paper_id
stringlengths
9
13
title
stringlengths
15
153
paper_url
stringlengths
41
45
authors
listlengths
1
21
type
stringclasses
3 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
310
2.43k
keywords
listlengths
1
32
TL;DR
large_stringlengths
23
250
submission_number
int64
1
11.7k
arxiv_id
stringlengths
10
10
arxiv_id_source
stringclasses
2 values
ZarM_uLVyGw
Contrastive Reinforcement Learning of Symbolic Reasoning Domains
https://openreview.net/forum?id=ZarM_uLVyGw
[ "Gabriel Poesia", "WenXin Dong", "Noah Goodman" ]
Poster
null
Abstract symbolic reasoning, as required in domains such as mathematics and logic, is a key component of human intelligence. Solvers for these domains have important applications, especially to computer-assisted education. But learning to solve symbolic problems is challenging for machine learning algorithms. Existing ...
[ "reinforcement learning", "education", "contrastive learning", "symbolic reasoning" ]
We propose a novel algorithm for reinforcement learning in discrete symbolic domains based on contrastive learning.
8,024
2106.09146
title_snapshot
81Erd42Wimi
Hyperparameter Tuning is All You Need for LISTA
https://openreview.net/forum?id=81Erd42Wimi
[ "Xiaohan Chen", "Jialin Liu", "Zhangyang Wang", "Wotao Yin" ]
Poster
null
Learned Iterative Shrinkage-Thresholding Algorithm (LISTA) introduces the concept of unrolling an iterative algorithm and training it like a neural network. It has had great success on sparse recovery. In this paper, we show that adding momentum to intermediate variables in the LISTA network achieves a better convergen...
[ "LISTA", "unrolling" ]
null
8,022
2110.15900
title_snapshot
yTXtUSV-gk4
Ising Model Selection Using $\ell_{1}$-Regularized Linear Regression: A Statistical Mechanics Analysis
https://openreview.net/forum?id=yTXtUSV-gk4
[ "Xiangming Meng", "Tomoyuki Obuchi", "Yoshiyuki Kabashima" ]
Poster
null
We theoretically analyze the typical learning performance of $\ell_{1}$-regularized linear regression ($\ell_1$-LinR) for Ising model selection using the replica method from statistical mechanics. For typical random regular graphs in the paramagnetic phase, an accurate estimate of the typical sample complexity of $\ell...
[ "Ising model", "Lasso", "replica method", "statistical physics" ]
Ising Model Selection Using $\ell_{1}$-Regularized Linear Regression
8,021
2102.03988
title_snapshot
xRJ_Xqmb6d
Perturb-and-max-product: Sampling and learning in discrete energy-based models
https://openreview.net/forum?id=xRJ_Xqmb6d
[ "Miguel Lazaro-Gredilla", "Antoine Dedieu", "Dileep George" ]
Poster
null
Perturb-and-MAP offers an elegant approach to approximately sample from a energy-based model (EBM) by computing the maximum-a-posteriori (MAP) configuration of a perturbed version of the model. Sampling in turn enables learning. However, this line of research has been hindered by the general intractability of the MAP c...
[ "max-product", "perturb-and-map", "discrete graphical models", "energy based models", "belief revision", "belief propagation" ]
Perturb-and-MAP combined with max-product yields better performance than Gibbs sampling or LP-based MAP methods on discrete EBMs
8,020
2111.02458
title_snapshot
WnJXcebN7hX
An Exponential Lower Bound for Linearly Realizable MDP with Constant Suboptimality Gap
https://openreview.net/forum?id=WnJXcebN7hX
[ "Yuanhao Wang", "Ruosong Wang", "Sham M. Kakade" ]
Oral
null
A fundamental question in the theory of reinforcement learning is: suppose the optimal $Q$-function lies in the linear span of a given $d$ dimensional feature mapping, is sample-efficient reinforcement learning (RL) possible? The recent and remarkable result of Weisz et al. (2020) resolves this question in the negative...
[ "Reinforcement Learning", "Linear Function Approximation", "Lower Bound" ]
null
8,007
2103.12690
title_judge
hhkyM3ib9e6
Permuton-induced Chinese Restaurant Process
https://openreview.net/forum?id=hhkyM3ib9e6
[ "Masahiro Nakano", "Yasuhiro Fujiwara", "Akisato Kimura", "Takeshi Yamada", "Naonori Ueda" ]
Poster
null
This paper proposes the permuton-induced Chinese restaurant process (PCRP), a stochastic process on rectangular partitioning of a matrix. This distribution is suitable for use as a prior distribution in Bayesian nonparametric relational model to find hidden clusters in matrices and network data. Our main contribution i...
[ "Bayesian nonparametrics", "Chinese restaurant process", "Relational model" ]
Multi-dimensional extension of the Chinese restaurant process, whose table coordinates arranged by the permuton.
7,987
null
null
Z_J5bCb4Rra
Divergence Frontiers for Generative Models: Sample Complexity, Quantization Effects, and Frontier Integrals
https://openreview.net/forum?id=Z_J5bCb4Rra
[ "Lang Liu", "Krishna Pillutla", "Sean Welleck", "Sewoong Oh", "Yejin Choi", "Zaid Harchaoui" ]
Poster
null
The spectacular success of deep generative models calls for quantitative tools to measure their statistical performance. Divergence frontiers have recently been proposed as an evaluation framework for generative models, due to their ability to measure the quality-diversity trade-off inherent to deep generative modeling...
[ "sample complexity", "precision-recall", "generative modeling", "divergence frontiers" ]
This paper studies the statistical behavior of the divergence frontiers and its statistical summary for evaluating generative models, where the sample complexity and the choice of quantization level are investigated.
