Unnamed: 0
int64 0
5k
| title
stringlengths 9
210
| abstract
stringlengths 164
1.92k
|
|---|---|---|
3,400 | VLUE: A Multi-Task Benchmark for Evaluating Vision-Language Models | Recent advances in vision-language pre-training (VLP) have demonstrated
impressive performance in a range of vision-language (VL) tasks. However, there
exist several challenges for measuring the community's progress in building
general multi-modal intelligence. First, most of the downstream VL datasets are
annotated using raw images that are already seen during pre-training, which may
result in an overestimation of current VLP models' generalization ability.
Second, recent VLP work mainly focuses on absolute performance but overlooks
the efficiency-performance trade-off, which is also an important indicator for
measuring progress.
To this end, we introduce the Vision-Language Understanding Evaluation (VLUE)
benchmark, a multi-task multi-dimension benchmark for evaluating the
generalization capabilities and the efficiency-performance trade-off (``Pareto
SOTA'') of VLP models. We demonstrate that there is a sizable generalization
gap for all VLP models when testing on out-of-distribution test sets annotated
on images from a more diverse distribution that spreads across cultures.
Moreover, we find that measuring the efficiency-performance trade-off of VLP
models leads to complementary insights for several design choices of VLP. We
release the VLUE benchmark to promote research on building vision-language
models that generalize well to more diverse images and concepts unseen during
pre-training, and are practical in terms of efficiency-performance trade-off. |
3,401 | RankSim: Ranking Similarity Regularization for Deep Imbalanced Regression | Data imbalance, in which a plurality of the data samples come from a small
proportion of labels, poses a challenge in training deep neural networks.
Unlike classification, in regression the labels are continuous, potentially
boundless, and form a natural ordering. These distinct features of regression
call for new techniques that leverage the additional information encoded in
label-space relationships. This paper presents the RankSim (ranking similarity)
regularizer for deep imbalanced regression, which encodes an inductive bias
that samples that are closer in label space should also be closer in feature
space. In contrast to recent distribution smoothing based approaches, RankSim
captures both nearby and distant relationships: for a given data sample,
RankSim encourages the sorted list of its neighbors in label space to match the
sorted list of its neighbors in feature space. RankSim is complementary to
conventional imbalanced learning techniques, including re-weighting, two-stage
training, and distribution smoothing, and lifts the state-of-the-art
performance on three imbalanced regression benchmarks: IMDB-WIKI-DIR,
AgeDB-DIR, and STS-B-DIR. |
3,402 | Non-convex online learning via algorithmic equivalence | We study an algorithmic equivalence technique between nonconvex gradient
descent and convex mirror descent. We start by looking at a harder problem of
regret minimization in online non-convex optimization. We show that under
certain geometric and smoothness conditions, online gradient descent applied to
non-convex functions is an approximation of online mirror descent applied to
convex functions under reparameterization. In continuous time, the gradient
flow with this reparameterization was shown to be exactly equivalent to
continuous-time mirror descent by Amid and Warmuth 2020, but theory for the
analogous discrete time algorithms is left as an open problem. We prove an
$O(T^{\frac{2}{3}})$ regret bound for non-convex online gradient descent in
this setting, answering this open problem. Our analysis is based on a new and
simple algorithmic equivalence method. |
3,403 | Few-Shot Adaptation of Pre-Trained Networks for Domain Shift | Deep networks are prone to performance degradation when there is a domain
shift between the source (training) data and target (test) data. Recent
test-time adaptation methods update batch normalization layers of pre-trained
source models deployed in new target environments with streaming data to
mitigate such performance degradation. Although such methods can adapt
on-the-fly without first collecting a large target domain dataset, their
performance is dependent on streaming conditions such as mini-batch size and
class-distribution, which can be unpredictable in practice. In this work, we
propose a framework for few-shot domain adaptation to address the practical
challenges of data-efficient adaptation. Specifically, we propose a constrained
optimization of feature normalization statistics in pre-trained source models
supervised by a small support set from the target domain. Our method is easy to
implement and improves source model performance with as few as one sample per
class for classification tasks. Extensive experiments on 5 cross-domain
classification and 4 semantic segmentation datasets show that our method
achieves more accurate and reliable performance than test-time adaptation,
while not being constrained by streaming conditions. |
3,404 | Prompting ELECTRA: Few-Shot Learning with Discriminative Pre-Trained Models | Pre-trained masked language models successfully perform few-shot learning by
formulating downstream tasks as text infilling. However, as a strong
alternative in full-shot settings, discriminative pre-trained models like
ELECTRA do not fit into the paradigm. In this work, we adapt prompt-based
few-shot learning to ELECTRA and show that it outperforms masked language
models in a wide range of tasks. ELECTRA is pre-trained to distinguish if a
token is generated or original. We naturally extend that to prompt-based
few-shot learning by training to score the originality of the target options
without introducing new parameters. Our method can be easily adapted to tasks
involving multi-token predictions without extra computation overhead. Analysis
shows that ELECTRA learns distributions that align better with downstream
tasks. |
3,405 | Automatic Short Math Answer Grading via In-context Meta-learning | Automatic short answer grading is an important research direction in the
exploration of how to use artificial intelligence (AI)-based tools to improve
education. Current state-of-the-art approaches use neural language models to
create vectorized representations of students responses, followed by
classifiers to predict the score. However, these approaches have several key
limitations, including i) they use pre-trained language models that are not
well-adapted to educational subject domains and/or student-generated text and
ii) they almost always train one model per question, ignoring the linkage
across a question and result in a significant model storage problem due to the
size of advanced language models. In this paper, we study the problem of
automatic short answer grading for students' responses to math questions and
propose a novel framework for this task. First, we use MathBERT, a variant of
the popular language model BERT adapted to mathematical content, as our base
model and fine-tune it for the downstream task of student response grading.
Second, we use an in-context learning approach that provides scoring examples
as input to the language model to provide additional context information and
promote generalization to previously unseen questions. We evaluate our
framework on a real-world dataset of student responses to open-ended math
questions and show that our framework (often significantly) outperforms
existing approaches, especially for new questions that are not seen during
training. |
3,406 | A Graph and Attentive Multi-Path Convolutional Network for Traffic Prediction | Traffic prediction is an important and yet highly challenging problem due to
the complexity and constantly changing nature of traffic systems. To address
the challenges, we propose a graph and attentive multi-path convolutional
network (GAMCN) model to predict traffic conditions such as traffic speed
across a given road network into the future. Our model focuses on the spatial
and temporal factors that impact traffic conditions. To model the spatial
factors, we propose a variant of the graph convolutional network (GCN) named
LPGCN to embed road network graph vertices into a latent space, where vertices
with correlated traffic conditions are close to each other. To model the
temporal factors, we use a multi-path convolutional neural network (CNN) to
learn the joint impact of different combinations of past traffic conditions on
the future traffic conditions. Such a joint impact is further modulated by an
attention} generated from an embedding of the prediction time, which encodes
the periodic patterns of traffic conditions. We evaluate our model on
real-world road networks and traffic data. The experimental results show that
our model outperforms state-of-art traffic prediction models by up to 18.9% in
terms of prediction errors and 23.4% in terms of prediction efficiency. |
3,407 | Support Recovery in Sparse PCA with Incomplete Data | We study a practical algorithm for sparse principal component analysis (PCA)
of incomplete and noisy data. Our algorithm is based on the semidefinite
program (SDP) relaxation of the non-convex $l_1$-regularized PCA problem. We
provide theoretical and experimental evidence that SDP enables us to exactly
recover the true support of the sparse leading eigenvector of the unknown true
matrix, despite only observing an incomplete (missing uniformly at random) and
noisy version of it. We derive sufficient conditions for exact recovery, which
involve matrix incoherence, the spectral gap between the largest and
second-largest eigenvalues, the observation probability and the noise variance.
We validate our theoretical results with incomplete synthetic data, and show
encouraging and meaningful results on a gene expression dataset. |
3,408 | Gradient Backpropagation Through Combinatorial Algorithms: Identity with Projection Works | Embedding discrete solvers as differentiable layers has given modern deep
learning architectures combinatorial expressivity and discrete reasoning
capabilities. The derivative of these solvers is zero or undefined, therefore a
meaningful replacement is crucial for effective gradient-based learning. Prior
works rely on smoothing the solver with input perturbations, relaxing the
solver to continuous problems, or interpolating the loss landscape with
techniques that typically require additional solver calls, introduce extra
hyper-parameters or compromise performance. We propose a principled approach to
exploit the geometry of the discrete solution space to treat the solver as a
negative identity on the backward pass and further provide a theoretical
justification. Our experiments demonstrate that such a straightforward
hyper-parameter-free approach is on-par with or outperforms previous more
complex methods on numerous experiments such as Traveling Salesman Problem,
Shortest Path, Deep Graph Matching, and backpropagating through discrete
samplers. Furthermore, we substitute the previously proposed problem-specific
and label-dependent margin by a generic regularization procedure that prevents
cost collapse and increases robustness. |
3,409 | Flowification: Everything is a Normalizing Flow | We develop a method that can be used to turn any multi-layer perceptron or
convolutional network into a normalizing flow. In some cases this requires the
addition of uncorrelated noise to the model but in the simplest case no
additional parameters. The technique we develop can be applied to a broad range
of architectures, allowing them to be used for a wide range of tasks. Our
models also allow existing density estimation techniques to be combined with
high performance feature extractors. In contrast to standard density estimation
techniques that require specific architectures and specialized knowledge, our
approach can leverage design knowledge from other domains and is a step closer
to the realization of general purpose architectures. We investigate the
efficacy of linear and convolutional layers for the task of density estimation
on standard datasets. Our results suggest standard layers lack something
fundamental in comparison to other normalizing flows. |
3,410 | Superposing Many Tickets into One: A Performance Booster for Sparse Neural Network Training | Recent works on sparse neural network training (sparse training) have shown
that a compelling trade-off between performance and efficiency can be achieved
by training intrinsically sparse neural networks from scratch. Existing sparse
training methods usually strive to find the best sparse subnetwork possible in
one single run, without involving any expensive dense or pre-training steps.
