look at the causal awareness of the learned representations, from computational formalism to controlled simulations to real-world practice. First, we cast doubt on the notion of non-causal robustness studied in the recent CausalAgents benchmark. We show that recent representations are already partially resilient to perturbations of non-causal agents, and yet modeling indirect causal effects involving mediator agents remains challenging. Further, we introduce a simple but effective regularization approach leveraging causal annotations of varying granularity. Through controlled experiments, we find that incorporating finer-grained causal annotations not only leads to higher degrees of causal awareness but also yields stronger out-of-distribution robustness. Finally, we extend our method to a sim-to-real causal transfer framework by means of cross-domain multi-task learning, which boosts generalization in practical settings even without real-world annotations. We hope our work provides more clarity to the challenges and opportunities of learning causally-aware representations in the multi-agent context while making a first step towards a practical solution. | Learning Causally-Aware Representations of Multi-Agent Interactions | [
"Yuejiang Liu",
"Ahmad Rahimi",
"Po-Chien Luan",
"Frano Rajič",
"Alexandre Alahi"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=1gY17EWQyl | @inproceedings{
kori2023a,
title={A Causal Ordering Prior for Unsupervised Representation Learning},
author={Avinash Kori and Pedro Sanchez and Konstantinos Vilouras and Ben Glocker and Sotirios A. Tsaftaris},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=1gY17EWQyl}
} | Unsupervised representation learning with variational inference relies heavily on independence assumptions over latent variables. Causal representation learning (CRL), however, argues that factors of variation in a dataset are, in fact, causally related. Allowing latent variables to be correlated, as a consequence of causal relationships, is more realistic and generalisable. So far, provably identifiable methods rely on: auxiliary information, weak labels, and interventional or even counterfactual data. Inspired by causal discovery with functional causal models, we propose a fully unsupervised representation learning method that considers a data generation process with a latent additive noise model (ANM). We encourage the latent space to follow a causal ordering via loss function based on the Hessian of the latent distribution. | A Causal Ordering Prior for Unsupervised Representation Learning | [
"Avinash Kori",
"Pedro Sanchez",
"Konstantinos Vilouras",
"Ben Glocker",
"Sotirios A. Tsaftaris"
] | Workshop/CRL | 2307.05704 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=0man1UezvH | @inproceedings{
sridhar2023learning,
title={Learning Macro Variables with Auto-encoders},
author={Dhanya Sridhar and Eric Elmoznino and Maitreyi Swaroop},
booktitle={Causal Representation Learning Workshop at NeurIPS 2023},
year={2023},
url={https://openreview.net/forum?id=0man1UezvH}
} | Most causal variables that we reason over, in both science and everyday life, are coarse abstractions of low-level data. However, despite their importance, the field of causality lacks a precise theory of abstract "macro" variables and their relation to low-level "micro" variables that can account for our intuitions. Here, we define a macro variable as something that (a) is simpler than its micro variable, (b) shares mutual information with its micro variable, and (c) is related to other macro variables via simple mechanisms. From this definition, we propose DeepCFL: a simple self-supervised method that learns macro variables and their relations. We empirically validate DeepCFL on synthetic tasks where the underlying macro variables are known, and find that they can be recovered with high fidelity. Given that the individual components of DeepCFL leverage standard and scalable techniques in deep learning, our preliminary results are encouraging signs that it can be successfully applied to real-world data. | Learning Macro Variables with Auto-encoders | [
"Dhanya Sridhar",
"Eric Elmoznino",
"Maitreyi Swaroop"
] | Workshop/CRL | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=xsyq6t10iI | @inproceedings{
liu2023navigating,
title={Navigating Beyond the Dead End: A Math Problem Solving Framework by Switching among Diverse Reasoning Thoughts},