7,985
2106.07898
title_snapshot
YOc9i6-NrQk
Topological Relational Learning on Graphs
https://openreview.net/forum?id=YOc9i6-NrQk
[ "Yuzhou Chen", "Baris Coskunuzer", "Yulia Gel" ]
Poster
null
Graph neural networks (GNNs) have emerged as a powerful tool for graph classification and representation learning. However, GNNs tend to suffer from over-smoothing problems and are vulnerable to graph perturbations. To address these challenges, we propose a novel topological neural framework of topological relational i...
[ "Topological relational learning", "Graph neural networks", "Persistence based graph representation", "Robust graph learning" ]
We introduce a new local topological representation of graph to enhance robustness of Graph Neural Networks in classification tasks and derive its theoretical guarrantees
7,973
2110.15529
title_snapshot
5-Of1DTlq
Preconditioned Gradient Descent for Over-Parameterized Nonconvex Matrix Factorization
https://openreview.net/forum?id=5-Of1DTlq
[ "Gavin Zhang", "Salar Fattahi", "Richard Y. Zhang" ]
Poster
null
In practical instances of nonconvex matrix factorization, the rank of the true solution $r^{\star}$ is often unknown, so the rank $r$ of the model can be over-specified as $r>r^{\star}$. This over-parameterized regime of matrix factorization significantly slows down the convergence of local search algorithms, from a li...
[ "low-rank matrix factorization", "matrix sensing", "nonconvex optimization", "semidefinite programming", "gradient descent" ]
We propose a preconditioned gradient descent that enjoys a provable linear convergence rate for over-parameterized low-rank matrix factorization.
7,969
2504.09708
title_snapshot
qPOeyokHXT8
Provably efficient multi-task reinforcement learning with model transfer
https://openreview.net/forum?id=qPOeyokHXT8
[ "Chicheng Zhang", "Zhi Wang" ]
Poster
null
We study multi-task reinforcement learning (RL) in tabular episodic Markov decision processes (MDPs). We formulate a heterogeneous multi-player RL problem, in which a group of players concurrently face similar but not necessarily identical MDPs, with a goal of improving their collective performance through inter-player...
[ "Multi-task learning", "Provably efficient reinforcement learning", "Model transfer" ]
We formulate a new multi-task episodic reinforcement learning problem, and provide a provably efficient algorithm using model transfer in this setting.
7,965
2107.08622
title_snapshot
Rav_oC35ToB
Locality Sensitive Teaching
https://openreview.net/forum?id=Rav_oC35ToB
[ "Zhaozhuo Xu", "Beidi Chen", "Chaojian Li", "Weiyang Liu", "Le Song", "Yingyan Lin", "Anshumali Shrivastava" ]
Poster
null
The emergence of the Internet-of-Things (IoT) sheds light on applying the machine teaching (MT) algorithms for online personalized education on home devices. This direction becomes more promising during the COVID-19 pandemic when in-person education becomes infeasible. However, as one of the most influential and practi...
[ "Locality Sensitive Hashing", "Machine Teaching", "Internet-of-Things" ]
An efficient machine teaching algorithm for edge devices.
7,954
null
null
xfDXF0I_bt
Locally Valid and Discriminative Prediction Intervals for Deep Learning Models
https://openreview.net/forum?id=xfDXF0I_bt
[ "Zhen Lin", "Shubhendu Trivedi", "Jimeng Sun" ]
Poster
null
Crucial for building trust in deep learning models for critical real-world applications is efficient and theoretically sound uncertainty quantification, a task that continues to be challenging. Useful uncertainty information is expected to have two key properties: It should be valid (guaranteeing coverage) and discrimi...
[ "Deep Learning", "Local Validity", "Conformal Prediction", "Uncertainty Quantification", "Prediction Interval" ]
null
7,941
2106.00225
title_snapshot
jBQaRXpEgO
Can Information Flows Suggest Targets for Interventions in Neural Circuits?
https://openreview.net/forum?id=jBQaRXpEgO
[ "Praveen Venkatesh", "Sanghamitra Dutta", "Neil Mehta", "Pulkit Grover" ]
Poster
null
Motivated by neuroscientific and clinical applications, we empirically examine whether observational measures of information flow can suggest interventions. We do so by performing experiments on artificial neural networks in the context of fairness in machine learning, where the goal is to induce fairness in the system...
[ "information flow", "interventions", "neuroscience", "information theory", "explainability", "fairness" ]
To test whether observational measures of information flow can predict the outcomes of interventions in neuroscience, we perform analogous experiments on artificial neural networks in a fairness context and analyze bias-accuracy tradeoffs.
7,933
2111.05299
title_snapshot
iPHnzuU6S94
Neural Active Learning with Performance Guarantees
https://openreview.net/forum?id=iPHnzuU6S94
[ "Zhilei Wang", "Pranjal Awasthi", "Christoph Dann", "Ayush Sekhari", "Claudio Gentile" ]
Poster
null
We investigate the problem of active learning in the streaming setting in non-parametric regimes, where the labels are stochastically generated from a class of functions on which we make no assumptions whatsoever. We rely on recently proposed Neural Tangent Kernel (NTK) approximation tools to construct a suitable neura...
[ "Active learning", "Deep Neural Networks", "Neural Tangent Kernel", "Selective Sampling", "theoretical guarantees" ]
We investigate online active learning in non-parametric regimes under the NTK approximation, and derive theoretical guarantees for the proposed algorithms.