For instance, dynamic sparse training (DST), as one of the most prominent
directions, is capable of reaching a competitive performance of dense training
by iteratively evolving the sparse topology during the course of training. In
this paper, we argue that it is better to allocate the limited resources to
create multiple low-loss sparse subnetworks and superpose them into a stronger
one, instead of allocating all resources entirely to find an individual
subnetwork. To achieve this, two desiderata are required: (1) efficiently
producing many low-loss subnetworks, the so-called cheap tickets, within one
training process limited to the standard training time used in dense training;
(2) effectively superposing these cheap tickets into one stronger subnetwork
without going over the constrained parameter budget. To corroborate our
conjecture, we present a novel sparse training approach, termed
\textbf{Sup-tickets}, which can satisfy the above two desiderata concurrently
in a single sparse-to-sparse training process. Across various modern
architectures on CIFAR-10/100 and ImageNet, we show that Sup-tickets integrates
seamlessly with the existing sparse training methods and demonstrates
consistent performance improvement. |
3,411 | DEP-RL: Embodied Exploration for Reinforcement Learning in Overactuated and Musculoskeletal Systems | Muscle-actuated organisms are capable of learning an unparalleled diversity
of dexterous movements despite their vast amount of muscles. Reinforcement
learning (RL) on large musculoskeletal models, however, has not been able to
show similar performance. We conjecture that ineffective exploration in large
overactuated action spaces is a key problem. This is supported by the finding
that common exploration noise strategies are inadequate in synthetic examples
of overactuated systems. We identify differential extrinsic plasticity (DEP), a
method from the domain of self-organization, as being able to induce
state-space covering exploration within seconds of interaction. By integrating
DEP into RL, we achieve fast learning of reaching and locomotion in
musculoskeletal systems, outperforming current approaches in all considered
tasks in sample efficiency and robustness. |
3,412 | PAC Generalization via Invariant Representations | One method for obtaining generalizable solutions to machine learning tasks
when presented with diverse training environments is to find invariant
representations of the data. These are representations of the covariates such
that the best model on top of the representation is invariant across training
environments. In the context of linear Structural Equation Models (SEMs),
invariant representations might allow us to learn models with
out-of-distribution guarantees, i.e., models that are robust to interventions
in the SEM. To address the invariant representation problem in a finite sample
setting, we consider the notion of $\epsilon$-approximate invariance. We study
the following question: If a representation is approximately invariant with
respect to a given number of training interventions, will it continue to be
approximately invariant on a larger collection of unseen SEMs? This larger
collection of SEMs is generated through a parameterized family of
interventions. Inspired by PAC learning, we obtain finite-sample
out-of-distribution generalization guarantees for approximate invariance that
holds probabilistically over a family of linear SEMs without faithfulness
assumptions. Our results show bounds that do not scale in ambient dimension
when intervention sites are restricted to lie in a constant size subset of
in-degree bounded nodes. We also show how to extend our results to a linear
indirect observation model that incorporates latent variables. |
3,413 | Do Deep Neural Networks Always Perform Better When Eating More Data? | Data has now become a shortcoming of deep learning. Researchers in their own
fields share the thinking that "deep neural networks might not always perform
better when they eat more data," which still lacks experimental validation and
a convincing guiding theory. Here to fill this lack, we design experiments from
Identically Independent Distribution(IID) and Out of Distribution(OOD), which
give powerful answers. For the purpose of guidance, based on the discussion of
results, two theories are proposed: under IID condition, the amount of
information determines the effectivity of each sample, the contribution of
samples and difference between classes determine the amount of sample
information and the amount of class information; under OOD condition, the
cross-domain degree of samples determine the contributions, and the
bias-fitting caused by irrelevant elements is a significant factor of
cross-domain. The above theories provide guidance from the perspective of data,
which can promote a wide range of practical applications of artificial
intelligence. |
3,414 | A Review and Evaluation of Elastic Distance Functions for Time Series Clustering | Time series clustering is the act of grouping time series data without
recourse to a label. Algorithms that cluster time series can be classified into
two groups: those that employ a time series specific distance measure; and
those that derive features from time series. Both approaches usually rely on
traditional clustering algorithms such as $k$-means. Our focus is on distance
based time series that employ elastic distance measures, i.e. distances that
perform some kind of realignment whilst measuring distance. We describe nine
commonly used elastic distance measures and compare their performance with
k-means and k-medoids clustering. Our findings are surprising. The most popular
technique, dynamic time warping (DTW), performs worse than Euclidean distance
with k-means, and even when tuned, is no better. Using k-medoids rather than
k-means improved the clusterings for all nine distance measures. DTW is not
significantly better than Euclidean distance with k-medoids. Generally,
distance measures that employ editing in conjunction with warping perform
better, and one distance measure, the move-split-merge (MSM) method, is the
best performing measure of this study. We also compare to clustering with DTW
using barycentre averaging (DBA). We find that DBA does improve DTW k-means,
but that the standard DBA is still worse than using MSM. Our conclusion is to
recommend MSM with k-medoids as the benchmark algorithm for clustering time
series with elastic distance measures. We provide implementations, results and
guidance on reproducing results on the associated GitHub repository. |
3,415 | Parameter Efficient Diff Pruning for Bias Mitigation | In recent years language models have achieved state of the art performance on
a wide variety of natural language processing tasks. As these models are
continuously growing in size it becomes increasingly important to explore
methods to make them more storage efficient. At the same time their increase
cognitive abilities increase the danger that societal bias existing in datasets
are implicitly encoded in the model weights. We propose an architecture which
deals with these two challenges at the same time using two techniques:
DiffPruning and Adverserial Training. The result is a modular architecture
which extends the original DiffPurning setup with and additional sparse
subnetwork applied as a mask to diminish the effects of a predefined protected
attribute at inference time. |
3,416 | Towards Efficient 3D Object Detection with Knowledge Distillation | Despite substantial progress in 3D object detection, advanced 3D detectors
often suffer from heavy computation overheads. To this end, we explore the
potential of knowledge distillation (KD) for developing efficient 3D object
detectors, focusing on popular pillar- and voxel-based detectors.Without
well-developed teacher-student pairs, we first study how to obtain student
models with good trade offs between accuracy and efficiency from the
perspectives of model compression and input resolution reduction. Then, we
build a benchmark to assess existing KD methods developed in the 2D domain for
3D object detection upon six well-constructed teacher-student pairs. Further,
we propose an improved KD pipeline incorporating an enhanced logit KD method
that performs KD on only a few pivotal positions determined by teacher
classification response, and a teacher-guided student model initialization to
facilitate transferring teacher model's feature extraction ability to students
through weight inheritance. Finally, we conduct extensive experiments on the
Waymo dataset. Our best performing model achieves $65.75\%$ LEVEL 2 mAPH,
surpassing its teacher model and requiring only $44\%$ of teacher flops. Our
most efficient model runs 51 FPS on an NVIDIA A100, which is $2.2\times$ faster
than PointPillar with even higher accuracy. Code will be available. |
3,417 | Principle Components Analysis based frameworks for efficient missing data imputation algorithms | Missing data is a commonly occurring problem in practice, and imputation,
i.e., filling the missing entries of the data, is a popular way to deal with
this problem. This motivates multiple works on imputation to deal with missing
data of various types and dimensions. However, for high-dimensional datasets,
these imputation methods can be computationally expensive. Therefore, in this
work, we propose Principle Component Analysis Imputation (PCAI), a simple
framework based on Principle Component Analysis (PCA) to speed up the
imputation process of many available imputation techniques. Next, based on
PCAI, we propose PCA Imputation - Classification (PIC), an imputation-dimension
reduction-classification framework to deal with missing data classification
problems where it is desirable to reduce the dimensions before training a
classification model. Our experiments show that the proposed frameworks can be
utilized with various imputation algorithms and improve the imputation speed
significantly. Interestingly, the frameworks aid imputation methods that rely
on many parameters by reducing the dimension of the data and hence, reducing
the number of parameters needed to be estimated. Moreover, they not only can
achieve compatible mean square error/higher classification accuracy compared to
the traditional imputation style on the original missing dataset but many times
deliver even better results. In addition, the frameworks also help to tackle
the memory issue that many imputation approaches have by reducing the number of
features. |
3,418 | Batch Normalization Is Blind to the First and Second Derivatives of the Loss | In this paper, we prove the effects of the BN operation on the
back-propagation of the first and second derivatives of the loss. When we do
the Taylor series expansion of the loss function, we prove that the BN
operation will block the influence of the first-order term and most influence
of the second-order term of the loss. We also find that such a problem is
caused by the standardization phase of the BN operation. Experimental results
have verified our theoretical conclusions, and we have found that the BN
operation significantly affects feature representations in specific tasks,
where losses of different samples share similar analytic formulas. |
3,419 | Machine Learning Methods for Health-Index Prediction in Coating Chambers | Coating chambers create thin layers that improve the mechanical and optical
surface properties in jewelry production using physical vapor deposition. In
such a process, evaporated material condensates on the walls of such chambers
and, over time, causes mechanical defects and unstable processes. As a result,
manufacturers perform extensive maintenance procedures to reduce production
loss. Current rule-based maintenance strategies neglect the impact of specific
recipes and the actual condition of the vacuum chamber. Our overall goal is to
predict the future condition of the coating chamber to allow cost and quality
optimized maintenance of the equipment. This paper describes the derivation of
a novel health indicator that serves as a step toward condition-based
maintenance for coating chambers. We indirectly use gas emissions of the
chamber's contamination to evaluate the machine's condition. Our approach
relies on process data and does not require additional hardware installation.
Further, we evaluated multiple machine learning algorithms for a
condition-based forecast of the health indicator that also reflects production
planning. Our results show that models based on decision trees are the most
effective and outperform all three benchmarks, improving at least $0.22$ in the
mean average error. Our work paves the way for cost and quality optimized
maintenance of coating applications. |
3,420 | On Avoiding Local Minima Using Gradient Descent With Large Learning Rates | It has been widely observed in training of neural networks that when applying
gradient descent (GD), a large step size is essential for obtaining superior
models. However, the effect of large step sizes on the success of GD is not
well understood theoretically. We argue that a complete understanding of the
mechanics leading to GD's success may indeed require considering effects of
using a large step size. To support this claim, we prove on a certain class of
functions that GD with large step size follows a different trajectory than GD
with a small step size, leading to convergence to the global minimum. We also
demonstrate the difference in trajectories for small and large learning rates
when GD is applied on a neural network, observing effects of an escape from a
local minimum with a large step size, which shows this behavior is indeed
relevant in practice. Finally, through a novel set of experiments, we show even
though stochastic noise is beneficial, it is not enough to explain success of
SGD and a large learning rate is essential for obtaining the best performance
even in stochastic settings. |
3,421 | Optimal Gradient Sliding and its Application to Distributed Optimization Under Similarity | We study structured convex optimization problems, with additive objective
$r:=p + q$, where $r$ is ($\mu$-strongly) convex, $q$ is $L_q$-smooth and
convex, and $p$ is $L_p$-smooth, possibly nonconvex. For such a class of
problems, we proposed an inexact accelerated gradient sliding method that can
skip the gradient computation for one of these components while still achieving
optimal complexity of gradient calls of $p$ and $q$, that is,
$\mathcal{O}(\sqrt{L_p/\mu})$ and $\mathcal{O}(\sqrt{L_q/\mu})$,
respectively. This result is much sharper than the classic black-box complexity
$\mathcal{O}(\sqrt{(L_p+L_q)/\mu})$, especially when the difference between
$L_q$ and $L_q$ is large. We then apply the proposed method to solve
distributed optimization problems over master-worker architectures, under
agents' function similarity, due to statistical data similarity or otherwise.
The distributed algorithm achieves for the first time lower complexity bounds
on {\it both} communication and local gradient calls, with the former having
being a long-standing open problem. Finally the method is extended to
distributed saddle-problems (under function similarity) by means of solving a
class of variational inequalities, achieving lower communication and
computation complexity bounds. |
3,422 | Group Probability-Weighted Tree Sums for Interpretable Modeling of Heterogeneous Data | Machine learning in high-stakes domains, such as healthcare, faces two
critical challenges: (1) generalizing to diverse data distributions given
limited training data while (2) maintaining interpretability. To address these
challenges, we propose an instance-weighted tree-sum method that effectively
pools data across diverse groups to output a concise, rule-based model. Given
distinct groups of instances in a dataset (e.g., medical patients grouped by
age or treatment site), our method first estimates group membership
probabilities for each instance. Then, it uses these estimates as instance
weights in FIGS (Tan et al. 2022), to grow a set of decision trees whose values
sum to the final prediction. We call this new method Group Probability-Weighted
Tree Sums (G-FIGS). G-FIGS achieves state-of-the-art prediction performance on
important clinical datasets; e.g., holding the level of sensitivity fixed at
92%, G-FIGS increases specificity for identifying cervical spine injury by up
to 10% over CART and up to 3% over FIGS alone, with larger gains at higher
sensitivity levels. By keeping the total number of rules below 16 in FIGS, the
final models remain interpretable, and we find that their rules match medical
domain expertise. All code, data, and models are released on Github. |
3,423 | Why Adversarial Training of ReLU Networks Is Difficult? | This paper mathematically derives an analytic solution of the adversarial
perturbation on a ReLU network, and theoretically explains the difficulty of
adversarial training. Specifically, we formulate the dynamics of the
adversarial perturbation generated by the multi-step attack, which shows that
the adversarial perturbation tends to strengthen eigenvectors corresponding to
a few top-ranked eigenvalues of the Hessian matrix of the loss w.r.t. the
input. We also prove that adversarial training tends to strengthen the
influence of unconfident input samples with large gradient norms in an
exponential manner. Besides, we find that adversarial training strengthens the
influence of the Hessian matrix of the loss w.r.t. network parameters, which
makes the adversarial training more likely to oscillate along directions of a
few samples, and boosts the difficulty of adversarial training. Crucially, our
proofs provide a unified explanation for previous findings in understanding
adversarial training. |
3,424 | Domain Constraints in Feature Space: Strengthening Robustness of Android Malware Detection against Realizable Adversarial Examples | Strengthening the robustness of machine learning-based malware detectors
against realistic evasion attacks remains one of the major obstacles for
Android malware detection. To that end, existing work has focused on
interpreting domain constraints of Android malware in the problem space, where
problem-space realizable adversarial examples are generated. In this paper, we
provide another promising way to achieve the same goal but based on
interpreting the domain constraints in the feature space, where feature-space
realizable adversarial examples are generated. Specifically, we present a novel
approach to extracting feature-space domain constraints by learning meaningful
feature dependencies from data, and applying them based on a novel robust
feature space. Experimental results successfully demonstrate the effectiveness
of our novel robust feature space in providing adversarial robustness for
DREBIN, a state-of-the-art Android malware detector. For example, it can
decrease the evasion rate of a realistic gradient-based attack by $96.4\%$ in a
limited-knowledge (transfer) setting and by $13.8\%$ in a more challenging,
perfect-knowledge setting. In addition, we show that directly using our learned
domain constraints in the adversarial retraining framework leads to about
$84\%$ improvement in a limited-knowledge setting, with up to $377\times$
faster implementation than using problem-space adversarial examples. |
3,425 | Universal Deep GNNs: Rethinking Residual Connection in GNNs from a Path Decomposition Perspective for Preventing the Over-smoothing | The performance of GNNs degrades as they become deeper due to the
over-smoothing. Among all the attempts to prevent over-smoothing, residual
connection is one of the promising methods due to its simplicity. However,
recent studies have shown that GNNs with residual connections only slightly
slow down the degeneration. The reason why residual connections fail in GNNs is
still unknown. In this paper, we investigate the forward and backward behavior
of GNNs with residual connections from a novel path decomposition perspective.