author={Tengxiao Liu and Qipeng Guo and Yuqing Yang and Xiangkun Hu and Yue Zhang and Xipeng Qiu and Zheng Zhang},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=xsyq6t10iI}
} | As large language models (LLMs) have shown effectiveness with different prompting methods, such as Chain of Thought, Program of Thought, we find that these methods have formed a great complementarity to each other on math reasoning tasks. In this work, we propose XoT, an automatic problem solving framework by prompting LLMs with diverse reasoning thoughts. For each question, XoT always begins with selecting the most suitable method then executes each method iteratively. Within each iteration, XoT actively checks the validity of the generated answer and incorporates the feedback from external executors, allowing it to dynamically switch among different prompting methods. Through extensive experiments on 9 popular math reasoning datasets, we demonstrate the effectiveness of our proposed approach and thoroughly analyze the strengths of each module. Furthermore, empirical results suggest that our framework is orthogonal to recent work that makes improvements on single reasoning methods. By allowing method switching, XoT provides a fresh perspective on the collaborative integration of diverse reasoning thoughts in a unified framework. | Plan, Verify and Switch: Integrated Reasoning with Diverse X-of-Thoughts | [
"Tengxiao Liu",
"Qipeng Guo",
"Yuqing Yang",
"Xiangkun Hu",
"Yue Zhang",
"Xipeng Qiu",
"Zheng Zhang"
] | Workshop/MATH-AI | 2310.14628 | [
"https://github.com/tengxiaoliu/xot"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=woQ9G0bzBM | @inproceedings{
mehrabian2023finding,
title={Finding Increasingly Large Extremal Graphs with AlphaZero and Tabu Search},
author={Abbas Mehrabian and Ankit Anand and Hyunjik Kim and Nicolas Sonnerat and Tudor Berariu and Matej Balog and Gheorghe Comanici and Andrew Lee and Anian Ruoss and Anna Bulanova and Daniel Toyama and Sam Blackwell and Bernardino Romera Paredes and Laurent Orseau and Petar Veli{\v{c}}kovi{\'c} and Anurag Murty Naredla and Joonkyung Lee and Adam Zsolt Wagner and Doina Precup},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=woQ9G0bzBM}
} | This work studies a central extremal graph theory problem inspired by a 1975 conjecture of Erdős, which aims to find graphs with a given size (number of nodes) that maximize the number of edges without having 3- or 4-cycles. We formulate this problem as a sequential decision-making problem and compare AlphaZero, a neural network-guided tree search, with tabu search, a heuristic local search method. Using either method, by introducing a curriculum---jump-starting the search for larger graphs using good graphs found at smaller sizes---we improve the state-of-the-art lower bounds for several sizes. Lastly, we propose a flexible graph-generation environment and a permutation-invariant network architecture for learning to search in the space of graphs. | Finding Increasingly Large Extremal Graphs with AlphaZero and Tabu Search | [
"Abbas Mehrabian",
"Ankit Anand",
"Hyunjik Kim",
"Nicolas Sonnerat",
"Tudor Berariu",
"Matej Balog",
"Gheorghe Comanici",
"Andrew Lee",
"Anian Ruoss",
"Anna Bulanova",
"Daniel Toyama",
"Sam Blackwell",
"Bernardino Romera Paredes",
"Laurent Orseau",
"Petar Veličković",
"Anurag Murty Naredla",
"Joonkyung Lee",
"Adam Zsolt Wagner",
"Doina Precup"
] | Workshop/MATH-AI | 2311.03583 | [
""
] | https://huggingface.co/papers/2311.03583 | 1 | 0 | 0 | 19 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=vibHb75kYq | @inproceedings{
al-negheimish2023augmenting,
title={Augmenting Large Language Models with Symbolic Rule Learning for Robust Numerical Reasoning},
author={Hadeel Al-Negheimish and Pranava Madhyastha and Alessandra Russo},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=vibHb75kYq}
} | While some prompting strategies have been proposed to elicit reasoning in Large Language Models (LLMs), numerical reasoning for machine reading comprehension remains a difficult challenge.
We propose a neuro-symbolic approach that uses in-context learning with LLMs to decompose complex questions into simpler ones and symbolic learning methods to learn rules for recomposing partial answers.
We evaluate it on different numerical subsets of the DROP benchmark; results show that it is competitive with DROP-specific SOTA models and significantly improves results over pure LLM prompting methods.