7,927
2106.03243
title_snapshot
l41jc6kUfKr
Pointwise Bounds for Distribution Estimation under Communication Constraints
https://openreview.net/forum?id=l41jc6kUfKr
[ "Wei-Ning Chen", "Peter Kairouz", "Ayfer Ozgur" ]
Poster
null
We consider the problem of estimating a $d$-dimensional discrete distribution from its samples observed under a $b$-bit communication constraint. In contrast to most previous results that largely focus on the global minimax error, we study the local behavior of the estimation error and provide \emph{pointwise} bounds t...
[ "distributed estimation", "communication complexity", "instance-optimal" ]
We characterize instance-optimal bounds on estimating distributions under communication constraints.
7,923
2110.03189
title_snapshot
6mUrD5rg-UU
Does enforcing fairness mitigate biases caused by subpopulation shift?
https://openreview.net/forum?id=6mUrD5rg-UU
[ "Subha Maity", "Debarghya Mukherjee", "Mikhail Yurochkin", "Yuekai Sun" ]
Poster
null
Many instances of algorithmic bias are caused by subpopulation shifts. For example, ML models often perform worse on demographic groups that are underrepresented in the training data. In this paper, we study whether enforcing algorithmic fairness during training improves the performance of the trained model in the \emp...
[ "algorithmic fairness", "domain adaptation", "sub-population shift", "fairness accuracy trade-off" ]
The paper presents a precise characterization of trade-off between accuracy and fairness under sub-population shift.
7,919
2011.03173
title_snapshot
FEhntTXAeHN
Object-Aware Regularization for Addressing Causal Confusion in Imitation Learning
https://openreview.net/forum?id=FEhntTXAeHN
[ "Jongjin Park", "Younggyo Seo", "Chang Liu", "Li Zhao", "Tao Qin", "Jinwoo Shin", "Tie-Yan Liu" ]
Poster
null
Behavioral cloning has proven to be effective for learning sequential decision-making policies from expert demonstrations. However, behavioral cloning often suffers from the causal confusion problem where a policy relies on the noticeable effect of expert actions due to the strong correlation but not the cause we desir...
[ "imitation learning", "behavioral cloning", "causal confusion", "regularization" ]
We propose an object-aware regularization technique to address the causal confusion problem in imitation learning.
7,914
2110.14118
title_snapshot
-mGv2KxQ43D
Learning MDPs from Features: Predict-Then-Optimize for Sequential Decision Making by Reinforcement Learning
https://openreview.net/forum?id=-mGv2KxQ43D
[ "Kai Wang", "Sanket Shah", "Haipeng Chen", "Andrew Perrault", "Finale Doshi-Velez", "Milind Tambe" ]
Spotlight
null
In the predict-then-optimize framework, the objective is to train a predictive model, mapping from environment features to parameters of an optimization problem, which maximizes decision quality when the optimization is subsequently solved. Recent work on decision-focused learning shows that embedding the optimization ...
[ "Optimization", "predict-then-optimize", "decision-focused learning", "differentiable optimization", "reinforcement learning", "MDP", "low-rank approximation", "Woodbury matrix identity", "KKT conditions", "optimality conditions", "sequential decision problems" ]
We extend decision-focused learning to MDPs with missing parameters. The key novelty is to approximate Hessian to address the high computational cost of differentiating through MDP layers, which arises from large state-action and policy spaces.
7,910
2106.03279
title_judge
KvjtYlrmAj
Stateful ODE-Nets using Basis Function Expansions
https://openreview.net/forum?id=KvjtYlrmAj
[ "Alejandro Francisco Queiruga", "N. Benjamin Erichson", "Liam Hodgkinson", "Michael W. Mahoney" ]
Poster
null
The recently-introduced class of ordinary differential equation networks (ODE-Nets) establishes a fruitful connection between deep learning and dynamical systems. In this work, we reconsider formulations of the weights as continuous-in-depth functions using linear combinations of basis functions which enables us to lev...
[ "Neural ODEs", "Dynamical Systems", "Differential Equations" ]
We take advantage of basis transformations to introduce stateful normalization layers as well as a methodology for compressing ODE-Nets.
7,904
2106.10820
title_snapshot
nehzxAdyJxF
Evaluating model performance under worst-case subpopulations
https://openreview.net/forum?id=nehzxAdyJxF
[ "Mike Li", "Hongseok Namkoong", "Shangzhou Xia" ]
Poster
null
The performance of ML models degrades when the training population is different from that seen under operation. Towards assessing distributional robustness, we study the worst-case performance of a model over all subpopulations of a given size, defined with respect to core attributes $Z$. This notion of robustness can ...
[ "reliability", "robustness", "safety", "distribution shift", "subpopulation shift", "fairness" ]
We develop a diagnostic for evaluating model performance under subpopulation shifts.
7,897
2407.01316
title_snapshot
8YSqxvRhi-Q
Online Meta-Learning via Learning with Layer-Distributed Memory
https://openreview.net/forum?id=8YSqxvRhi-Q
[ "Sudarshan Babu", "Pedro Henrique Pamplona Savarese", "Michael Maire" ]
Poster
null
We demonstrate that efficient meta-learning can be achieved via end-to-end training of deep neural networks with memory distributed across layers. The persistent state of this memory assumes the entire burden of guiding task adaptation. Moreover, its distributed nature is instrumental in orchestrating adaptation. Ab...
[ "Meta-learning", "Memory models", "Online learning" ]
Efficient and simplified meta-learning via distributed memory networks.
7,881
null
null
_y2G1-i7L8
Asymptotically Exact Error Characterization of Offline Policy Evaluation with Misspecified Linear Models
https://openreview.net/forum?id=_y2G1-i7L8
[ "Kohei Miyaguchi" ]
Poster
null
We consider the problem of offline policy evaluation~(OPE) with Markov decision processes~(MDPs), where the goal is to estimate the utility of given decision-making policies based on static datasets. Recently, theoretical understanding of OPE has been rapidly advanced under (approximate) realizability assumptions, i.e....