We find that the recursive aggregation of the median length paths from the
binomial distribution of residual connection paths dominates output
representation, resulting in over-smoothing as GNNs go deeper. Entangled
propagation and weight matrices cause gradient smoothing and prevent GNNs with
residual connections from optimizing to the identity mapping. Based on these
findings, we present a Universal Deep GNNs (UDGNN) framework with cold-start
adaptive residual connections (DRIVE) and feedforward modules. Extensive
experiments demonstrate the effectiveness of our method, which achieves
state-of-the-art results over non-smooth heterophily datasets by simply
stacking standard GNNs. |
3,426 | Generalizing Hierarchical Bayesian Bandits | A contextual bandit is a popular and practical framework for online learning
to act under uncertainty. In many problems, the number of actions is huge and
their mean rewards are correlated. In this work, we introduce a general
framework for capturing such correlations through a two-level graphical model
where actions are related through multiple shared latent parameters. We propose
a Thompson sampling algorithm G-HierTS that uses this structure to explore
efficiently and bound its Bayes regret. The regret has two terms, one for
learning action parameters and the other for learning the shared latent
parameters. The terms reflect the structure of our model as well as the quality
of priors. Our theoretical findings are validated empirically using both
synthetic and real-world problems. We also experiment with G-HierTS that
maintains a factored posterior over latent parameters. While this approximation
does not come with guarantees, it improves computational efficiency with a
minimal impact on empirical regret. |
3,427 | Payday loans -- blessing or growth suppressor? Machine Learning Analysis | The upsurge of real estate involves a variety of factors that have got
influenced by many domains. Indeed, the unrecognized sector that would affect
the economy for which regulatory proposals are being drafted to keep this in
control is the payday loans. This research paper revolves around the impact of
payday loans in the real estate market. The research paper draws a first-hand
experience of obtaining the index for the concentration of real estate in an
area of reference by virtue of payday loans in Toronto, Ontario in particular,
which sets out an ideology to create, evaluate and demonstrate the scenario
through research analysis. The purpose of this indexing via payday loans is the
basic - debt: income ratio which states that when the income of the person
bound to pay the interest of payday loans increases, his debt goes down
marginally which hence infers that the person invests in fixed assets like real
estate which hikes up its growth. |
3,428 | OOD Link Prediction Generalization Capabilities of Message-Passing GNNs in Larger Test Graphs | This work provides the first theoretical study on the ability of graph
Message Passing Neural Networks (gMPNNs) -- such as Graph Neural Networks
(GNNs) -- to perform inductive out-of-distribution (OOD) link prediction tasks,
where deployment (test) graph sizes are larger than training graphs. We first
prove non-asymptotic bounds showing that link predictors based on
permutation-equivariant (structural) node embeddings obtained by gMPNNs can
converge to a random guess as test graphs get larger. We then propose a
theoretically-sound gMPNN that outputs structural pairwise (2-node) embeddings
and prove non-asymptotic bounds showing that, as test graphs grow, these
embeddings converge to embeddings of a continuous function that retains its
ability to predict links OOD. Empirical results on random graphs show agreement
with our theoretical results. |
3,429 | Learning Adaptive Propagation for Knowledge Graph Reasoning | Due to the success of Graph Neural Networks (GNNs) in learning from
graph-structured data, various GNN-based methods have been introduced to learn
from knowledge graphs (KGs). In this paper, to reveal the key factors
underneath existing GNN-based methods, we revisit exemplar works from the lens
of the propagation path. We find that the answer entity can be close to queried
one, but the information dependency can be long. Thus, better reasoning
performance can be obtained by exploring longer propagation paths. However,
identifying such a long-range dependency in KG is hard since the number of
involved entities grows exponentially. This motivates us to learn an adaptive
propagation path that filters out irrelevant entities while preserving
promising targets during the propagation. First, we design an incremental
sampling mechanism where the close and promising target can be preserved.
Second, we design a learning-based sampling distribution to identify the
targets with fewer involved entities. In this way, GNN can go deeper to capture
long-range information. Extensive experiments show that our method is efficient
and achieves state-of-the-art performances in both transductive and inductive
reasoning settings, benefiting from the deeper propagation. |
3,430 | Online Agnostic Multiclass Boosting | Boosting is a fundamental approach in machine learning that enjoys both
strong theoretical and practical guarantees. At a high-level, boosting
algorithms cleverly aggregate weak learners to generate predictions with
arbitrarily high accuracy. In this way, boosting algorithms convert weak
learners into strong ones. Recently, Brukhim et al. extended boosting to the
online agnostic binary classification setting. A key ingredient in their
approach is a clean and simple reduction to online convex optimization, one
that efficiently converts an arbitrary online convex optimizer to an agnostic
online booster. In this work, we extend this reduction to multiclass problems
and give the first boosting algorithm for online agnostic mutliclass
classification. Our reduction also enables the construction of algorithms for
statistical agnostic, online realizable, and statistical realizable multiclass
boosting. |
3,431 | A k nearest neighbours classifiers ensemble based on extended neighbourhood rule and features subsets | kNN based ensemble methods minimise the effect of outliers by identifying a
set of data points in the given feature space that are nearest to an unseen
observation in order to predict its response by using majority voting. The
ordinary ensembles based on kNN find out the k nearest observations in a region
(bounded by a sphere) based on a predefined value of k. This scenario, however,
might not work in situations when the test observation follows the pattern of
the closest data points with the same class that lie on a certain path not
contained in the given sphere. This paper proposes a k nearest neighbour
ensemble where the neighbours are determined in k steps. Starting from the
first nearest observation of the test point, the algorithm identifies a single
observation that is closest to the observation at the previous step. At each
base learner in the ensemble, this search is extended to k steps on a random
bootstrap sample with a random subset of features selected from the feature
space. The final predicted class of the test point is determined by using a
majority vote in the predicted classes given by all base models. This new
ensemble method is applied on 17 benchmark datasets and compared with other
classical methods, including kNN based models, in terms of classification
accuracy, kappa and Brier score as performance metrics. Boxplots are also
utilised to illustrate the difference in the results given by the proposed and
other state-of-the-art methods. The proposed method outperformed the rest of
the classical methods in the majority of cases. The paper gives a detailed
simulation study for further assessment. |
3,432 | FLICU: A Federated Learning Workflow for Intensive Care Unit Mortality Prediction | Although Machine Learning (ML) can be seen as a promising tool to improve
clinical decision-making for supporting the improvement of medication plans,
clinical procedures, diagnoses, or medication prescriptions, it remains limited
by access to healthcare data. Healthcare data is sensitive, requiring strict
privacy practices, and typically stored in data silos, making traditional
machine learning challenging. Federated learning can counteract those
limitations by training machine learning models over data silos while keeping
the sensitive data localized. This study proposes a federated learning workflow
for ICU mortality prediction. Hereby, the applicability of federated learning
as an alternative to centralized machine learning and local machine learning is
investigated by introducing federated learning to the binary classification
problem of predicting ICU mortality. We extract multivariate time series data
from the MIMIC-III database (lab values and vital signs), and benchmark the
predictive performance of four deep sequential classifiers (FRNN, LSTM, GRU,
and 1DCNN) varying the patient history window lengths (8h, 16h, 24h, 48h) and
the number of FL clients (2, 4, 8). The experiments demonstrate that both
centralized machine learning and federated learning are comparable in terms of
AUPRC and F1-score. Furthermore, the federated approach shows superior
performance over local machine learning. Thus, the federated approach can be
seen as a valid and privacy-preserving alternative to centralized machine
learning for classifying ICU mortality when sharing sensitive patient data
between hospitals is not possible. |
3,433 | Meta Representation Learning with Contextual Linear Bandits | Meta-learning seeks to build algorithms that rapidly learn how to solve new
learning problems based on previous experience. In this paper we investigate
meta-learning in the setting of stochastic linear bandit tasks. We assume that
the tasks share a low dimensional representation, which has been partially
acquired from previous learning tasks. We aim to leverage this information in
order to learn a new downstream bandit task, which shares the same
representation. Our principal contribution is to show that if the learned
representation estimates well the unknown one, then the downstream task can be
efficiently learned by a greedy policy that we propose in this work. We derive
an upper bound on the regret of this policy, which is, up to logarithmic
factors, of order $r\sqrt{N}(1\vee \sqrt{d/T})$, where $N$ is the horizon of
the downstream task, $T$ is the number of training tasks, $d$ the ambient
dimension and $r \ll d$ the dimension of the representation. We highlight that
our strategy does not need to know $r$. We note that if $T> d$ our bound
achieves the same rate of optimal minimax bandit algorithms using the true
underlying representation. Our analysis is inspired and builds in part upon
previous work on meta-learning in the i.i.d. full information setting
\citep{tripuraneni2021provable,boursier2022trace}. As a separate contribution
we show how to relax certain assumptions in those works, thereby improving
their representation learning and risk analysis. |
3,434 | CHALLENGER: Training with Attribution Maps | We show that utilizing attribution maps for training neural networks can
improve regularization of models and thus increase performance. Regularization
is key in deep learning, especially when training complex models on relatively
small datasets. In order to understand inner workings of neural networks,
attribution methods such as Layer-wise Relevance Propagation (LRP) have been
extensively studied, particularly for interpreting the relevance of input
features. We introduce Challenger, a module that leverages the explainable
power of attribution maps in order to manipulate particularly relevant input
patterns. Therefore, exposing and subsequently resolving regions of ambiguity
towards separating classes on the ground-truth data manifold, an issue that
arises particularly when training models on rather small datasets. Our
Challenger module increases model performance through building more diverse
filters within the network and can be applied to any input data domain. We
demonstrate that our approach results in substantially better classification as
well as calibration performance on datasets with only a few samples up to
datasets with thousands of samples. In particular, we show that our generic
domain-independent approach yields state-of-the-art results in vision, natural
language processing and on time series tasks. |
3,435 | Retrieving and Ranking Relevant JavaScript Technologies from Web Repositories | The selection of software technologies is an important but complex task. We
consider developers of JavaScript (JS) applications, for whom the assessment of
JS libraries has become difficult and time-consuming due to the growing number
of technology options available. A common strategy is to browse software
repositories via search engines (e.g., NPM, or Google), although it brings some
problems. First, given a technology need, the engines might return a long list
of results, which often causes information overload issues. Second, the results
should be ranked according to criteria of interest for the developer. However,
deciding how to weight these criteria to make a decision is not
straightforward. In this work, we propose a two-phase approach for assisting
developers to retrieve and rank JS technologies in a semi-automated fashion.
The first-phase (ST-Retrieval) uses a meta-search technique for collecting JS
technologies that meet the developer's needs. The second-phase (called
ST-Rank), relies on a machine learning technique to infer, based on criteria
used by other projects in the Web, a ranking of the output of ST-Retrieval. We
evaluated our approach with NPM and obtained satisfactory results in terms of
the accuracy of the technologies retrieved and the order in which they were
ranked. |
3,436 | CGMN: A Contrastive Graph Matching Network for Self-Supervised Graph Similarity Learning | Graph similarity learning refers to calculating the similarity score between
two graphs, which is required in many realistic applications, such as visual
tracking, graph classification, and collaborative filtering. As most of the
existing graph neural networks yield effective graph representations of a
single graph, little effort has been made for jointly learning two graph
representations and calculating their similarity score. In addition, existing
unsupervised graph similarity learning methods are mainly clustering-based,
which ignores the valuable information embodied in graph pairs. To this end, we
propose a contrastive graph matching network (CGMN) for self-supervised graph
similarity learning in order to calculate the similarity between any two input
graph objects. Specifically, we generate two augmented views for each graph in
a pair respectively. Then, we employ two strategies, namely cross-view
interaction and cross-graph interaction, for effective node representation
learning. The former is resorted to strengthen the consistency of node
representations in two views. The latter is utilized to identify node
differences between different graphs. Finally, we transform node
representations into graph-level representations via pooling operations for
graph similarity computation. We have evaluated CGMN on eight real-world
datasets, and the experiment results show that the proposed new approach is
superior to the state-of-the-art methods in graph similarity learning
downstream tasks. |
3,437 | Improved Algorithms for Bandit with Graph Feedback via Regret Decomposition | The problem of bandit with graph feedback generalizes both the multi-armed
bandit (MAB) problem and the learning with expert advice problem by encoding in
a directed graph how the loss vector can be observed in each round of the game.