Our approach boasts data efficiency, since it does not involve any additional training or fine-tuning. Additionally, the neuro-symbolic approach facilitates robust numerical reasoning; the model is faithful to the passage it has been presented, and provides interpretable and verifiable reasoning traces. | Augmenting Large Language Models with Symbolic Rule Learning for Robust Numerical Reasoning | [
"Hadeel Al-Negheimish",
"Pranava Madhyastha",
"Alessandra Russo"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=tEUJiua8ir | @inproceedings{
zhou2023understanding,
title={Understanding Length Generalization by Thinking Like Transformers},
author={Hattie Zhou and Arwen Bradley and Etai Littwin and Noam Razin and Omid Saremi and Joshua Susskind and Samy Bengio and Preetum Nakkiran},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=tEUJiua8ir}
} | Large language models exhibit surprising emergent generalization properties, yet also struggle on many simple reasoning tasks such as arithmetic and parity. In this work, we focus on length generalization, and we propose a unifying framework to understand when and how Transformers can be expected to length generalize on a given task. First, we show that there exist algorithmic tasks for which standard decoder-only Transformers trained from scratch naturally exhibit strong length generalization. For these tasks, we leverage the RASP programming language (Weiss et al., 2021) to show that the correct algorithmic solution which solves the task can be represented by a simple Transformer. We thus propose and give evidence for the RASP-Generalization Conjecture: Transformers tend to learn a length-generalizing solution if there exists a short RASP-L program that works for all input lengths. We then leverage our insights to give new scratchpad formats which yield strong length generalization on traditionally hard tasks (such as parity and addition). Overall, our work provides a novel perspective on the mechanisms of length generalization and the algorithmic capabilities of Transformers. | What Algorithms can Transformers Learn? A Study in Length Generalization | [
"Hattie Zhou",
"Arwen Bradley",
"Etai Littwin",
"Noam Razin",
"Omid Saremi",
"Joshua Susskind",
"Samy Bengio",
"Preetum Nakkiran"
] | Workshop/MATH-AI | 2310.16028 | [
""
] | https://huggingface.co/papers/2310.16028 | 1 | 2 | 0 | 8 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=sSgdyY0YJR | @inproceedings{
gloeckle2023temperaturescaled,
title={Temperature-scaled large language models for Lean proofstep prediction},
author={Fabian Gloeckle and Baptiste Roziere and Amaury Hayat and Gabriel Synnaeve},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=sSgdyY0YJR}
} | Leveraging the reasoning capabilities of large language models (LLMs) for theorem proving is a promising but challenging task because it requires in-domain finetunings on which LLMs are known to be prone to overfit. This issue is exacerbated by two properties that set theorem proving apart from more mainstream applications of LLMs: training data in formal environments like Lean or Isabelle is very scarce and evaluation benchmarks are prohibitively costly to be used extensively for hyperparameter search and model selection. In this work, we propose temperature scaling as a regularization method for multi-epoch training on small datasets. We explain its theoretical purpose heuristically and demonstrate its effectiveness empirically, obtaining state-of-the-art supervised tactic generation models for Lean 3 of sizes 1.5B, 7B and 13B parameters. Model selection based on temperature-scaled perplexity increases scores on theorem proving benchmarks by up to four percentage points. We provide detailed ablations and analyses of the proof search behaviors of the resulting models, allowing practitioners to pick optimal model sizes for their respective use cases. | Temperature-scaled large language models for Lean proofstep prediction | [
"Fabian Gloeckle",
"Baptiste Roziere",
"Amaury Hayat",
"Gabriel Synnaeve"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=rTz88hpGxc | @inproceedings{
hersche2023probabilistic,
title={Probabilistic Abduction for Visual Abstract Reasoning via Learning Rules in Vector-symbolic Architectures},
author={Michael Hersche and Francesco di Stefano and Thomas Hofmann and Abu Sebastian and Abbas Rahimi},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=rTz88hpGxc}
} | Abstract reasoning is a cornerstone of human intelligence, and replicating it with artificial intelligence (AI) presents an ongoing challenge. This study focuses on efficiently solving Raven's progressive matrices (RPM), a visual test for assessing abstract reasoning abilities, by using distributed computation and operators provided by vector-symbolic architectures (VSA). Instead of hard-coding the rule formulations associated with RPMs, our approach can learn the VSA rule formulations (hence the name Learn-VRF) with just one pass through the training data. Yet, our approach, with compact parameters, remains transparent and interpretable. Learn-VRF yields accurate predictions on I-RAVEN's in-distribution data, and exhibits strong out-of-distribution capabilities concerning unseen attribute-rule pairs, significantly outperforming pure connectionist baselines including large language models. Our code is available at https://github.com/IBM/learn-vector-symbolic-architectures-rule-formulations. | Probabilistic Abduction for Visual Abstract Reasoning via Learning Rules in Vector-symbolic Architectures | [
"Michael Hersche",
"Francesco di Stefano",
"Thomas Hofmann",
"Abu Sebastian",
"Abbas Rahimi"
] | Workshop/MATH-AI | 2401.16024 | [
"https://github.com/ibm/learn-vector-symbolic-architectures-rule-formulations"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=qRu578CgJD | @inproceedings{
bidi2023reinforcement,
title={Reinforcement Learning in Control Theory: A New Approach to Mathematical Problem Solving},