[ "reinforcement learning", "offline policy evaluation", "linear function approximation" ]
We characterize the error of a simple linear OPE method when the function approximation may be completely wrong. As a result, we found a new interpretation and new error bounds.
7,879
null
null
j4oYd8SGop
Inverse-Weighted Survival Games
https://openreview.net/forum?id=j4oYd8SGop
[ "Xintian Han", "Mark Goldstein", "Aahlad Manas Puli", "Thomas Wies", "Adler J Perotte", "Rajesh Ranganath" ]
Poster
null
Deep models trained through maximum likelihood have achieved state-of-the-art results for survival analysis. Despite this training scheme, practitioners evaluate models under other criteria, such as binary classification losses at a chosen set of time horizons, e.g. Brier score (BS) and Bernoulli log likelihood (BLL). ...
[ "Survival Analysis", "Healthcare", "Censoring", "Games", "IPCW", "Brier Score" ]
A new inverse-weighted training method in survival analysis for optimizing criteria such as Brier score under censoring.
7,867
2111.08175
title_snapshot
Tqx7nJp7PR
MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers
https://openreview.net/forum?id=Tqx7nJp7PR
[ "Krishna Pillutla", "Swabha Swayamdipta", "Rowan Zellers", "John Thickstun", "Sean Welleck", "Yejin Choi", "Zaid Harchaoui" ]
Oral
null
As major progress is made in open-ended text generation, measuring how close machine-generated text is to human language remains a critical open problem. We introduce Mauve, a comparison measure for open-ended text generation, which directly compares the learnt distribution from a text generation model to the distribut...
[ "Open-ended generation", "neural text generation", "evaluation", "divergence frontier", "human judgement" ]
A comparison measure for open-ended text generation by directly comparing the distribution of neural machine-generated text to that of human-written text.
7,853
2102.01454
title_snapshot
iNUKmzaL-M5
Exploring Social Posterior Collapse in Variational Autoencoder for Interaction Modeling
https://openreview.net/forum?id=iNUKmzaL-M5
[ "Chen Tang", "Wei Zhan", "Masayoshi Tomizuka" ]
Poster
null
Multi-agent behavior modeling and trajectory forecasting are crucial for the safe navigation of autonomous agents in interactive scenarios. Variational Autoencoder (VAE) has been widely applied in multi-agent interaction modeling to generate diverse behavior and learn a low-dimensional representation for interacting sy...
[ "multi-agent interaction modeling", "variational autoencoder", "applications" ]
null
7,851
2112.00298
title_snapshot
VJ7u6SbqorK
MEST: Accurate and Fast Memory-Economic Sparse Training Framework on the Edge
https://openreview.net/forum?id=VJ7u6SbqorK
[ "Geng Yuan", "Xiaolong Ma", "Wei Niu", "Zhengang Li", "Zhenglun Kong", "Ning Liu", "Yifan Gong", "Zheng Zhan", "Chaoyang He", "Qing Jin", "Siyue Wang", "Minghai Qin", "Bin Ren", "Yanzhi Wang", "Sijia Liu", "Xue Lin" ]
Spotlight
null
Recently, a new trend of exploring sparsity for accelerating neural network training has emerged, embracing the paradigm of training on the edge. This paper proposes a novel Memory-Economic Sparse Training (MEST) framework targeting for accurate and fast execution on edge devices. The proposed MEST framework consists o...
[ "dynamic sparse training", "memory-economic training", "edge device training", "mobile training acceleration" ]
null
7,849
2110.14032
title_snapshot
rm0I5y2zkG8
Learning to delegate for large-scale vehicle routing
https://openreview.net/forum?id=rm0I5y2zkG8
[ "Sirui Li", "Zhongxia Yan", "Cathy Wu" ]
Spotlight
null
Vehicle routing problems (VRPs) form a class of combinatorial problems with wide practical applications. While previous heuristic or learning-based works achieve decent solutions on small problem instances, their performance deteriorates in large problems. This article presents a novel learning-augmented local search f...
[ "machine learning", "combinatorial optimization", "vehicle routing", "decomposition" ]
null
7,840
2107.04139
title_snapshot
QZpx42n0BWr
FINE Samples for Learning with Noisy Labels
https://openreview.net/forum?id=QZpx42n0BWr
[ "Taehyeon Kim", "Jongwoo Ko", "Sangwook Cho", "JinHwan Choi", "Se-Young Yun" ]
Poster
null
Modern deep neural networks (DNNs) become frail when the datasets contain noisy (incorrect) class labels. Robust techniques in the presence of noisy labels can be categorized into two folds: developing noise-robust functions or using noise-cleansing methods by detecting the noisy data. Recently, noise-cleansing methods...
[ "Learning with noisy labels", "Image Classification", "Deep Neural Networks", "Eigen Decomposition" ]
null
7,839
2102.11628
title_snapshot
YzasumDKCWV
Modified Frank Wolfe in Probability Space
https://openreview.net/forum?id=YzasumDKCWV
[ "Carson Kent", "Jiajin Li", "Jose Blanchet", "Peter Glynn" ]
Poster
null
We propose a novel Frank-Wolfe (FW) procedure for the optimization of infinite-dimensional functionals of probability measures - a task which arises naturally in a wide range of areas including statistical learning (e.g. variational inference) and artificial intelligence (e.g. generative adversarial networks). Our FW p...