The mini-max regret is closely related to the structure of the feedback graph
and their connection is far from being fully understood. We propose a new
algorithmic framework for the problem based on a partition of the feedback
graph. Our analysis reveals the interplay between various parts of the graph by
decomposing the regret to the sum of the regret caused by small parts and the
regret caused by their interaction. As a result, our algorithm can be viewed as
an interpolation and generalization of the optimal algorithms for MAB and
learning with expert advice. Our framework unifies previous algorithms for both
strongly observable graphs and weakly observable graphs, resulting in improved
and optimal regret bounds on a wide range of graph families including graphs of
bounded degree and strongly observable graphs with a few corrupted arms. |
3,438 | Align then Fusion: Generalized Large-scale Multi-view Clustering with Anchor Matching Correspondences | Multi-view anchor graph clustering selects representative anchors to avoid
full pair-wise similarities and therefore reduce the complexity of graph
methods. Although widely applied in large-scale applications, existing
approaches do not pay sufficient attention to establishing correct
correspondences between the anchor sets across views. To be specific, anchor
graphs obtained from different views are not aligned column-wisely. Such an
Anchor-Unaligned Problem (AUP) would cause inaccurate graph fusion and degrade
the clustering performance. Under multi-view scenarios, generating correct
correspondences could be extremely difficult since anchors are not consistent
in feature dimensions. To solve this challenging issue, we propose the first
study of a generalized and flexible anchor graph fusion framework termed Fast
Multi-View Anchor-Correspondence Clustering (FMVACC). Specifically, we show how
to find anchor correspondence with both feature and structure information,
after which anchor graph fusion is performed column-wisely. Moreover, we
theoretically show the connection between FMVACC and existing multi-view late
fusion and partial view-aligned clustering, which further demonstrates our
generality. Extensive experiments on seven benchmark datasets demonstrate the
effectiveness and efficiency of our proposed method. Moreover, the proposed
alignment module also shows significant performance improvement applying to
existing multi-view anchor graph competitors indicating the importance of
anchor alignment. |
3,439 | Embedding Graphs on Grassmann Manifold | Learning efficient graph representation is the key to favorably addressing
downstream tasks on graphs, such as node or graph property prediction. Given
the non-Euclidean structural property of graphs, preserving the original graph
data's similarity relationship in the embedded space needs specific tools and a
similarity metric. This paper develops a new graph representation learning
scheme, namely EGG, which embeds approximated second-order graph
characteristics into a Grassmann manifold. The proposed strategy leverages
graph convolutions to learn hidden representations of the corresponding
subspace of the graph, which is then mapped to a Grassmann point of a low
dimensional manifold through truncated singular value decomposition (SVD). The
established graph embedding approximates denoised correlationship of node
attributes, as implemented in the form of a symmetric matrix space for
Euclidean calculation. The effectiveness of EGG is demonstrated using both
clustering and classification tasks at the node level and graph level. It
outperforms baseline models on various benchmarks. |
3,440 | SEREN: Knowing When to Explore and When to Exploit | Efficient reinforcement learning (RL) involves a trade-off between
"exploitative" actions that maximise expected reward and "explorative'" ones
that sample unvisited states. To encourage exploration, recent approaches
proposed adding stochasticity to actions, separating exploration and
exploitation phases, or equating reduction in uncertainty with reward. However,
these techniques do not necessarily offer entirely systematic approaches making
this trade-off. Here we introduce SElective Reinforcement Exploration Network
(SEREN) that poses the exploration-exploitation trade-off as a game between an
RL agent -- \exploiter, which purely exploits known rewards, and another RL
agent -- \switcher, which chooses at which states to activate a pure
exploration policy that is trained to minimise system uncertainty and override
Exploiter. Using a form of policies known as impulse control, \switcher is able
to determine the best set of states to switch to the exploration policy while
Exploiter is free to execute its actions everywhere else. We prove that SEREN
converges quickly and induces a natural schedule towards pure exploitation.
Through extensive empirical studies in both discrete (MiniGrid) and continuous
(MuJoCo) control benchmarks, we show that SEREN can be readily combined with
existing RL algorithms to yield significant improvement in performance relative
to state-of-the-art algorithms. |
3,441 | A Transistor Operations Model for Deep Learning Energy Consumption Scaling | Deep Learning (DL) has transformed the automation of a wide range of
industries and finds increasing ubiquity in society. The increasing complexity
of DL models and its widespread adoption has led to the energy consumption
doubling every 3-4 months. Currently, the relationship between DL model
configuration and energy consumption is not well established. Current FLOPs and
MACs based methods only consider the linear operations. In this paper, we
develop a bottom-level Transistor Operations (TOs) method to expose the role of
activation functions and neural network structure in energy consumption scaling
with DL model configuration. TOs allows us uncovers the role played by
non-linear operations (e.g. division/root operations performed by activation
functions and batch normalisation). As such, our proposed TOs model provides
developers with a hardware-agnostic index for how energy consumption scales
with model settings. To validate our work, we analyse the TOs energy scaling of
a feed-forward DNN model set and achieve a 98.2% - 99.97% precision in
estimating its energy consumption. We believe this work can be extended to any
DL model. |
3,442 | Hilbert Curve Projection Distance for Distribution Comparison | Distribution comparison plays a central role in many machine learning tasks
like data classification and generative modeling. In this study, we propose a
novel metric, called Hilbert curve projection (HCP) distance, to measure the
distance between two probability distributions with high robustness and low
complexity. In particular, we first project two high-dimensional probability
densities using Hilbert curve to obtain a coupling between them, and then
calculate the transport distance between these two densities in the original
space, according to the coupling. We show that HCP distance is a proper metric
and is well-defined for absolutely continuous probability measures.
Furthermore, we demonstrate that the empirical HCP distance converges to its
population counterpart at a rate of no more than $O(n^{-1/2d})$ under
regularity conditions. To suppress the curse-of-dimensionality, we also develop
two variants of the HCP distance using (learnable) subspace projections.
Experiments on both synthetic and real-world data show that our HCP distance
works as an effective surrogate of the Wasserstein distance with low complexity
and overcomes the drawbacks of the sliced Wasserstein distance. |
3,443 | Stock Trading Optimization through Model-based Reinforcement Learning with Resistance Support Relative Strength | Reinforcement learning (RL) is gaining attention by more and more researchers
in quantitative finance as the agent-environment interaction framework is
aligned with decision making process in many business problems. Most of the
current financial applications using RL algorithms are based on model-free
method, which still faces stability and adaptivity challenges. As lots of
cutting-edge model-based reinforcement learning (MBRL) algorithms mature in
applications such as video games or robotics, we design a new approach that
leverages resistance and support (RS) level as regularization terms for action
in MBRL, to improve the algorithm's efficiency and stability. From the
experiment results, we can see RS level, as a market timing technique, enhances
the performance of pure MBRL models in terms of various measurements and
obtains better profit gain with less riskiness. Besides, our proposed method
even resists big drop (less maximum drawdown) during COVID-19 pandemic period
when the financial market got unpredictable crisis. Explanations on why control
of resistance and support level can boost MBRL is also investigated through
numerical experiments, such as loss of actor-critic network and prediction
error of the transition dynamical model. It shows that RS indicators indeed
help the MBRL algorithms to converge faster at early stage and obtain smaller
critic loss as training episodes increase. |
3,444 | Metrizing Fairness | We study supervised learning problems for predicting properties of
individuals who belong to one of two demographic groups, and we seek predictors
that are fair according to statistical parity. This means that the
distributions of the predictions within the two groups should be close with
respect to the Kolmogorov distance, and fairness is achieved by penalizing the
dissimilarity of these two distributions in the objective function of the
learning problem. In this paper, we showcase conceptual and computational
benefits of measuring unfairness with integral probability metrics (IPMs) other
than the Kolmogorov distance. Conceptually, we show that the generator of any
IPM can be interpreted as a family of utility functions and that unfairness
with respect to this IPM arises if individuals in the two demographic groups
have diverging expected utilities. We also prove that the
unfairness-regularized prediction loss admits unbiased gradient estimators if
unfairness is measured by the squared $\mathcal L^2$-distance or by a squared
maximum mean discrepancy. In this case, the fair learning problem is
susceptible to efficient stochastic gradient descent (SGD) algorithms.
Numerical experiments on real data show that these SGD algorithms outperform
state-of-the-art methods for fair learning in that they achieve superior
accuracy-unfairness trade-offs -- sometimes orders of magnitude faster.
Finally, we identify conditions under which statistical parity can improve
prediction accuracy. |
3,445 | RLx2: Training a Sparse Deep Reinforcement Learning Model from Scratch | Training deep reinforcement learning (DRL) models usually requires high
computation costs. Therefore, compressing DRL models possesses immense
potential for training acceleration and model deployment. However, existing
methods that generate small models mainly adopt the knowledge distillation
based approach by iteratively training a dense network, such that the training
process still demands massive computing resources. Indeed, sparse training from
scratch in DRL has not been well explored and is particularly challenging due
to non-stationarity in bootstrap training. In this work, we propose a novel
sparse DRL training framework, "the \textbf{R}igged \textbf{R}einforcement
\textbf{L}earning \textbf{L}ottery" (RLx2), which is capable of training a DRL
agent \emph{using an ultra-sparse network throughout} for off-policy
reinforcement learning. The systematic RLx2 framework contains three key
components: gradient-based topology evolution, multi-step Temporal Difference
(TD) targets, and dynamic-capacity replay buffer. RLx2 enables efficient
topology exploration and robust Q-value estimation simultaneously. We
demonstrate state-of-the-art sparse training performance in several continuous
control tasks using RLx2, showing $7.5\times$-$20\times$ model compression with
less than $3\%$ performance degradation, and up to $20\times$ and $50\times$
FLOPs reduction for training and inference, respectively. |
3,446 | Agnostic Physics-Driven Deep Learning | This work establishes that a physical system can perform statistical learning
without gradient computations, via an Agnostic Equilibrium Propagation
(Aeqprop) procedure that combines energy minimization, homeostatic control, and
nudging towards the correct response. In Aeqprop, the specifics of the system
do not have to be known: the procedure is based only on external manipulations,
and produces a stochastic gradient descent without explicit gradient
computations. Thanks to nudging, the system performs a true, order-one gradient
step for each training sample, in contrast with order-zero methods like
reinforcement or evolutionary strategies, which rely on trial and error. This
procedure considerably widens the range of potential hardware for statistical
learning to any system with enough controllable parameters, even if the details
of the system are poorly known. Aeqprop also establishes that in natural
(bio)physical systems, genuine gradient-based statistical learning may result
from generic, relatively simple mechanisms, without backpropagation and its
requirement for analytic knowledge of partial derivatives. |
3,447 | Task-Prior Conditional Variational Auto-Encoder for Few-Shot Image Classification | Transductive methods always outperform inductive methods in few-shot image
classification scenarios. However, the existing few-shot methods contain a
latent condition: the number of samples in each class is the same, which may be
unrealistic. To cope with those cases where the query shots of each class are
nonuniform (i.e. nonuniform few-shot learning), we propose a Task-Prior
Conditional Variational Auto-Encoder model named TP-VAE, conditioned on support
shots and constrained by a task-level prior regularization. Our method obtains
high performance in the more challenging nonuniform few-shot scenarios.
Moreover, our method outperforms the state-of-the-art in a wide range of
standard few-shot image classification scenarios. Among them, the accuracy of
1-shot increased by about 3\%. |
3,448 | Efficient Transformed Gaussian Processes for Non-Stationary Dependent Multi-class Classification | This work introduces the Efficient Transformed Gaussian Process (ETGP), a new
way of creating C stochastic processes characterized by: 1) the C processes are
non-stationary, 2) the C processes are dependent by construction without
needing a mixing matrix, 3) training and making predictions is very efficient
since the number of Gaussian Processes (GP) operations (e.g. inverting the
inducing point's covariance matrix) do not depend on the number of processes.
This makes the ETGP particularly suited for multi-class problems with a very
large number of classes, which are the problems studied in this work. ETGPs
exploit the recently proposed Transformed Gaussian Process (TGP), a stochastic
process specified by transforming a Gaussian Process using an invertible
transformation. However, unlike TGPs, ETGPs are constructed by transforming a
single sample from a GP using C invertible transformations. We derive an
efficient sparse variational inference algorithm for the proposed model and
demonstrate its utility in 5 classification tasks which include
low/medium/large datasets and a different number of classes, ranging from just
a few to hundreds. Our results show that ETGPs, in general, outperform
state-of-the-art methods for multi-class classification based on GPs, and have
a lower computational cost (around one order of magnitude smaller). |
3,449 | Chefs' Random Tables: Non-Trigonometric Random Features | We introduce chefs' random tables (CRTs), a new class of non-trigonometric
random features (RFs) to approximate Gaussian and softmax kernels. CRTs are an
alternative to standard random kitchen sink (RKS) methods, which inherently
rely on the trigonometric maps. We present variants of CRTs where RFs are
positive, a key requirement for applications in recent low-rank Transformers.
Further variance reduction is possible by leveraging statistics which are
simple to compute. One instantiation of CRTs, the optimal positive random
features (OPRFs), is to our knowledge the first RF method for unbiased softmax
kernel estimation with positive and bounded RFs, resulting in exponentially
small tails and much lower variance than its counterparts. As we show,
orthogonal random features applied in OPRFs provide additional variance
reduction for any dimensionality $d$ (not only asymptotically for sufficiently
large $d$, as for RKS). We test CRTs on many tasks ranging from non-parametric
classification to training Transformers for text, speech and image data,
obtaining new state-of-the-art results for low-rank text Transformers, while
providing linear space and time complexity. |
3,450 | Running the Dual-PQC GAN on noisy simulators and real quantum hardware | In an earlier work, we introduced dual-Parameterized Quantum Circuit (PQC)
Generative Adversarial Networks (GAN), an advanced prototype of a quantum GAN.