author={Kala Agbo Bidi and Jean-Michel Coron and Amaury Hayat and Nathan Lichtl{\'e}},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=qRu578CgJD}
} | One of the central questions in control theory is achieving stability through feedback control. This paper introduces a novel approach that combines Reinforcement Learning (RL) with mathematical analysis to address this challenge, with a specific focus on the Sterile Insect Technique (SIT) system. The objective is to find a feedback control that stabilizes the mosquito population model. Despite the mathematical complexities and the absence of known solutions for this specific problem, our RL approach identifies a candidate solution for an explicit stabilizing control. This study underscores the synergy between AI and mathematics, opening new avenues for tackling intricate mathematical problems. | Reinforcement Learning in Control Theory: A New Approach to Mathematical Problem Solving | [
"Kala Agbo Bidi",
"Jean-Michel Coron",
"Amaury Hayat",
"Nathan Lichtlé"
] | Workshop/MATH-AI | 2310.13072 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=oX1ANuWuIn | @inproceedings{
jenne2023can,
title={Can We Count on Deep Learning: Exploring and Characterizing Combinatorial Structures Using Machine Learning},
author={Helen Jenne and Herman Chau and Davis Brown and Jackson Warley and Timothy Doster and Henry Kvinge},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=oX1ANuWuIn}
} | With its exceptional pattern matching ability, deep learning has proven to be a powerful tool in a range of scientific domains. This is increasingly true in research mathematics, where recent work has demonstrated deep learning's ability to highlight subtle connections between mathematical objects that might escape a human expert. In this work we describe a simple method to help domain experts characterize a set of mathematical objects using deep learning. Such *characterization problems* often occur when some particular class of function, space, linear representation, etc. naturally emerges in calculations or other means but lacks a simple description. The goal is to find simple rules that also ideally shed light on the underlying mathematics. Our method, which we call *Feature Attribution Clustering for Exploration (FACE)*, clusters the feature attribution representations extracted from a trained model, arriving at a short list of prototype attributions that the domain expert can then try to convert into formal and rigorous rules. As a case study, we use our method to derive a new result in combinatorics by characterizing a subset of 0-1 matrices that corresponds to certain representations of permutations known as two-sided ordered words. | Can We Count on Deep Learning: Exploring and Characterizing Combinatorial Structures Using Machine Learning | [
"Helen Jenne",
"Herman Chau",
"Davis Brown",
"Jackson Warley",
"Timothy Doster",
"Henry Kvinge"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=m7m14acWQi | @inproceedings{
he-yueya2023solving,
title={Solving Math Word Problems by Combining Language Models With Symbolic Solvers},
author={Joy He-Yueya and Gabriel Poesia and Rose Wang and Noah Goodman},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=m7m14acWQi}
} | Automatically generating high-quality step-by-step solutions to math word problems has many applications in education. Recently, combining large language models (LLMs) with external tools to perform complex reasoning and calculation has emerged as a promising direction for solving math word problems, but prior approaches such as Program-Aided Language model (PAL) are biased towards simple procedural problems and less effective for problems that require declarative reasoning. We propose an approach that combines an LLM that can incrementally formalize word problems as a set of variables and equations with an external symbolic solver that can solve the equations. Our approach achieves comparable accuracy to the original PAL on the GSM8K benchmark of math word problems and outperforms PAL by an absolute 20% on ALGEBRA, a new dataset of more challenging word problems extracted from Algebra textbooks. Our work highlights the benefits of using declarative and incremental representations when interfacing with an external tool for solving complex math word problems. Our data and prompts are publicly available at https://github.com/joyheyueya/declarative-math-word-problem. | Solving Math Word Problems by Combining Language Models With Symbolic Solvers | [
"Joy He-Yueya",
"Gabriel Poesia",
"Rose Wang",
"Noah Goodman"
] | Workshop/MATH-AI | 2304.09102 | [
"https://github.com/joyheyueya/declarative-math-word-problem"
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=loDIYsNLbw | @inproceedings{
mao2023wamp,
title={{WAMP}: A Competition-Level Dataset for Assessing the Mathematical Reasoning Capabilities of {LLM}s},
author={Yujun Mao and Yoon Kim and Yilun Zhou},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=loDIYsNLbw}
} | Recent large language models (LLMs) have shown indications of math deduction abilities. However, it is unclear that for challenging math problems, what information about the problem helps (or hurts). In this paper, we propose a challenging benchmark dataset for such analyses. The Concept and Hint-Annotated Math Problems, or CHAMP, consists of competition-level math problems annotated with "concepts," or general math facts, and "hints," or problem-specific tricks. These entities and their interconnections allow us to explore the effects of additional information, such as relevant hints, misleading concepts, or related problems. We conduct 12 preliminary studies with 4 models, summarize our findings and discuss how CHAMP supports general discussions around LLMs' capabilities to understand and use contexts. The dataset, code and an extended version of the paper are available on the project website at https://yujunmao1.github.io/CHAMP. | CHAMP: A Competition-Level Dataset for Fine-Grained Analyses of LLMs' Mathematical Reasoning Capabilities | [
"Yujun Mao",
"Yoon Kim",
"Yilun Zhou"