[ "Frank-Wolfe", "Variational Methods", "Wasserstein Gradient Flows", "Distributionally Robust Optimization" ]
null
7,838
null
null
lS-ocNKV-u
Recurrent Submodular Welfare and Matroid Blocking Semi-Bandits
https://openreview.net/forum?id=lS-ocNKV-u
[ "Orestis Papadigenopoulos", "Constantine Caramanis" ]
Poster
null
A recent line of research focuses on the study of stochastic multi-armed bandits (MAB), in the case where temporal correlations of specific structure are imposed between the player's actions and the reward distributions of the arms. These correlations lead to (sub-)optimal solutions that exhibit interesting dynamical p...
[ "non-stationary", "multi-armed", "bandits", "blocking", "regret", "submodular", "welfare", "maximization", "approximation" ]
null
7,834
2102.00321
title_judge
owQmPJ9q9u
Controlling Neural Networks with Rule Representations
https://openreview.net/forum?id=owQmPJ9q9u
[ "Sungyong Seo", "Sercan O Arik", "Jinsung Yoon", "Xiang Zhang", "Kihyuk Sohn", "Tomas Pfister" ]
Poster
null
We propose a novel training method that integrates rules into deep learning, in a way the strengths of the rules are controllable at inference. Deep Neural Networks with Controllable Rule Representations (DeepCTRL) incorporates a rule encoder into the model coupled with a rule-based objective, enabling a shared represe...
[ "Controlling neural networks", "Incorporating rules", "Unsupervised adaptation" ]
We propose a novel training method to integrate rules into deep learning, in a way their strengths are controllable at inference.
7,829
2106.07804
title_snapshot
t9gKUW9T8fX
On Model Calibration for Long-Tailed Object Detection and Instance Segmentation
https://openreview.net/forum?id=t9gKUW9T8fX
[ "Tai-Yu Pan", "Cheng Zhang", "YANDONG LI", "Hexiang Hu", "Dong Xuan", "Soravit Changpinyo", "Boqing Gong", "Wei-Lun Chao" ]
Poster
null
Vanilla models for object detection and instance segmentation suffer from the heavy bias toward detecting frequent objects in the long-tailed setting. Existing methods address this issue mostly during training, e.g., by re-sampling or re-weighting. In this paper, we investigate a largely overlooked approach --- post-pr...
[ "Long-tailed distribution", "Object detection", "Instance segmentation", "Calibration" ]
We propose a simple and effective post-processing calibration method for long-tailed object detection and instance segmentation
7,823
2107.02170
title_snapshot
napaTaDQ0lY
Lower Bounds on Metropolized Sampling Methods for Well-Conditioned Distributions
https://openreview.net/forum?id=napaTaDQ0lY
[ "Yin Tat Lee", "Ruoqi Shen", "Kevin Tian" ]
Oral
null
We give lower bounds on the performance of two of the most popular sampling methods in practice, the Metropolis-adjusted Langevin algorithm (MALA) and multi-step Hamiltonian Monte Carlo (HMC) with a leapfrog integrator, when applied to well-conditioned distributions. Our main result is a nearly-tight lower bound of $\w...
[ "sampling", "computational statistics", "Bayesian methods", "Langevin dynamics", "Hamiltonian Monte Carlo" ]
We give lower bounds showing the current analyses of MALA for sampling well-conditioned distributions are nearly-tight, and that HMC incurs a polynomial dimension dependence for any number of steps.
7,820
2106.05480
title_snapshot
15HPeY8MGQ
Bayesian Adaptation for Covariate Shift
https://openreview.net/forum?id=15HPeY8MGQ
[ "Aurick Zhou", "Sergey Levine" ]
Poster
null
When faced with distribution shift at test time, deep neural networks often make inaccurate predictions with unreliable uncertainty estimates. While improving the robustness of neural networks is one promising approach to mitigate this issue, an appealing alternate to robustifying networks against all possible test-tim...
[ "uncertainty estimation", "calibration", "test-time adaptation", "distribution shift" ]
We present a Bayesian approach to test-time adaptation in order to improve both accuracy and calibration under distribution shift.
7,810
2109.12746
title_judge
l-0rLXvctI
Fair Sparse Regression with Clustering: An Invex Relaxation for a Combinatorial Problem
https://openreview.net/forum?id=l-0rLXvctI
[ "Adarsh Barik", "Jean Honorio" ]
Spotlight
null
In this paper, we study the problem of fair sparse regression on a biased dataset where bias depends upon a hidden binary attribute. The presence of a hidden attribute adds an extra layer of complexity to the problem by combining sparse regression and clustering with unknown binary labels. The corresponding optimizatio...
[ "Fairness", "Invexity", "Combinatorial Optimization", "Primal Dual Witness Method" ]
We provide an invex relaxation with provable theoretical guarantees for a combinatorial problem while ensuring fairness.
7,805
2102.09704
title_snapshot
1LCtHgPC-l4
Representation Learning on Spatial Networks
https://openreview.net/forum?id=1LCtHgPC-l4
[ "Zheng Zhang", "Liang Zhao" ]
Poster
null
Spatial networks are networks for which the nodes and edges are constrained by geometry and embedded in real space, which has crucial effects on their topological properties. Although tremendous success has been achieved in spatial and network representation separately in recent years, there exist very little works on ...
[ "Spatial networks", "graph neural network", "representation learning" ]
We propose a new message passing neural network model with theoretical guarantee to handle the spatial-graph topology coupled network.