We applied the model on a realistic High-Energy Physics (HEP) use case: the
exact theoretical simulation of a calorimeter response with a reduced problem
size. This paper explores the dual- PQC GAN for a more practical usage by
testing its performance in the presence of different types of quantum noise,
which are the major obstacles to overcome for successful deployment using
near-term quantum devices. The results propose the possibility of running the
model on current real hardware, but improvements are still required in some
areas. |
3,451 | Quantum Multi-Armed Bandits and Stochastic Linear Bandits Enjoy Logarithmic Regrets | Multi-arm bandit (MAB) and stochastic linear bandit (SLB) are important
models in reinforcement learning, and it is well-known that classical
algorithms for bandits with time horizon $T$ suffer $\Omega(\sqrt{T})$ regret.
In this paper, we study MAB and SLB with quantum reward oracles and propose
quantum algorithms for both models with $O(\mbox{poly}(\log T))$ regrets,
exponentially improving the dependence in terms of $T$. To the best of our
knowledge, this is the first provable quantum speedup for regrets of bandit
problems and in general exploitation in reinforcement learning. Compared to
previous literature on quantum exploration algorithms for MAB and reinforcement
learning, our quantum input model is simpler and only assumes quantum oracles
for each individual arm. |
3,452 | A Continuous Time Framework for Discrete Denoising Models | We provide the first complete continuous time framework for denoising
diffusion models of discrete data. This is achieved by formulating the forward
noising process and corresponding reverse time generative process as Continuous
Time Markov Chains (CTMCs). The model can be efficiently trained using a
continuous time version of the ELBO. We simulate the high dimensional CTMC
using techniques developed in chemical physics and exploit our continuous time
framework to derive high performance samplers that we show can outperform
discrete time methods for discrete data. The continuous time treatment also
enables us to derive a novel theoretical result bounding the error between the
generated sample distribution and the true data distribution. |
3,453 | Fast Nonlinear Vector Quantile Regression | Quantile regression (QR) is a powerful tool for estimating one or more
conditional quantiles of a target variable $\mathrm{Y}$ given explanatory
features $\boldsymbol{\mathrm{X}}$. A limitation of QR is that it is only
defined for scalar target variables, due to the formulation of its objective
function, and since the notion of quantiles has no standard definition for
multivariate distributions. Recently, vector quantile regression (VQR) was
proposed as an extension of QR for high-dimensional target variables, thanks to
a meaningful generalization of the notion of quantiles to multivariate
distributions. Despite its elegance, VQR is arguably not applicable in practice
due to several limitations: (i) it assumes a linear model for the quantiles of
the target $\mathrm{Y}$ given the features $\boldsymbol{\mathrm{X}}$; (ii) its
exact formulation is intractable even for modestly-sized problems in terms of
target dimensions, number of regressed quantile levels, or number of features,
and its relaxed dual formulation may violate the monotonicity of the estimated
quantiles; (iii) no fast or scalable solvers for VQR currently exist. In this
work we fully address these limitations, namely: (i) We extend VQR to the
non-linear case, showing substantial improvement over linear VQR; (ii) We
propose vector monotone rearrangement, a method which ensures the estimates
obtained by VQR relaxations are monotone functions; (iii) We provide fast,
GPU-accelerated solvers for linear and nonlinear VQR which maintain a fixed
memory footprint with number of samples and quantile levels, and demonstrate
that they scale to millions of samples and thousands of quantile levels; (iv)
We release an optimized python package of our solvers as to widespread the use
of VQR in real-world applications. |
3,454 | Rethinking Saliency Map: An Context-aware Perturbation Method to Explain EEG-based Deep Learning Model | Deep learning is widely used to decode the electroencephalogram (EEG) signal.
However, there are few attempts to specifically investigate how to explain the
EEG-based deep learning models. We conduct a review to summarize the existing
works explaining the EEG-based deep learning model. Unfortunately, we find that
there is no appropriate method to explain them. Based on the characteristic of
EEG data, we suggest a context-aware perturbation method to generate a saliency
map from the perspective of the raw EEG signal. Moreover, we also justify that
the context information can be used to suppress the artifacts in the EEG-based
deep learning model. In practice, some users might want a simple version of the
explanation, which only indicates a few features as salient points. To this
end, we propose an optional area limitation strategy to restrict the
highlighted region. To validate our idea and make a comparison with the other
methods, we select three representative EEG-based models to implement
experiments on the emotional EEG dataset DEAP. The results of the experiments
support the advantages of our method. |
3,455 | Knowledge Distillation for 6D Pose Estimation by Keypoint Distribution Alignment | Knowledge distillation facilitates the training of a compact student network
by using a deep teacher one. While this has achieved great success in many
tasks, it remains completely unstudied for image-based 6D object pose
estimation. In this work, we introduce the first knowledge distillation method
for 6D pose estimation. Specifically, we follow a standard approach to 6D pose
estimation, consisting of predicting the 2D image locations of object
keypoints. In this context, we observe the compact student network to struggle
predicting precise 2D keypoint locations. Therefore, to address this, instead
of training the student with keypoint-to-keypoint supervision, we introduce a
strategy based the optimal transport theory that distills the teacher's
keypoint \emph{distribution} into the student network, facilitating its
training. Our experiments on several benchmarks show that our distillation
method yields state-of-the-art results with different compact student models. |
3,456 | Sampling-free Inference for Ab-Initio Potential Energy Surface Networks | Obtaining the energy of molecular systems typically requires solving the
associated Schr\"odinger equation. Unfortunately, analytical solutions only
exist for single-electron systems, and accurate approximate solutions are
expensive. In recent work, the potential energy surface network (PESNet) has
been proposed to reduce training time by solving the Schr\"odinger equation for
many geometries simultaneously. While training significantly faster, inference
still required numerical integration limiting the evaluation to a few
geometries. Here, we address the inference shortcomings by proposing the
Potential learning from ab-initio Networks (PlaNet) framework to simultaneously
train a surrogate model that avoids expensive Monte-Carlo integration and,
thus, reduces inference time from minutes or even hours to milliseconds. In
this way, we can accurately model high-resolution multi-dimensional energy
surfaces that previously would have been unobtainable via neural wave
functions. Finally, we present PESNet++, an architectural improvement to
PESNet, that reduces errors by up to 39% and provides new state-of-the-art
results for neural wave functions across all systems evaluated. |
3,457 | FedAUXfdp: Differentially Private One-Shot Federated Distillation | Federated learning suffers in the case of non-iid local datasets, i.e., when
the distributions of the clients' data are heterogeneous. One promising
approach to this challenge is the recently proposed method FedAUX, an
augmentation of federated distillation with robust results on even highly
heterogeneous client data. FedAUX is a partially $(\epsilon,
\delta)$-differentially private method, insofar as the clients' private data is
protected in only part of the training it takes part in. This work contributes
a fully differentially private modification, termed FedAUXfdp. We further
contribute an upper bound on the $l_2$-sensitivity of regularized multinomial
logistic regression. In experiments with deep networks on large-scale image
datasets, FedAUXfdp with strong differential privacy guarantees performs
significantly better than other equally privatized SOTA baselines on non-iid
client data in just a single communication round. Full privatization of the
modified method results in a negligible reduction in accuracy at all levels of
data heterogeneity. |
3,458 | White-box Membership Attack Against Machine Learning Based Retinopathy Classification | The advances in machine learning (ML) have greatly improved AI-based
diagnosis aid systems in medical imaging. However, being based on collecting
medical data specific to individuals induces several security issues,
especially in terms of privacy. Even though the owner of the images like a
hospital put in place strict privacy protection provisions at the level of its
information system, the model trained over his images still holds disclosure
potential. The trained model may be accessible to an attacker as: 1) White-box:
accessing to the model architecture and parameters; 2) Black box: where he can
only query the model with his own inputs through an appropriate interface.
Existing attack methods include: feature estimation attacks (FEA), membership
inference attack (MIA), model memorization attack (MMA) and identification
attacks (IA). In this work we focus on MIA against a model that has been
trained to detect diabetic retinopathy from retinal images. Diabetic
retinopathy is a condition that can cause vision loss and blindness in the
people who have diabetes. MIA is the process of determining whether a data
sample comes from the training data set of a trained ML model or not. From a
privacy perspective in our use case where a diabetic retinopathy classification
model is given to partners that have at their disposal images along with
patients' identifiers, inferring the membership status of a data sample can
help to state if a patient has contributed or not to the training of the model. |
3,459 | Dataset Condensation via Efficient Synthetic-Data Parameterization | The great success of machine learning with massive amounts of data comes at a
price of huge computation costs and storage for training and tuning. Recent
studies on dataset condensation attempt to reduce the dependence on such
massive data by synthesizing a compact training dataset. However, the existing
approaches have fundamental limitations in optimization due to the limited
representability of synthetic datasets without considering any data regularity
characteristics. To this end, we propose a novel condensation framework that
generates multiple synthetic data with a limited storage budget via efficient
parameterization considering data regularity. We further analyze the
shortcomings of the existing gradient matching-based condensation methods and
develop an effective optimization technique for improving the condensation of
training data information. We propose a unified algorithm that drastically
improves the quality of condensed data against the current state-of-the-art on
CIFAR-10, ImageNet, and Speech Commands. |
3,460 | AttentionCode: Ultra-Reliable Feedback Codes for Short-Packet Communications | Ultra-reliable short-packet communication is a major challenge in future
wireless networks with critical applications. To achieve ultra-reliable
communications beyond 99.999%, this paper envisions a new interaction-based
communication paradigm that exploits the feedback from the receiver for the
sixth generation (6G) communication networks and beyond. We present
AttentionCode, a new class of feedback codes leveraging deep learning (DL)
technologies. The underpinnings of AttentionCode are three architectural
innovations: AttentionNet, input restructuring, and adaptation to fading
channels, accompanied by several training methods, including large-batch
training, distributed learning, look-ahead optimizer, training-test
signal-to-noise ratio (SNR) mismatch, and curriculum learning. The training
methods can potentially be generalized to other wireless communication
applications with machine learning. Numerical experiments verify that
AttentionCode establishes a new state of the art among all DL-based feedback
codes in both additive white Gaussian noise (AWGN) channels and fading
channels. In AWGN channels with noiseless feedback, for example, AttentionCode
achieves a block error rate (BLER) of $10^{-7}$ when the forward channel SNR is
0dB for a block size of 50 bits, demonstrating the potential of AttentionCode
to provide ultra-reliable short-packet communications for 6G. |
3,461 | Multi-Agent Reinforcement Learning is a Sequence Modeling Problem | Large sequence model (SM) such as GPT series and BERT has displayed
outstanding performance and generalization capabilities on vision, language,
and recently reinforcement learning tasks. A natural follow-up question is how
to abstract multi-agent decision making into an SM problem and benefit from the
prosperous development of SMs. In this paper, we introduce a novel architecture
named Multi-Agent Transformer (MAT) that effectively casts cooperative
multi-agent reinforcement learning (MARL) into SM problems wherein the task is
to map agents' observation sequence to agents' optimal action sequence. Our
goal is to build the bridge between MARL and SMs so that the modeling power of
modern sequence models can be unleashed for MARL. Central to our MAT is an
encoder-decoder architecture which leverages the multi-agent advantage
decomposition theorem to transform the joint policy search problem into a
sequential decision making process; this renders only linear time complexity
for multi-agent problems and, most importantly, endows MAT with monotonic
performance improvement guarantee. Unlike prior arts such as Decision
Transformer fit only pre-collected offline data, MAT is trained by online
trials and errors from the environment in an on-policy fashion. To validate
MAT, we conduct extensive experiments on StarCraftII, Multi-Agent MuJoCo,
Dexterous Hands Manipulation, and Google Research Football benchmarks. Results
demonstrate that MAT achieves superior performance and data efficiency compared
to strong baselines including MAPPO and HAPPO. Furthermore, we demonstrate that
MAT is an excellent few-short learner on unseen tasks regardless of changes in
the number of agents. See our project page at
https://sites.google.com/view/multi-agent-transformer. |
3,462 | Data-driven Numerical Invariant Synthesis with Automatic Generation of Attributes | We propose a data-driven algorithm for numerical invariant synthesis and
verification. The algorithm is based on the ICE-DT schema for learning decision
trees from samples of positive and negative states and implications
corresponding to program transitions. The main issue we address is the
discovery of relevant attributes to be used in the learning process of
numerical invariants. We define a method for solving this problem guided by the
data sample. It is based on the construction of a separator that covers
positive states and excludes negative ones, consistent with the implications.
The separator is constructed using an abstract domain representation of convex
sets. The generalization mechanism of the decision tree learning from the
constraints of the separator allows the inference of general invariants,
accurate enough for proving the targeted property. We implemented our algorithm
and showed its efficiency. |
3,463 | Edge YOLO: Real-Time Intelligent Object Detection System Based on Edge-Cloud Cooperation in Autonomous Vehicles | Driven by the ever-increasing requirements of autonomous vehicles, such as
traffic monitoring and driving assistant, deep learning-based object detection
(DL-OD) has been increasingly attractive in intelligent transportation systems.