] | Workshop/MATH-AI | 2401.06961 | [
""
] | https://huggingface.co/papers/2401.06961 | 3 | 0 | 0 | 3 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=j3mujdoAob | @inproceedings{
zhang2023ai,
title={{AI} for Mathematics: A Cognitive Science Perspective},
author={Cedegao Zhang and Katherine Collins and Adrian Weller and Joshua Tenenbaum},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=j3mujdoAob}
} | Mathematics is one of the most powerful conceptual systems developed and used by the human species. Dreams of automated mathematicians have a storied history in artificial intelligence (AI). Rapid progress in AI, particularly propelled by advances in large language models (LLMs), has sparked renewed, widespread interest in building such systems. In this work, we reflect on these goals from a \textit{cognitive science} perspective. We call attention to several classical and ongoing research directions from cognitive science, which we believe are valuable for AI practitioners to consider when seeking to build truly human (or superhuman)-level mathematical systems. We close with open discussions and questions that we believe necessitate a multi-disciplinary perspective---cognitive scientists working in tandem with AI researchers and mathematicians---as we move toward better mathematical AI systems which not only help us push the frontier of the mathematics, but also offer glimpses into how we as humans are even capable of such great cognitive feats. | AI for Mathematics: A Cognitive Science Perspective | [
"Cedegao Zhang",
"Katherine Collins",
"Adrian Weller",
"Joshua Tenenbaum"
] | Workshop/MATH-AI | 2310.13021 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=itphvM79Fx | @inproceedings{
thakur2023a,
title={A Language-Agent Approach to Formal Theorem-Proving},
author={Amitayush Thakur and Yeming Wen and Swarat Chaudhuri},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=itphvM79Fx}
} | Language agents, which use a large language model (LLM) capable of in-context learning to interact with an external environment, have emerged as a promising approach to control tasks. We present a language-agent approach that offers state-of-the-art performance in formal theorem-proving. Our method, COPRA, uses a high-capacity, black-box LLM (GPT-4) as part of a policy for a stateful backtracking search. During the search, the policy can select proof tactics and retrieve lemmas and definitions from an external database. Each selected tactic is executed in the underlying proof framework, and the execution feedback is used to build the prompt for the next policy invocation. The search also tracks selected information from its history and uses it to reduce hallucinations and unnecessary LLM queries.
We evaluate COPRA on the miniF2F benchmark for Lean and a set of Coq tasks from the Compcert project. On these benchmarks, COPRA is significantly better than one-shot invocations of GPT-4, as well as state-of-the-art models fine-tuned on proof data, at finding correct proofs quickly. | A Language-Agent Approach to Formal Theorem-Proving | [
"Amitayush Thakur",
"Yeming Wen",
"Swarat Chaudhuri"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=hrI14X0Ltk | @inproceedings{
lu2023mathvista,
title={MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts},
author={Pan Lu and Hritik Bansal and Tony Xia and Jiacheng Liu and Chunyuan Li and Hannaneh Hajishirzi and Hao Cheng and Kai-Wei Chang and Michel Galley and Jianfeng Gao},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=hrI14X0Ltk}
} | Although Large Language Models (LLMs) and Large Multimodal Models (LMMs) exhibit impressive skills in various domains, their ability for mathematical reasoning within visual contexts has not been formally examined. Equipping LLMs and LMMs with this capability is vital for general-purpose AI assistants and showcases promising potential in education, data analysis, and scientific discovery. To bridge this gap, we present MathVista, a benchmark designed to amalgamate challenges from diverse mathematical and visual tasks. We first taxonomize the key task types, reasoning skills, and visual contexts from the literature to guide our selection from 28 existing math-focused and visual question answering datasets. Then, we construct three new datasets, IQTest, FunctionQA, and PaperQA, to accommodate for missing types of visual contexts. The problems featured often require deep visual understanding beyond OCR or image captioning, and compositional reasoning with rich domain-specific tools, thus posing a notable challenge to existing models. We conduct a comprehensive evaluation of 11 prominent open-source and proprietary foundation models (LLMs, LLMs augmented with tools, and LMMs), and early experiments with GPT-4V. The best-performing model, Multimodal Bard, achieves only 58% of human performance (34.8% vs 60.3%), indicating ample room for further improvement. Given this significant gap, MathVista fuels future research in the development of general-purpose AI agents capable of tackling mathematically intensive and visually rich real-world tasks. Preliminary tests show that MathVista also presents challenges to GPT-4V, underscoring the benchmark's importance. | MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts | [
"Pan Lu",
"Hritik Bansal",
"Tony Xia",
"Jiacheng Liu",
"Chunyuan Li",
"Hannaneh Hajishirzi",
"Hao Cheng",
"Kai-Wei Chang",
"Michel Galley",
"Jianfeng Gao"
] | Workshop/MATH-AI | 2310.02255 | [
""
] | https://huggingface.co/papers/2310.02255 | 2 | 2 | 0 | 10 | [] | [
"AI4Math/MathVista"
] | [] | [] | [
"AI4Math/MathVista"
] | [] | 1 | poster |
null | https://openreview.net/forum?id=gUJLxVpWi2 | @inproceedings{
chen2023tooldec,
title={ToolDec: Syntax Error-Free and Generalizable Tool Use for {LLM}s via Finite-State Decoding},
author={Hongqiao Chen and Kexun Zhang and Lei Li and William Yang Wang},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=gUJLxVpWi2}
} | Large language models (LLMs) have shown promising capabilities in using external tools.