7,803
null
null
jZ6FlEB78CG
Gradual Domain Adaptation without Indexed Intermediate Domains
https://openreview.net/forum?id=jZ6FlEB78CG
[ "Hong-You Chen", "Wei-Lun Chao" ]
Poster
null
The effectiveness of unsupervised domain adaptation degrades when there is a large discrepancy between the source and target domains. Gradual domain adaption (GDA) is one promising way to mitigate such an issue, by leveraging additional unlabeled data that gradually shift from the source to the target. Through sequenti...
[ "Gradual Domain Adaptation", "classification" ]
We propose a novel algorithm IDOL to bypass the need of pre-defined domain sequences in gradual domain adaptation (GDA).
7,795
2207.04587
title_snapshot
3qMwV98zLIk
FlexMatch: Boosting Semi-Supervised Learning with Curriculum Pseudo Labeling
https://openreview.net/forum?id=3qMwV98zLIk
[ "Bowen Zhang", "Yidong Wang", "Wenxin Hou", "Hao Wu", "Jindong Wang", "Manabu Okumura", "Takahiro Shinozaki" ]
Poster
null
The recently proposed FixMatch achieved state-of-the-art results on most semi-supervised learning (SSL) benchmarks. However, like other modern SSL algorithms, FixMatch uses a pre-defined constant threshold for all classes to select unlabeled data that contribute to the training, thus failing to consider different learn...
[ "semi-supervised learning", "pseudo labeling", "consistency regularization" ]
A simple and cost-free approach that can dramatically boost the performance and convergence speed of semi-supervised learning.
7,790
2110.08263
title_snapshot
QgNAUqQLh4
Structured Dropout Variational Inference for Bayesian Neural Networks
https://openreview.net/forum?id=QgNAUqQLh4
[ "Son Nguyen", "Duong Nguyen", "Khai Nguyen", "Khoat Than", "Hung Bui", "Nhat Ho" ]
Poster
null
Approximate inference in Bayesian deep networks exhibits a dilemma of how to yield high fidelity posterior approximations while maintaining computational efficiency and scalability. We tackle this challenge by introducing a novel variational structured approximation inspired by the Bayesian interpretation of Dropout re...
[ "Bayesian neural networks", "Variational Dropout", "Structured Approximate inference" ]
Improving approximate inference in Bayesian neural networks with Variational Structured Dropout
7,786
2102.07927
title_snapshot
-Z7FuZGUzv
You are caught stealing my winning lottery ticket! Making a lottery ticket claim its ownership
https://openreview.net/forum?id=-Z7FuZGUzv
[ "Xuxi Chen", "Tianlong Chen", "Zhenyu Zhang", "Zhangyang Wang" ]
Poster
null
Despite tremendous success in many application scenarios, the training and inference costs of using deep learning are also rapidly increasing over time. The lottery ticket hypothesis (LTH) emerges as a promising framework to leverage a special sparse subnetwork (i.e., $\textit{winning ticket}$) instead of a full model ...
[ "Lottery Ticket Hypothesis", "Ownership Verification" ]
A topology-based ownership verification mechanism that can prevent lottery-ticket theft under various verification schemes and attacks
7,781
2111.00162
title_snapshot
iKYO63MOWwi
Online Active Learning with Surrogate Loss Functions
https://openreview.net/forum?id=iKYO63MOWwi
[ "Giulia DeSalvo", "Claudio Gentile", "Tobias Sommer Thune" ]
Spotlight
null
We derive a novel active learning algorithm in the streaming setting for binary classification tasks. The algorithm leverages weak labels to minimize the number of label requests, and trains a model to optimize a surrogate loss on a resulting set of labeled and weak-labeled points. Our algorithm jointly admits two cr...
[ "active learning", "streaming", "weak labels" ]
We introduce a novel active learning algorithm with provable guarantees that works with surrogate loss functions and achieving compelling experimental performances
7,774
null
null
m8KpGet0Etq
Structured in Space, Randomized in Time: Leveraging Dropout in RNNs for Efficient Training
https://openreview.net/forum?id=m8KpGet0Etq
[ "Anup Sarma", "Sonali Singh", "Huaipan Jiang", "Rui Zhang", "Mahmut Kandemir", "Chita Das" ]
Poster
null
Recurrent Neural Networks (RNNs), more specifically their Long Short-Term Memory (LSTM) variants, have been widely used as a deep learning tool for tackling sequence-based learning tasks in text and speech. Training of such LSTM applications is computationally intensive due to the recurrent nature of hidden state compu...
[ "LSTM", "TRAINING", "DROPOUT", "STRUCTURED", "SPARSITY" ]
null
7,771
2106.12089
title_snapshot
BR5bZGhzrel
The Value of Information When Deciding What to Learn
https://openreview.net/forum?id=BR5bZGhzrel
[ "Dilip Arumugam", "Benjamin Van Roy" ]
Poster
null
All sequential decision-making agents explore so as to acquire knowledge about a particular target. It is often the responsibility of the agent designer to construct this target which, in rich and complex environments, constitutes a onerous burden; without full knowledge of the environment itself, a designer may forge ...
[ "Multi-armed bandits", "sequential decision-making under uncertainty", "information theory", "efficient exploration", "reinforcement learning" ]
Coupling information-theoretic principles for the design of optimal learning targets and optimal information acquisition.
7,756
2110.13973
title_snapshot
tNT4APQ0Wgj
Time-independent Generalization Bounds for SGLD in Non-convex Settings
https://openreview.net/forum?id=tNT4APQ0Wgj
[ "Tyler Farghly", "Patrick Rebeschini" ]
Poster
null
We establish generalization error bounds for stochastic gradient Langevin dynamics (SGLD) with constant learning rate under the assumptions of dissipativity and smoothness, a setting that has received increased attention in the sampling/optimization literature. Unlike existing bounds for SGLD in non-convex settings, ou...