However, it is difficult for the existing DL-OD schemes to realize the
responsible, cost-saving, and energy-efficient autonomous vehicle systems due
to low their inherent defects of low timeliness and high energy consumption. In
this paper, we propose an object detection (OD) system based on edge-cloud
cooperation and reconstructive convolutional neural networks, which is called
Edge YOLO. This system can effectively avoid the excessive dependence on
computing power and uneven distribution of cloud computing resources.
Specifically, it is a lightweight OD framework realized by combining pruning
feature extraction network and compression feature fusion network to enhance
the efficiency of multi-scale prediction to the largest extent. In addition, we
developed an autonomous driving platform equipped with NVIDIA Jetson for
system-level verification. We experimentally demonstrate the reliability and
efficiency of Edge YOLO on COCO2017 and KITTI data sets, respectively.
According to COCO2017 standard datasets with a speed of 26.6 frames per second
(FPS), the results show that the number of parameters in the entire network is
only 25.67 MB, while the accuracy (mAP) is up to 47.3%. |
3,464 | Detecting Unknown DGAs without Context Information | New malware emerges at a rapid pace and often incorporates Domain Generation
Algorithms (DGAs) to avoid blocking the malware's connection to the command and
control (C2) server. Current state-of-the-art classifiers are able to separate
benign from malicious domains (binary classification) and attribute them with
high probability to the DGAs that generated them (multiclass classification).
While binary classifiers can label domains of yet unknown DGAs as malicious,
multiclass classifiers can only assign domains to DGAs that are known at the
time of training, limiting the ability to uncover new malware families. In this
work, we perform a comprehensive study on the detection of new DGAs, which
includes an evaluation of 59,690 classifiers. We examine four different
approaches in 15 different configurations and propose a simple yet effective
approach based on the combination of a softmax classifier and regular
expressions (regexes) to detect multiple unknown DGAs with high probability. At
the same time, our approach retains state-of-the-art classification performance
for known DGAs. Our evaluation is based on a leave-one-group-out
cross-validation with a total of 94 DGA families. By using the maximum number
of known DGAs, our evaluation scenario is particularly difficult and close to
the real world. All of the approaches examined are privacy-preserving, since
they operate without context and exclusively on a single domain to be
classified. We round up our study with a thorough discussion of
class-incremental learning strategies that can adapt an existing classifier to
newly discovered classes. |
3,465 | Harnessing spectral representations for subgraph alignment | With the rise and advent of graph learning techniques, graph data has become
ubiquitous. However, while several efforts are being devoted to the design of
new convolutional architectures, pooling or positional encoding schemes, less
effort is being spent on problems involving maps between (possibly very large)
graphs, such as signal transfer, graph isomorphism and subgraph correspondence.
With this paper, we anticipate the need for a convenient framework to deal with
such problems, and focus in particular on the challenging subgraph alignment
scenario. We claim that, first and foremost, the representation of a map plays
a central role on how these problems should be modeled. Taking the hint from
recent work in geometry processing, we propose the adoption of a spectral
representation for maps that is compact, easy to compute, robust to topological
changes, easy to plug into existing pipelines, and is especially effective for
subgraph alignment problems. We report for the first time a surprising
phenomenon where the partiality arising in the subgraph alignment task is
manifested as a special structure of the map coefficients, even in the absence
of exact subgraph isomorphism, and which is consistently observed over
different families of graphs up to several thousand nodes. |
3,466 | Deep Learning Methods for Fingerprint-Based Indoor Positioning: A Review | Outdoor positioning systems based on the Global Navigation Satellite System
have several shortcomings that have deemed their use for indoor positioning
impractical. Location fingerprinting, which utilizes machine learning, has
emerged as a viable method and solution for indoor positioning due to its
simple concept and accurate performance. In the past, shallow learning
algorithms were traditionally used in location fingerprinting. Recently, the
research community started utilizing deep learning methods for fingerprinting
after witnessing the great success and superiority these methods have over
traditional/shallow machine learning algorithms. This paper provides a
comprehensive review of deep learning methods in indoor positioning. First, the
advantages and disadvantages of various fingerprint types for indoor
positioning are discussed. The solutions proposed in the literature are then
analyzed, categorized, and compared against various performance evaluation
metrics. Since data is key in fingerprinting, a detailed review of publicly
available indoor positioning datasets is presented. While incorporating deep
learning into fingerprinting has resulted in significant improvements, doing
so, has also introduced new challenges. These challenges along with the common
implementation pitfalls are discussed. Finally, the paper is concluded with
some remarks as well as future research trends. |
3,467 | Neural Volumetric Object Selection | We introduce an approach for selecting objects in neural volumetric 3D
representations, such as multi-plane images (MPI) and neural radiance fields
(NeRF). Our approach takes a set of foreground and background 2D user scribbles
in one view and automatically estimates a 3D segmentation of the desired
object, which can be rendered into novel views. To achieve this result, we
propose a novel voxel feature embedding that incorporates the neural volumetric
3D representation and multi-view image features from all input views. To
evaluate our approach, we introduce a new dataset of human-provided
segmentation masks for depicted objects in real-world multi-view scene
captures. We show that our approach out-performs strong baselines, including 2D
segmentation and 3D segmentation approaches adapted to our task. |
3,468 | CalFAT: Calibrated Federated Adversarial Training with Label Skewness | Recent studies have shown that, like traditional machine learning, federated
learning (FL) is also vulnerable to adversarial attacks. To improve the
adversarial robustness of FL, few federated adversarial training (FAT) methods
have been proposed to apply adversarial training locally before global
aggregation. Although these methods demonstrate promising results on
independent identically distributed (IID) data, they suffer from training
instability issues on non-IID data with label skewness, resulting in much
degraded natural accuracy. This tends to hinder the application of FAT in
real-world applications where the label distribution across the clients is
often skewed. In this paper, we study the problem of FAT under label skewness,
and firstly reveal one root cause of the training instability and natural
accuracy degradation issues: skewed labels lead to non-identical class
probabilities and heterogeneous local models. We then propose a Calibrated FAT
(CalFAT) approach to tackle the instability issue by calibrating the logits
adaptively to balance the classes. We show both theoretically and empirically
that the optimization of CalFAT leads to homogeneous local models across the
clients and much improved convergence rate and final performance. |
3,469 | Unbalanced CO-Optimal Transport | Optimal transport (OT) compares probability distributions by computing a
meaningful alignment between their samples. CO-optimal transport (COOT) takes
this comparison further by inferring an alignment between features as well.
While this approach leads to better alignments and generalizes both OT and
Gromov-Wasserstein distances, we provide a theoretical result showing that it
is sensitive to outliers that are omnipresent in real-world data. This prompts
us to propose unbalanced COOT for which we provably show its robustness to
noise in the compared datasets. To the best of our knowledge, this is the first
such result for OT methods in incomparable spaces. With this result in hand, we
provide empirical evidence of this robustness for the challenging tasks of
heterogeneous domain adaptation with and without varying proportions of classes
and simultaneous alignment of samples and features across single-cell
measurements. |
3,470 | ACIL: Analytic Class-Incremental Learning with Absolute Memorization and Privacy Protection | Class-incremental learning (CIL) learns a classification model with training
data of different classes arising progressively. Existing CIL either suffers
from serious accuracy loss due to catastrophic forgetting, or invades data
privacy by revisiting used exemplars. Inspired by linear learning formulations,
we propose an analytic class-incremental learning (ACIL) with absolute
memorization of past knowledge while avoiding breaching of data privacy (i.e.,
without storing historical data). The absolute memorization is demonstrated in
the sense that class-incremental learning using ACIL given present data would
give identical results to that from its joint-learning counterpart which
consumes both present and historical samples. This equality is theoretically
validated. Data privacy is ensured since no historical data are involved during
the learning process. Empirical validations demonstrate ACIL's competitive
accuracy performance with near-identical results for various incremental task
settings (e.g., 5-50 phases). This also allows ACIL to outperform the
state-of-the-art methods for large-phase scenarios (e.g., 25 and 50 phases). |
3,471 | A Deep Learning Approach for Automatic Detection of Qualitative Features of Lecturing | Artificial Intelligence in higher education opens new possibilities for
improving the lecturing process, such as enriching didactic materials, helping
in assessing students' works or even providing directions to the teachers on
how to enhance the lectures. We follow this research path, and in this work, we
explore how an academic lecture can be assessed automatically by quantitative
features. First, we prepare a set of qualitative features based on teaching
practices and then annotate the dataset of academic lecture videos collected
for this purpose. We then show how these features could be detected
automatically using machine learning and computer vision techniques. Our
results show the potential usefulness of our work. |
3,472 | Uncertainty Quantification and Resource-Demanding Computer Vision Applications of Deep Learning | Bringing deep neural networks (DNNs) into safety critical applications such
as automated driving, medical imaging and finance, requires a thorough
treatment of the model's uncertainties. Training deep neural networks is
already resource demanding and so is also their uncertainty quantification. In
this overview article, we survey methods that we developed to teach DNNs to be
uncertain when they encounter new object classes. Additionally, we present
training methods to learn from only a few labels with help of uncertainty
quantification. Note that this is typically paid with a massive overhead in
computation of an order of magnitude and more compared to ordinary network
training. Finally, we survey our work on neural architecture search which is
also an order of magnitude more resource demanding then ordinary network
training. |
3,473 | Confederated Learning: Federated Learning with Decentralized Edge Servers | Federated learning (FL) is an emerging machine learning paradigm that allows
to accomplish model training without aggregating data at a central server. Most
studies on FL consider a centralized framework, in which a single server is
endowed with a central authority to coordinate a number of devices to perform
model training in an iterative manner. Due to stringent communication and
bandwidth constraints, such a centralized framework has limited scalability as
the number of devices grows. To address this issue, in this paper, we propose a
ConFederated Learning (CFL) framework. The proposed CFL consists of multiple
servers, in which each server is connected with an individual set of devices as
in the conventional FL framework, and decentralized collaboration is leveraged
among servers to make full use of the data dispersed throughout the network. We
develop an alternating direction method of multipliers (ADMM) algorithm for
CFL. The proposed algorithm employs a random scheduling policy which randomly
selects a subset of devices to access their respective servers at each
iteration, thus alleviating the need of uploading a huge amount of information
from devices to servers. Theoretical analysis is presented to justify the
proposed method. Numerical results show that the proposed method can converge
to a decent solution significantly faster than gradient-based FL algorithms,
thus boasting a substantial advantage in terms of communication efficiency. |
3,474 | FRAug: Tackling Federated Learning with Non-IID Features via Representation Augmentation | Federated Learning (FL) is a decentralized learning paradigm in which
multiple clients collaboratively train deep learning models without
centralizing their local data and hence preserve data privacy. Real-world
applications usually involve a distribution shift across the datasets of the
different clients, which hurts the generalization ability of the clients to
unseen samples from their respective data distributions. In this work, we
address the recently proposed feature shift problem where the clients have
different feature distributions while the label distribution is the same. We
propose Federated Representation Augmentation (FRAug) to tackle this practical
and challenging problem. Our approach generates synthetic client-specific
samples in the embedding space to augment the usually small client datasets.
For that, we train a shared generative model to fuse the clients' knowledge,
learned from different feature distributions, to synthesize client-agnostic
embeddings, which are then locally transformed into client-specific embeddings
by Representation Transformation Networks (RTNets). By transferring knowledge
across the clients, the generated embeddings act as a regularizer for the
client models and reduce overfitting to the local original datasets, hence
improving generalization. Our empirical evaluation on multiple benchmark
datasets demonstrates the effectiveness of the proposed method, which
substantially outperforms the current state-of-the-art FL methods for non-IID
features, including PartialFed and FedBN. |
3,475 | Daisy Bloom Filters | Weighted Bloom filters (Bruck, Gao and Jiang, ISIT 2006) are Bloom filters
that adapt the number of hash functions according to the query element. That
is, they use a sequence of hash functions $h_1, h_2, \dots$ and insert $x$ by
setting the bits in $k_x$ positions $h_1(x), h_2(x), \dots, h_{k_x}(x)$ to 1,
where the parameter $k_x$ depends on $x$. Similarly, a query for $x$ checks
whether the bits at positions $h_1(x), h_2(x), \dots, h_{k_x}(x)$ contain a $0$
(in which case we know that $x$ was not inserted), or contains only $1$s (in
which case $x$ may have been inserted, but it could also be a false positive).
In this paper, we determine a near-optimal choice of the parameters $k_x$ in
a model where $n$ elements are inserted independently from a probability
distribution $\mathcal{P}$ and query elements are chosen from a probability
distribution $\mathcal{Q}$, under a bound on the false positive probability
$F$. In contrast, the parameter choice of Bruck et al., as well as follow-up
work by Wang et al., does not guarantee a nontrivial bound on the false
positive rate. We refer to our parameterization of the weighted Bloom filter as
a $\textit{Daisy Bloom filter}$.
For many distributions $\mathcal{P}$ and $\mathcal{Q}$, the Daisy Bloom
filter space usage is significantly smaller than that of Standard Bloom
filters. Our upper bound is complemented with an information-theoretical lower
bound, showing that (with mild restrictions on the distributions $\mathcal{P}$
and $\mathcal{Q}$), the space usage of Daisy Bloom filters is the best possible
up to a constant factor.