However, existing approaches rely on fine-tuning or in-context learning to use tools, which make syntactic mistakes and are difficult to generalize.
In this paper, we propose ToolDec, a finite-state machine-guided decoding algorithm for tool-augmented LLMs.
ToolDec eliminates tool-related errors by ensuring valid tool names and type-conforming arguments.
Furthermore, ToolDec enables LLM to effectively select tools using only the information contained in their names, with no need for tool-specific fine-tuning.
Our experiments on multiple word problem datasets show that ToolDec reduces syntactic errors to zero, consequently achieving significantly better performance and as much as a 2x speedup.
We also show that ToolDec achieves superior generalization performance on unseen tools, performing up to 8x better than the baseline. | ToolDec: Syntax Error-Free and Generalizable Tool Use for LLMs via Finite-State Decoding | [
"Hongqiao Chen",
"Kexun Zhang",
"Lei Li",
"William Yang Wang"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=btQ7Bt1NLF | @inproceedings{
charton2023teaching,
title={Teaching small transformers to rewrite {ZX} diagrams},
author={Francois Charton and Alexandre Krajenbrink and Konstantinos Meichanetzidis and Richie Yeung},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=btQ7Bt1NLF}
} | ZX calculus is a graphical language for reasoning about linear maps. Maps are represented as graphs, and reasoning amounts to graph rewrites. The main applications of ZX calculus are in quantum computation. We train small transformers to simplify ZX graphs, i.e. perform resource optimisation of quantum circuits.
Preliminary experiments show that transformers can be trained to simplify CNOT and Clifford circuits with high accuracy. These are the simplest kinds of ZX graphs, in the sense that there exists an efficient rewrite strategy. We also show evidence that transformers learn to simplify the more complex Clifford+T graphs, for which in general there does not exist an efficient simplification algorithm. | Teaching small transformers to rewrite ZX diagrams | [
"Francois Charton",
"Alexandre Krajenbrink",
"Konstantinos Meichanetzidis",
"Richie Yeung"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=YfhuG7xHQ8 | @inproceedings{
lee2023teaching,
title={Teaching Arithmetic to Small Transformers},
author={Nayoung Lee and Kartik Sreenivasan and Jason Lee and Kangwook Lee and Dimitris Papailiopoulos},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=YfhuG7xHQ8}
} | Large language models like GPT-4 exhibit emergent capabilities across general-purpose tasks, such as basic arithmetic, when trained on extensive text data, even though these tasks are not explicitly encoded by the unsupervised, next-token prediction objective. This study investigates how even small transformers, trained from random initialization, can efficiently learn arithmetic operations such as addition, multiplication, and elementary functions like square root, using the next-token prediction objective. We first demonstrate that conventional training data is not the most effective for arithmetic learning, and simple formatting changes can significantly improve accuracy. This leads to sharp phase transitions as a function of training data scale, which, in some cases, can be explained through connections to low-rank matrix completion. Building on prior work, we then train on chain-of-thought style data that includes intermediate step results. Even in the complete absence of pretraining, this approach significantly and simultaneously improves accuracy, sample complexity, and convergence speed. We also study the interplay between arithmetic and text data during training and examine the effects of few-shot prompting, pretraining, and parameter scaling. Additionally, we discuss the challenges associated with length generalization. Our work highlights the importance of high-quality, instructive data that considers the particular characteristics of the next-word prediction loss for rapidly eliciting arithmetic capabilities. | Teaching Arithmetic to Small Transformers | [
"Nayoung Lee",
"Kartik Sreenivasan",
"Jason Lee",
"Kangwook Lee",
"Dimitris Papailiopoulos"
] | Workshop/MATH-AI | 2307.03381 | [
"https://github.com/lee-ny/teaching_arithmetic"
] | https://huggingface.co/papers/2307.03381 | 3 | 17 | 0 | 5 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=WgaVCqZeIU | @inproceedings{
miku{\l}a2023magnushammer,
title={Magnushammer: A Transformer-Based Approach to Premise Selection},
author={Maciej Miku{\l}a and Szymon Antoniak and Szymon Tworkowski and Bartosz Piotrowski and Albert Jiang and Jin Peng Zhou and Christian Szegedy and {\L}ukasz Kuci{\'n}ski and Piotr Mi{\l}o{\'s} and Yuhuai Wu},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=WgaVCqZeIU}