[ "SGLD", "Langevin", "stochastic gradient", "generalization", "stability", "non-convex", "wasserstein", "optimization" ]
In the setting of non-convex learning, we derive generalization error bounds for SGLD that are time-independent and decay to zero as the sample size increases
7,751
2111.12876
title_snapshot
dTp-VUFDIB
PTR: A Benchmark for Part-based Conceptual, Relational, and Physical Reasoning
https://openreview.net/forum?id=dTp-VUFDIB
[ "Yining Hong", "Li Yi", "Joshua B. Tenenbaum", "Antonio Torralba", "Chuang Gan" ]
Poster
null
A critical aspect of human visual perception is the ability to parse visual scenes into individual objects and further into object parts, forming part-whole hierarchies. Such composite structures could induce a rich set of semantic concepts and relations, thus playing an important role in the interpretation and organiz...
[ "Part", "visual reasoning", "relational reasoning", "concept learning", "model diagnostics" ]
We introduce a new dataset for Part-based Conceptual, Relational, and Physical Reasoning
7,750
2112.05136
title_snapshot
vCWztO0ppL
A Convergence Analysis of Gradient Descent on Graph Neural Networks
https://openreview.net/forum?id=vCWztO0ppL
[ "Pranjal Awasthi", "Abhimanyu Das", "Sreenivas Gollapudi" ]
Poster
null
Graph Neural Networks~(GNNs) are a powerful class of architectures for solving learning problems on graphs. While many variants of GNNs have been proposed in the literature and have achieved strong empirical performance, their theoretical properties are less well understood. In this work we study the convergence proper...
[ "Learning Theory", "Graph Neural Networks", "Convergence Analysis" ]
We provide a convergence analysis of gradient descent for graph neural networks.
7,747
null
null
7RIYO406DB-
Dynamic influence maximization
https://openreview.net/forum?id=7RIYO406DB-
[ "Binghui Peng" ]
Poster
null
We initiate a systematic study on {\em dynamic influence maximization} (DIM). In the DIM problem, one maintains a seed set $S$ of at most $k$ nodes in a dynamically involving social network, with the goal of maximizing the expected influence spread while minimizing the amortized updating cost. We consider two evolution...
[ "Influence maximization", "submodular maximization", "dynamic algorithm" ]
Near optimal algorithm/hardness for dynamic influence maximization
7,745
2110.12602
title_snapshot
z9Xs6T0y9Eg
Improved Guarantees for Offline Stochastic Matching via new Ordered Contention Resolution Schemes
https://openreview.net/forum?id=z9Xs6T0y9Eg
[ "Brian Brubach", "Nathaniel Grammel", "Will Ma", "Aravind Srinivasan" ]
Poster
null
Matching is one of the most fundamental and broadly applicable problems across many domains. In these diverse real-world applications, there is often a degree of uncertainty in the input which has led to the study of stochastic matching models. Here, each edge in the graph has a known, independent probability of existi...
[ "Stochastic Optimization", "Combinatorial Optimization", "Discrete Optimization", "Approximation Algorithms", "Matching", "Stochastic Matching", "Prophet Inequality", "Prophet Secretary", "Contention Resolution" ]
null
7,742
2106.06892
title_snapshot
7BlQMwp_44p
ReLU Regression with Massart Noise
https://openreview.net/forum?id=7BlQMwp_44p
[ "Ilias Diakonikolas", "Jongho Park", "Christos Tzamos" ]
Poster
null
We study the fundamental problem of ReLU regression, where the goal is to fit Rectified Linear Units (ReLUs) to data. This supervised learning task is efficiently solvable in the realizable setting, but is known to be computationally hard with adversarial label noise. In this work, we focus on ReLU regression in the Ma...
[ "Robust learning" ]
This work studies ReLU regression with semi-random noise.
7,733
2109.04623
title_snapshot
jVzGglbNuW5
Learning Markov State Abstractions for Deep Reinforcement Learning
https://openreview.net/forum?id=jVzGglbNuW5
[ "Cameron Allen", "Neev Parikh", "Omer Gottesman", "George Konidaris" ]
Poster
null
A fundamental assumption of reinforcement learning in Markov decision processes (MDPs) is that the relevant decision process is, in fact, Markov. However, when MDPs have rich observations, agents typically learn by way of an abstract state representation, and such representations are not guaranteed to preserve the Mark...
[ "State Abstraction", "Representation Learning", "Markov Decision Processes", "Reinforcement Learning", "Deep RL", "Markov Property" ]
We introduce a theoretically-grounded and practical training objective for learning Markov abstract state representations that does not require pixel reconstruction, transition modeling, or reward information.
7,724
2106.04379
title_snapshot
5KWmB6JePx
End-to-End Training of Multi-Document Reader and Retriever for Open-Domain Question Answering
https://openreview.net/forum?id=5KWmB6JePx
[ "Devendra Singh Sachan", "Siva Reddy", "William L. Hamilton", "Chris Dyer", "Dani Yogatama" ]
Poster
null
We present an end-to-end differentiable training method for retrieval-augmented open-domain question answering systems that combine information from multiple retrieved documents when generating answers. We model retrieval decisions as latent variables over sets of relevant documents. Since marginalizing over sets of re...
[ "open-domain question answering", "information retrieval", "closed-book QA" ]
We propose an algorithm to train a multi-document reader and retriever for open-domain question answering in an end-to-end fashion.