Daisy Bloom filters can be seen as a fine-grained variant of a recent data
structure of Vaidya, Knorr, Mitzenmacher and Kraska. Like their work, we are
motivated by settings in which we have prior knowledge of the workload of the
filter, possibly in the form of advice from a machine learning algorithm. |
3,476 | Exploring the Open World Using Incremental Extreme Value Machines | Dynamic environments require adaptive applications. One particular machine
learning problem in dynamic environments is open world recognition. It
characterizes a continuously changing domain where only some classes are seen
in one batch of the training data and such batches can only be learned
incrementally. Open world recognition is a demanding task that is, to the best
of our knowledge, addressed by only a few methods. This work introduces a
modification of the widely known Extreme Value Machine (EVM) to enable open
world recognition. Our proposed method extends the EVM with a partial model
fitting function by neglecting unaffected space during an update. This reduces
the training time by a factor of 28. In addition, we provide a modified model
reduction using weighted maximum K-set cover to strictly bound the model
complexity and reduce the computational effort by a factor of 3.5 from 2.1 s to
0.6 s. In our experiments, we rigorously evaluate openness with two novel
evaluation protocols. The proposed method achieves superior accuracy of about
12 % and computational efficiency in the tasks of image classification and face
recognition. |
3,477 | Blind Estimation of a Doubly Selective OFDM Channel: A Deep Learning Algorithm and Theory | We provide a new generation solution to the fundamental old problem of a
doubly selective fading channel estimation for orthogonal frequency division
multiplexing (OFDM) systems. For systems based on OFDM, we propose a deep
learning (DL)-based blind doubly selective channel estimator. This estimator
does require no pilot symbols, unlike the corresponding state-of-the-art
estimators, even during the estimation of a deep fading doubly selective
channel. We also provide the first of its kind theory on the testing mean
squared error (MSE) performance of our investigated blind OFDM channel
estimator based on over-parameterized ReLU FNNs. |
3,478 | Neural Shape Mating: Self-Supervised Object Assembly with Adversarial Shape Priors | Learning to autonomously assemble shapes is a crucial skill for many robotic
applications. While the majority of existing part assembly methods focus on
correctly posing semantic parts to recreate a whole object, we interpret
assembly more literally: as mating geometric parts together to achieve a snug
fit. By focusing on shape alignment rather than semantic cues, we can achieve
across-category generalization. In this paper, we introduce a novel task,
pairwise 3D geometric shape mating, and propose Neural Shape Mating (NSM) to
tackle this problem. Given the point clouds of two object parts of an unknown
category, NSM learns to reason about the fit of the two parts and predict a
pair of 3D poses that tightly mate them together. We couple the training of NSM
with an implicit shape reconstruction task to make NSM more robust to imperfect
point cloud observations. To train NSM, we present a self-supervised data
collection pipeline that generates pairwise shape mating data with ground truth
by randomly cutting an object mesh into two parts, resulting in a dataset that
consists of 200K shape mating pairs from numerous object meshes with diverse
cut types. We train NSM on the collected dataset and compare it with several
point cloud registration methods and one part assembly baseline. Extensive
experimental results and ablation studies under various settings demonstrate
the effectiveness of the proposed algorithm. Additional material is available
at: https://neural-shape-mating.github.io/ |
3,479 | Adversarial synthesis based data-augmentation for code-switched spoken language identification | Spoken Language Identification (LID) is an important sub-task of Automatic
Speech Recognition(ASR) that is used to classify the language(s) in an audio
segment. Automatic LID plays an useful role in multilingual countries. In
various countries, identifying a language becomes hard, due to the multilingual
scenario where two or more than two languages are mixed together during
conversation. Such phenomenon of speech is called as code-mixing or
code-switching. This nature is followed not only in India but also in many
Asian countries. Such code-mixed data is hard to find, which further reduces
the capabilities of the spoken LID. Hence, this work primarily addresses this
problem using data augmentation as a solution on the on the data scarcity of
the code-switched class. This study focuses on Indic language code-mixed with
English. Spoken LID is performed on Hindi, code-mixed with English. This
research proposes Generative Adversarial Network (GAN) based data augmentation
technique performed using Mel spectrograms for audio data. GANs have already
been proven to be accurate in representing the real data distribution in the
image domain. Proposed research exploits these capabilities of GANs in speech
domains such as speech classification, automatic speech recognition, etc. GANs
are trained to generate Mel spectrograms of the minority code-mixed class which
are then used to augment data for the classifier. Utilizing GANs give an
overall improvement on Unweighted Average Recall by an amount of 3.5% as
compared to a Convolutional Recurrent Neural Network (CRNN) classifier used as
the baseline reference. |
3,480 | Measuring and mitigating voting access disparities: a study of race and polling locations in Florida and North Carolina | Voter suppression and associated racial disparities in access to voting are
long-standing civil rights concerns in the United States. Barriers to voting
have taken many forms over the decades. A history of violent explicit
discouragement has shifted to more subtle access limitations that can include
long lines and wait times, long travel times to reach a polling station, and
other logistical barriers to voting. Our focus in this work is on quantifying
disparities in voting access pertaining to the overall time-to-vote, and how
they could be remedied via a better choice of polling location or provisioning
more sites where voters can cast ballots. However, appropriately calibrating
access disparities is difficult because of the need to account for factors such
as population density and different community expectations for reasonable
travel times.
In this paper, we quantify access to polling locations, developing a
methodology for the calibrated measurement of racial disparities in polling
location "load" and distance to polling locations. We apply this methodology to
a study of real-world data from Florida and North Carolina to identify
disparities in voting access from the 2020 election. We also introduce
algorithms, with modifications to handle scale, that can reduce these
disparities by suggesting new polling locations from a given list of identified
public locations (including schools and libraries). Applying these algorithms
on the 2020 election location data also helps to expose and explore tradeoffs
between the cost of allocating more polling locations and the potential impact
on access disparities. The developed voting access measurement methodology and
algorithmic remediation technique is a first step in better polling location
assignment. |
3,481 | Rites de Passage: Elucidating Displacement to Emplacement of Refugees on Twitter | Social media deliberations allow to explore refugee-related is-sues. AI-based
studies have investigated refugee issues mostly around a specific event and
considered unimodal approaches. Contrarily, we have employed a multimodal
architecture for probing the refugee journeys from their home to host nations.
We draw insights from Arnold van Gennep's anthropological work 'Les Rites de
Passage', which systematically analyzed an individual's transition from one
group or society to another. Based on Gennep's
separation-transition-incorporation framework, we have identified four phases
of refugee journeys: Arrival of Refugees, Temporal stay at Asylums,
Rehabilitation, and Integration of Refugees into the host nation. We collected
0.23 million multimodal tweets from April 2020 to March 2021 for testing this
proposed frame-work. We find that a combination of transformer-based language
models and state-of-the-art image recognition models, such as fusion of
BERT+LSTM and InceptionV4, can out-perform unimodal models. Subsequently, to
test the practical implication of our proposed model in real-time, we have
considered 0.01 million multimodal tweets related to the 2022 Ukrainian refugee
crisis. An F1-score of 71.88 % for this 2022 crisis confirms the
generalizability of our proposed framework. |
3,482 | Leave-one-out Singular Subspace Perturbation Analysis for Spectral Clustering | The singular subspaces perturbation theory is of fundamental importance in
probability and statistics. It has various applications across different
fields. We consider two arbitrary matrices where one is a leave-one-column-out
submatrix of the other one and establish a novel perturbation upper bound for
the distance between two corresponding singular subspaces. It is well-suited
for mixture models and results in a sharper and finer statistical analysis than
classical perturbation bounds such as Wedin's Theorem. Powered by this
leave-one-out perturbation theory, we provide a deterministic entrywise
analysis for the performance of the spectral clustering under mixture models.
Our analysis leads to an explicit exponential error rate for the clustering of
sub-Gaussian mixture models. For the mixture of isotropic Gaussians, the rate
is optimal under a weaker signal-to-noise condition than that of L\"offler et
al. (2021). |
3,483 | Anti-virus Autobots: Predicting More Infectious Virus Variants for Pandemic Prevention through Deep Learning | More infectious virus variants can arise from rapid mutations in their
proteins, creating new infection waves. These variants can evade one's immune
system and infect vaccinated individuals, lowering vaccine efficacy. Hence, to
improve vaccine design, this project proposes Optimus PPIme - a deep learning
approach to predict future, more infectious variants from an existing virus
(exemplified by SARS-CoV-2). The approach comprises an algorithm which acts as
a "virus" attacking a host cell. To increase infectivity, the "virus" mutates
to bind better to the host's receptor. 2 algorithms were attempted - greedy
search and beam search. The strength of this variant-host binding was then
assessed by a transformer network we developed, with a high accuracy of 90%.
With both components, beam search eventually proposed more infectious variants.
Therefore, this approach can potentially enable researchers to develop vaccines
that provide protection against future infectious variants before they emerge,
pre-empting outbreaks and saving lives. |
3,484 | Play it by Ear: Learning Skills amidst Occlusion through Audio-Visual Imitation Learning | Humans are capable of completing a range of challenging manipulation tasks
that require reasoning jointly over modalities such as vision, touch, and
sound. Moreover, many such tasks are partially-observed; for example, taking a
notebook out of a backpack will lead to visual occlusion and require reasoning
over the history of audio or tactile information. While robust tactile sensing
can be costly to capture on robots, microphones near or on a robot's gripper
are a cheap and easy way to acquire audio feedback of contact events, which can
be a surprisingly valuable data source for perception in the absence of vision.