} | We present Magnushammer: a novel approach to premise selection -- a crucial task in automated theorem proving. Traditionally, symbolic methods that rely on domain knowledge and engineering effort are applied to this task. In contrast, this work demonstrates that contrastive training with the transformer architecture can achieve higher-quality retrieval of relevant premises, without the domain knowledge or feature engineering overhead. Magnushammer outperforms the most advanced and widely used automation tool in interactive theorem proving: Sledgehammer. On the PISA and miniF2F benchmarks Magnushammer achieves $59.5\%$ (against $38.3\%$) and $34.0\%$ (against $20.9\%$) success rates, respectively. By combining Magnushammer with a language-model-based theorem prover, we further improve the state-of-the-art proof success rate from $57.0\%$ to $71.0\%$ on the PISA benchmark. Moreover, we develop and open source a novel, large dataset for premise selection. | Magnushammer: A Transformer-Based Approach to Premise Selection | [
"Maciej Mikuła",
"Szymon Antoniak",
"Szymon Tworkowski",
"Bartosz Piotrowski",
"Albert Jiang",
"Jin Peng Zhou",
"Christian Szegedy",
"Łukasz Kuciński",
"Piotr Miłoś",
"Yuhuai Wu"
] | Workshop/MATH-AI | 2303.04488 | [
""
] | https://huggingface.co/papers/2303.04488 | 0 | 0 | 0 | 9 | [] | [
"Simontwice/premise_selection_in_isabelle"
] | [] | [] | [
"Simontwice/premise_selection_in_isabelle"
] | [] | 1 | poster |
null | https://openreview.net/forum?id=WWDsbsgyhS | @inproceedings{
sharma2023sird,
title={{SIRD}: Symbolic Integration Rules Dataset},
author={Vaibhav Sharma and abhinav nagpal and Muhammed Fatih Balin},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=WWDsbsgyhS}
} | Advancements in neural networks and computer hardware lead to new use cases for deep learning in the natural sciences every day. Even though symbolic
mathematics tasks have been explored, symbolic integration only has a few studies using black box models and currently lacks explainability. Symbolic integration
is a challenging search problem and the final result is obtained by applying different integration rules at each step. We propose a novel and interpretable approach to perform symbolic integration using deep learning through integral rule prediction to speed up the search. We introduce the first-of-its-kind symbolic integration rules dataset comprising two million distinct functions and integration rule pairs. For complex rules
such as u-substitution and integration by parts, it also includes the expression needed for rule application.
We also train a transformer model on our proposed dataset and incorporate it into
SymPy's integral\_steps function to get guided\_integral\_steps, resulting in $6\times$ fewer branches explored by allowing our model to guide the depth-first-search procedure. | SIRD: Symbolic Integration Rules Dataset | [
"Vaibhav Sharma",
"abhinav nagpal",
"Muhammed Fatih Balin"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
||
null | https://openreview.net/forum?id=VlUAcTr881 | @inproceedings{
liang2023mint,
title={MinT: Boosting Generalization in Mathematical Reasoning via Multi-View Fine-Tuning},
author={Zhenwen Liang and Dian Yu and Xiaoman Pan and Wenlin Yao and Qingkai Zeng and Xiangliang Zhang and Dong Yu},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=VlUAcTr881}
} | Reasoning in mathematical domains remains a significant challenge for relatively small language models (LMs). Many current methods focus on specializing LMs in mathematical reasoning and rely heavily on knowledge distillation from powerful but inefficient large LMs (LLMs). In this work, we explore a new direction that avoids over-reliance on LLM teachers, introducing a multi-view fine-tuning method that efficiently exploits existing mathematical problem datasets with diverse annotation styles. Our approach uniquely considers the various annotation formats as different "views" and leverages them in training the model. By postpending distinct instructions to input questions, models can learn to generate solutions in diverse formats in a flexible manner. Experimental results show that our strategy enables a LLaMA-7B model to outperform prior approaches that utilize knowledge distillation, as well as carefully established baselines. Additionally, the proposed method grants the models promising generalization ability across various views and datasets, and the capability to learn from inaccurate or incomplete noisy data. We hope our multi-view training paradigm could inspire future studies in other machine reasoning domains. | MinT: Boosting Generalization in Mathematical Reasoning via Multi-View Fine-Tuning | [
"Zhenwen Liang",
"Dian Yu",
"Xiaoman Pan",
"Wenlin Yao",
"Qingkai Zeng",
"Xiangliang Zhang",
"Dong Yu"
] | Workshop/MATH-AI | 2307.07951 | [
""
] | https://huggingface.co/papers/2307.07951 | 2 | 0 | 0 | 7 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=VjKvYKCXuc | @inproceedings{
liu2023exploration,
title={Exploration with Principles for Diverse {AI} Supervision},
author={Hao Liu and Matei Zaharia and Pieter Abbeel},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=VjKvYKCXuc}
} | Training large transformers using next-token prediction has given rise to groundbreaking advancements in AI. While this generative AI approach has produced impressive results, it heavily leans on human supervision. Even state-of-the-art AI models like ChatGPT depend on fine-tuning through human demonstrations, demanding extensive human input and domain expertise. This strong reliance on human oversight poses a significant hurdle to the advancement of AI innovation. To address this limitation, we propose a novel paradigm termed Exploratory AI (EAI) aimed at autonomously generating high-quality training data. Drawing inspiration from the principles of unsupervised reinforcement learning (RL) pretraining, EAI achieves exploration within the natural language space. We accomplish this by harnessing large language models to assess the novelty of generated content. Our approach employs two key components: an actor that generates novel content and a critic that evaluates the generated content, offering critiques to guide the actor. Empirical evaluations demonstrate that EAI significantly boosts model performance on complex reasoning tasks, addressing the limitations of human-intensive supervision. | Exploration with Principles for Diverse AI Supervision | [
"Hao Liu",
"Matei Zaharia",
"Pieter Abbeel"
] | Workshop/MATH-AI | 2310.08899 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
|
null | https://openreview.net/forum?id=Rf9bHUBJUR | @inproceedings{
mekala2023echoprompt,
title={EchoPrompt: Instructing the Model to Rephrase Queries for Improved In-context Learning},
author={Raja Sekhar Reddy Mekala and Yasaman Razeghi and Sameer Singh},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=Rf9bHUBJUR}
} | Language models are achieving impressive performance on various tasks by aggressively adopting inference-time prompting techniques, such as zero-shot and few-shot prompting. In this work, we introduce EchoPrompt, a simple yet effective approach that prompts the model to rephrase its queries before answering them. EchoPrompt is adapted for both zero-shot and few-shot in-context learning with standard and chain-of-thought prompting. Experimental results show that EchoPrompt yields substantial improvements across all these settings for four families of causal language models. These improvements are observed across various numerical reasoning (e.g. GSM8K, SVAMP), reading comprehension (e.g. DROP), and logical reasoning (e.g. Coin Flipping) tasks. On average, EchoPrompt improves the Zero-shot-CoT performance of code-davinci-002 by 5% in numerical tasks and 13% in reading comprehension tasks. Our empirical results indicate that EchoPrompt is an effective technique that enhances in-context learning performance. We recommend incorporating EchoPrompt into various baseline prompting strategies to achieve performance boosts. | EchoPrompt: Instructing the Model to Rephrase Queries for Improved In-context Learning | [
"Raja Sekhar Reddy Mekala",
"Yasaman Razeghi",
"Sameer Singh"
] | Workshop/MATH-AI | 2309.10687 | [
"https://github.com/rajasekharmekala/query-rephrasing-subtask-cot"
] | https://huggingface.co/papers/2309.10687 | 2 | 1 | 0 | 3 | [] | [] | [] | [] | [] | [] | 1 | poster |
null | https://openreview.net/forum?id=ROOVUBZp8v | @inproceedings{
liu2023tinygsm,
title={Tiny{GSM}: achieving 80\% on {GSM}8k with one billion parameters},
author={Bingbin Liu and Sebastien Bubeck and Ronen Eldan and Janardhan Kulkarni and Yuanzhi Li and Anh Nguyen and Rachel Ward and Yi Zhang},
booktitle={The 3rd Workshop on Mathematical Reasoning and AI at NeurIPS'23},
year={2023},
url={https://openreview.net/forum?id=ROOVUBZp8v}
} | Small models offer various computational advantages, yet the extent to which size is critical for problem-solving abilities remains an open question. This work studies the performance of small models on mathematical reasoning. Specifically, for solving math word problems, we find that a 1.3B model can achieve 80.1% accuracy on GSM8K, outperforming existing models that are orders of magnitude larger, and even rivaling the performance of the GPT-3.5-turbo teacher model from which the training data is generated. Our approach is simple and has two key components: The first is the use of a GPT-3.5-turbo-generated synthetic dataset of math word problem with solutions, which we will fully release. The second component is the use of a verifier, which selects the final outputs from multiple candidate generations. | TinyGSM: achieving 80 | [
"Bingbin Liu",
"Sebastien Bubeck",
"Ronen Eldan",
"Janardhan Kulkarni",
"Yuanzhi Li",
"Anh Nguyen",
"Rachel Ward",
"Yi Zhang"
] | Workshop/MATH-AI | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | [] | [] | [] | 0 | poster |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.