Motivated by the potential for sound to mitigate visual occlusion, we aim to
learn a set of challenging partially-observed manipulation tasks from visual
and audio inputs. Our proposed system learns these tasks by combining offline
imitation learning from a modest number of tele-operated demonstrations and
online finetuning using human provided interventions. In a set of simulated
tasks, we find that our system benefits from using audio, and that by using
online interventions we are able to improve the success rate of offline
imitation learning by ~20%. Finally, we find that our system can complete a set
of challenging, partially-observed tasks on a Franka Emika Panda robot, like
extracting keys from a bag, with a 70% success rate, 50% higher than a policy
that does not use audio. |
3,485 | Precise Learning Curves and Higher-Order Scaling Limits for Dot Product Kernel Regression | As modern machine learning models continue to advance the computational
frontier, it has become increasingly important to develop precise estimates for
expected performance improvements under different model and data scaling
regimes. Currently, theoretical understanding of the learning curves that
characterize how the prediction error depends on the number of samples is
restricted to either large-sample asymptotics ($m\to\infty$) or, for certain
simple data distributions, to the high-dimensional asymptotics in which the
number of samples scales linearly with the dimension ($m\propto d$). There is a
wide gulf between these two regimes, including all higher-order scaling
relations $m\propto d^r$, which are the subject of the present paper. We focus
on the problem of kernel ridge regression for dot-product kernels and present
precise formulas for the test error, bias, and variance, for data drawn
uniformly from the sphere in the $r$th-order asymptotic scaling regime
$m\to\infty$ with $m/d^r$ held constant. We observe a peak in the learning
curve whenever $m \approx d^r/r!$ for any integer $r$, leading to multiple
sample-wise descent and nontrivial behavior at multiple scales. |
3,486 | Efficient Reward Poisoning Attacks on Online Deep Reinforcement Learning | We study data poisoning attacks on online deep reinforcement learning (DRL)
where the attacker is oblivious to the learning algorithm used by the agent and
does not necessarily have full knowledge of the environment. We demonstrate the
intrinsic vulnerability of state-of-the-art DRL algorithms by designing a
general reward poisoning framework called adversarial MDP attacks. We
instantiate our framework to construct several new attacks which only corrupt
the rewards for a small fraction of the total training timesteps and make the
agent learn a low-performing policy. Our key insight is that the
state-of-the-art DRL algorithms strategically explore the environment to find a
high-performing policy. Our attacks leverage this insight to construct a
corrupted environment for misleading the agent towards learning low-performing
policies with a limited attack budget. We provide a theoretical analysis of the
efficiency of our attack and perform an extensive evaluation. Our results show
that our attacks efficiently poison agents learning with a variety of
state-of-the-art DRL algorithms, such as DQN, PPO, SAC, etc. under several
popular classical control and MuJoCo environments. |
3,487 | To Federate or Not To Federate: Incentivizing Client Participation in Federated Learning | Federated learning (FL) facilitates collaboration between a group of clients
who seek to train a common machine learning model without directly sharing
their local data. Although there is an abundance of research on improving the
speed, efficiency, and accuracy of federated training, most works implicitly
assume that all clients are willing to participate in the FL framework. Due to
data heterogeneity, however, the global model may not work well for some
clients, and they may instead choose to use their own local model. Such
disincentivization of clients can be problematic from the server's perspective
because having more participating clients yields a better global model, and
offers better privacy guarantees to the participating clients. In this paper,
we propose an algorithm called IncFL that explicitly maximizes the fraction of
clients who are incentivized to use the global model by dynamically adjusting
the aggregation weights assigned to their updates. Our experiments show that
IncFL increases the number of incentivized clients by 30-55% compared to
standard federated training algorithms, and can also improve the generalization
performance of the global model on unseen clients. |
3,488 | Adversarial Bandits Robust to Switching Targets | We study the adversarial bandit problem under $S$ number of switching best
arms for unknown $S$. For handling this problem, we adopt the master-base
framework using the online mirror descent method (OMD). We first provide a
master-base algorithm with basic OMD, achieving
$\tilde{O}(S^{1/2}K^{1/3}T^{2/3})$. For improving the regret bound with respect
to $T$, we propose to use adaptive learning rates for OMD to control variance
of loss estimators, and achieve
$\tilde{O}(\min\{\mathbb{E}[\sqrt{SKT\rho_T(h^\dagger)}],S\sqrt{KT}\})$, where
$\rho_T(h^\dagger)$ is a variance term for loss estimators. |
3,489 | GraMeR: Graph Meta Reinforcement Learning for Multi-Objective Influence Maximization | Influence maximization (IM) is a combinatorial problem of identifying a
subset of nodes called the seed nodes in a network (graph), which when
activated, provide a maximal spread of influence in the network for a given
diffusion model and a budget for seed set size. IM has numerous applications
such as viral marketing, epidemic control, sensor placement and other
network-related tasks. However, the uses are limited due to the computational
complexity of current algorithms. Recently, learning heuristics for IM have
been explored to ease the computational burden. However, there are serious
limitations in current approaches such as: (1) IM formulations only consider
influence via spread and ignore self activation; (2) scalability to large
graphs; (3) generalizability across graph families; (4) low computational
efficiency with a large running time to identify seed sets for every test
network. In this work, we address each of these limitations through a unique
approach that involves (1) formulating a generic IM problem as a Markov
decision process that handles both intrinsic and influence activations; (2)
employing double Q learning to estimate seed nodes; (3) ensuring scalability
via sub-graph based representations; and (4) incorporating generalizability via
meta-learning across graph families. Extensive experiments are carried out in
various standard networks to validate performance of the proposed Graph Meta
Reinforcement learning (GraMeR) framework. The results indicate that GraMeR is
multiple orders faster and generic than conventional approaches. |
3,490 | Walle: An End-to-End, General-Purpose, and Large-Scale Production System for Device-Cloud Collaborative Machine Learning | To break the bottlenecks of mainstream cloud-based machine learning (ML)
paradigm, we adopt device-cloud collaborative ML and build the first end-to-end
and general-purpose system, called Walle, as the foundation. Walle consists of
a deployment platform, distributing ML tasks to billion-scale devices in time;
a data pipeline, efficiently preparing task input; and a compute container,
providing a cross-platform and high-performance execution environment, while
facilitating daily task iteration. Specifically, the compute container is based
on Mobile Neural Network (MNN), a tensor compute engine along with the data
processing and model execution libraries, which are exposed through a refined
Python thread-level virtual machine (VM) to support diverse ML tasks and
concurrent task execution. The core of MNN is the novel mechanisms of operator
decomposition and semi-auto search, sharply reducing the workload in manually
optimizing hundreds of operators for tens of hardware backends and further
quickly identifying the best backend with runtime optimization for a
computation graph. The data pipeline introduces an on-device stream processing
framework to enable processing user behavior data at source. The deployment
platform releases ML tasks with an efficient push-then-pull method and supports
multi-granularity deployment policies. We evaluate Walle in practical
e-commerce application scenarios to demonstrate its effectiveness, efficiency,
and scalability. Extensive micro-benchmarks also highlight the superior
performance of MNN and the Python thread-level VM. Walle has been in
large-scale production use in Alibaba, while MNN has been open source with a
broad impact in the community. |
3,491 | Temporal Multiresolution Graph Neural Networks For Epidemic Prediction | In this paper, we introduce Temporal Multiresolution Graph Neural Networks
(TMGNN), the first architecture that both learns to construct the multiscale
and multiresolution graph structures and incorporates the time-series signals
to capture the temporal changes of the dynamic graphs. We have applied our
proposed model to the task of predicting future spreading of epidemic and
pandemic based on the historical time-series data collected from the actual
COVID-19 pandemic and chickenpox epidemic in several European countries, and
have obtained competitive results in comparison to other previous
state-of-the-art temporal architectures and graph learning algorithms. We have
shown that capturing the multiscale and multiresolution structures of graphs is
important to extract either local or global information that play a critical
role in understanding the dynamic of a global pandemic such as COVID-19 which
started from a local city and spread to the whole world. Our work brings a
promising research direction in forecasting and mitigating future epidemics and
pandemics. |
3,492 | Adaptive Learning for Discovery | In this paper, we study a sequential decision-making problem, called Adaptive
Sampling for Discovery (ASD). Starting with a large unlabeled dataset,
algorithms for ASD adaptively label the points with the goal to maximize the
sum of responses.
This problem has wide applications to real-world discovery problems, for
example drug discovery with the help of machine learning models. ASD algorithms
face the well-known exploration-exploitation dilemma. The algorithm needs to
choose points that yield information to improve model estimates but it also
needs to exploit the model. We rigorously formulate the problem and propose a
general information-directed sampling (IDS) algorithm. We provide theoretical
guarantees for the performance of IDS in linear, graph and low-rank models. The
benefits of IDS are shown in both simulation experiments and real-data
experiments for discovering chemical reaction conditions. |
3,493 | Lepton Flavour Violation Identification in Tau Decay ($τ^{-} \rightarrow μ^{-}μ^{-}μ^{+}$) Using Artificial Intelligence | The discovery of neutrino oscillation, proving that neutrinos do have masses,
reveals the misfits of particles in the current Standard Model (SM) theory. In
theory, neutrinos having masses could result in lepton flavour not being a
symmetry called Lepton Flavour Violation (LFV). While SM theory extensions
allowed LFV processes, their branching fractions are too small, making them
unobservable even with the strongest equipment up-to-date. With that,
scientists in recent years have generated LFV-like processes from the combined
LHCb and Monte-Carlo-Simulated data in an attempt to identify LFV using
Artificial Intelligence (AI), specifically Machine Learning (ML) and Deep
Learning (DL). In this paper, the performance of several algorithms in AI has
been presented, such as XGBoost, LightGBM, custom 1-D Dense Block Neural
Networks (DBNNs), and custom 1-D Convolutional Neural Networks (CNNs) in
identifying LFV signals, specifically $\tau^{-} \rightarrow
\mu^{-}\mu^{-}\mu^{+}$ decay from the combined LHCb and Monte-Carlo-Simulated
data that imitates the signatures of the said decay. Kolmogorov-Smirnov (KS)
and Cramer-von Mises (CvM) tests were also conducted to verify the validity of
predictions for each of the trained algorithms. The result shows decent
performances among algorithms, except for the LightGBM, for failing the CvM
test, and a 20-layered CNN for having recorded a considerably low AUC.
Meanwhile, XGBoost and a 10-layered DBNN recorded the highest AUC of 0.88. The
main contribution of this paper is the extensive experiment involving custom
DBNN and CNN algorithms in different layers, all of which have been rarely used
in the past years in identifying LFV-like signatures, unlike GBMs and
tree-based algorithms, which have been more popular in the said task. |
3,494 | Robust Weight Perturbation for Adversarial Training | Overfitting widely exists in adversarial robust training of deep networks. An
effective remedy is adversarial weight perturbation, which injects the
worst-case weight perturbation during network training by maximizing the
classification loss on adversarial examples. Adversarial weight perturbation
helps reduce the robust generalization gap; however, it also undermines the
robustness improvement. A criterion that regulates the weight perturbation is
therefore crucial for adversarial training. In this paper, we propose such a
criterion, namely Loss Stationary Condition (LSC) for constrained perturbation.
With LSC, we find that it is essential to conduct weight perturbation on
adversarial data with small classification loss to eliminate robust
overfitting. Weight perturbation on adversarial data with large classification
loss is not necessary and may even lead to poor robustness. Based on these
observations, we propose a robust perturbation strategy to constrain the extent
of weight perturbation. The perturbation strategy prevents deep networks from
overfitting while avoiding the side effect of excessive weight perturbation,
significantly improving the robustness of adversarial training. Extensive
experiments demonstrate the superiority of the proposed method over the
state-of-the-art adversarial training methods. |
3,495 | Bayesian Low-Rank Interpolative Decomposition for Complex Datasets | In this paper, we introduce a probabilistic model for learning interpolative
decomposition (ID), which is commonly used for feature selection, low-rank
approximation, and identifying hidden patterns in data, where the matrix
factors are latent variables associated with each data dimension. Prior
densities with support on the specified subspace are used to address the
constraint for the magnitude of the factored component of the observed matrix.
Bayesian inference procedure based on Gibbs sampling is employed. We evaluate
the model on a variety of real-world datasets including CCLE EC50, CCLE IC50,
CTRP EC50,and MovieLens 100K datasets with different sizes, and dimensions, and
show that the proposed Bayesian ID GBT and GBTN models lead to smaller
reconstructive errors compared to existing randomized approaches. |
3,496 | Universality of group convolutional neural networks based on ridgelet analysis on groups | We investigate the approximation property of group convolutional neural
networks (GCNNs) based on the ridgelet theory. We regard a group convolution as
a matrix element of a group representation, and formulate a versatile GCNN as a
nonlinear mapping between group representations, which covers typical GCNN
literatures such as a cyclic convolution on a multi-channel image,
permutation-invariant datasets (Deep Sets), and $\mathrm{E}(n)$-equivariant
convolutions. The ridgelet transform is an analysis operator of a depth-2
network, namely, it maps an arbitrary given target function $f$ to the weight
$\gamma$ of a network $S[\gamma]$ so that the network represents the function
as $S[\gamma]=f$. It has been known only for fully-connected networks, and this
study is the first to present the ridgelet transform for (G)CNNs. Since the
ridgelet transform is given as a closed-form integral operator, it provides a
constructive proof of the $cc$-universality of GCNNs. Unlike previous
universality arguments on CNNs, we do not need to convert/modify the networks
into other universal approximators such as invariant polynomials and
fully-connected networks. |
3,497 | Excess Risk of Two-Layer ReLU Neural Networks in Teacher-Student Settings and its Superiority to Kernel Methods | While deep learning has outperformed other methods for various tasks,
theoretical frameworks that explain its reason have not been fully established.
To address this issue, we investigate the excess risk of two-layer ReLU neural
networks in a teacher-student regression model, in which a student network
learns an unknown teacher network through its outputs. Especially, we consider
the student network that has the same width as the teacher network and is
trained in two phases: first by noisy gradient descent and then by the vanilla
gradient descent. Our result shows that the student network provably reaches a
near-global optimal solution and outperforms any kernel methods estimator (more
generally, linear estimators), including neural tangent kernel approach, random
feature model, and other kernel methods, in a sense of the minimax optimal
rate. The key concept inducing this superiority is the non-convexity of the
neural network models. Even though the loss landscape is highly non-convex, the
student network adaptively learns the teacher neurons. |
3,498 | Mitigating Out-of-Distribution Data Density Overestimation in Energy-Based Models | Deep energy-based models (EBMs), which use deep neural networks (DNNs) as
energy functions, are receiving increasing attention due to their ability to
learn complex distributions. To train deep EBMs, the maximum likelihood
estimation (MLE) with short-run Langevin Monte Carlo (LMC) is often used. While
the MLE with short-run LMC is computationally efficient compared to an MLE with
full Markov Chain Monte Carlo (MCMC), it often assigns high density to
out-of-distribution (OOD) data. To address this issue, here we systematically
investigate why the MLE with short-run LMC can converge to EBMs with wrong
density estimates, and reveal that the heuristic modifications to LMC
introduced by previous works were the main problem. We then propose a Uniform
Support Partitioning (USP) scheme that optimizes a set of points to evenly
partition the support of the EBM and then uses the resulting points to
approximate the EBM-MLE loss gradient. We empirically demonstrate that USP
avoids the pitfalls of short-run LMC, leading to significantly improved OOD
data detection performance on Fashion-MNIST. |
3,499 | Your Contrastive Learning Is Secretly Doing Stochastic Neighbor Embedding | Contrastive learning, especially Self-Supervised Contrastive Learning (SSCL),
has achieved great success in extracting powerful features from unlabeled data,
enabling comparable performance to the supervised counterpart. In this work, we
contribute to the theoretical understanding of SSCL and uncover its connection
to the classic data visualization method, Stochastic Neighbor Embedding (SNE).
In the perspective of SNE, whose goal is matching pairwise distance, SSCL can
be viewed as a special case with the input space pairwise distance specified by
constructed "positive" pairs from data augmentation. The established
correspondence facilitates deeper theoretical understandings of learned
features of SSCL, as well as methodological guidelines for practical
improvement. Specifically, through the lens of SNE, not only can we re-derive
the alignment and uniformity principle, but also provide novel analysis on
domain-agnostic augmentations and implicit bias. To illustrate the practical
advantage, we demonstrate that the modifications from SNE to $t$-SNE can also
be adopted in the SSCL setting, achieving significant improvement in both
in-distribution and out-of-distribution generalization. |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.