7,723
2106.05346
title_snapshot
nUtLCcV24hL
Reinforcement Learning Enhanced Explainer for Graph Neural Networks
https://openreview.net/forum?id=nUtLCcV24hL
[ "Caihua Shan", "Yifei Shen", "Yao Zhang", "Xiang Li", "Dongsheng Li" ]
Poster
null
Graph neural networks (GNNs) have recently emerged as revolutionary technologies for machine learning tasks on graphs. In GNNs, the graph structure is generally incorporated with node representation via the message passing scheme, making the explanation much more challenging. Given a trained GNN model, a GNN explainer ...
[ "Graph neural networks", "Interpretability", "Reinforcement Learning" ]
We generate explanations for graph neural networks by a reinforcement learning enhanced method.
7,722
null
null
e8WWUBeafM
Bellman-consistent Pessimism for Offline Reinforcement Learning
https://openreview.net/forum?id=e8WWUBeafM
[ "Tengyang Xie", "Ching-An Cheng", "Nan Jiang", "Paul Mineiro", "Alekh Agarwal" ]
Oral
null
The use of pessimism, when reasoning about datasets lacking exhaustive exploration has recently gained prominence in offline reinforcement learning. Despite the robustness it adds to the algorithm, overly pessimistic reasoning can be equally damaging in precluding the discovery of good policies, which is an issue for t...
[ "offline reinforcement learning", "Bellman-consistent pessimism", "sample complexity bounds", "linear MDP", "function approximation" ]
We introduce the notion of Bellman-consistent pessimism for general function approximation in offline reinforcement learning.
7,718
2106.06926
title_snapshot
h7-XixPCAL
Structured Denoising Diffusion Models in Discrete State-Spaces
https://openreview.net/forum?id=h7-XixPCAL
[ "Jacob Austin", "Daniel D. Johnson", "Jonathan Ho", "Daniel Tarlow", "Rianne van den Berg" ]
Poster
null
Denoising diffusion probabilistic models (DDPMs) [Ho et al. 2021] have shown impressive results on image and waveform generation in continuous state spaces. Here, we introduce Discrete Denoising Diffusion Probabilistic Models (D3PMs), diffusion-like generative models for discrete data that generalize the multinomial di...
[ "diffusion models", "generative models", "text generation", "probabilistic models" ]
We explore diffusion models for discrete data, exploiting structured transition matrices to achieve strong results on image and text generation.
7,715
2107.03006
title_snapshot
D5APl1Yixnc
Taming Communication and Sample Complexities in Decentralized Policy Evaluation for Cooperative Multi-Agent Reinforcement Learning
https://openreview.net/forum?id=D5APl1Yixnc
[ "Xin Zhang", "Zhuqing Liu", "Jia Liu", "Zhengyuan Zhu", "Songtao Lu" ]
Poster
null
Cooperative multi-agent reinforcement learning (MARL) has received increasing attention in recent years and has found many scientific and engineering applications. However, a key challenge arising from many cooperative MARL algorithm designs (e.g., the actor-critic framework) is the policy evaluation problem, which can...
[ "Decentralized Mini-max Optimization", "Variance Reduction", "Nonlinear Function Approximation", "Policy Evaluation" ]
null
7,704
null
null
XeM4Lld0zTR
Policy Optimization in Adversarial MDPs: Improved Exploration via Dilated Bonuses
https://openreview.net/forum?id=XeM4Lld0zTR
[ "Haipeng Luo", "Chen-Yu Wei", "Chung-Wei Lee" ]
Poster
null
Policy optimization is a widely-used method in reinforcement learning. Due to its local-search nature, however, theoretical guarantees on global optimality often rely on extra assumptions on the Markov Decision Processes (MDPs) that bypass the challenge of global exploration. To eliminate the need of such assumptions, ...
[ "adversarial MDP", "policy optimization", "bandit feedback", "dilated bonuses" ]
null
7,703
2107.08346
title_snapshot
Wkq4hKpGxWv
Auditing Black-Box Prediction Models for Data Minimization Compliance
https://openreview.net/forum?id=Wkq4hKpGxWv
[ "Bashir Rastegarpanah", "Krishna P. Gummadi", "Mark Crovella" ]
Spotlight
null
In this paper, we focus on auditing black-box prediction models for compliance with the GDPR’s data minimization principle. This principle restricts prediction models to use the minimal information that is necessary for performing the task at hand. Given the challenge of the black-box setting, our key idea is to check ...
[ "Responsible ML", "Privacy", "Data Minimization" ]
This paper proposes a new operational definition of the data minimization principle (eg, from GDPR) that is appropriate for auditing black-box prediction models, and develops efficient algorithms for that setting.
7,696
null
null
8gmBGNeOfT
Observation-Free Attacks on Stochastic Bandits
https://openreview.net/forum?id=8gmBGNeOfT
[ "Yinglun Xu", "Bhuvesh Kumar", "Jacob Abernethy" ]
Poster
null
We study data corruption attacks on stochastic multi arm bandit algorithms. Existing attack methodologies assume that the attacker can observe the multi arm bandit algorithm's realized behavior which is in contrast to the adversaries modeled in the robust multi arm bandit algorithms literature. To the best of our knowl...
[ "Multi arm bandit", "Data poisoning attack" ]
null
7,694
null
null
NP-9Ppxdca
Tracking People with 3D Representations
https://openreview.net/forum?id=NP-9Ppxdca
[ "Jathushan Rajasegaran", "Georgios Pavlakos", "Angjoo Kanazawa", "Jitendra Malik" ]
Poster
null
We present a novel approach for tracking multiple people in video. Unlike past approaches which employ 2D representations, we focus on using 3D representations of people, located in three-dimensional space. To this end, we develop a method, Human Mesh and Appearance Recovery (HMAR) which in addition to extracting the 3...
[ "Tracking", "3D", "Humans", "Transformers" ]
null
7,691
2111.07868
title_snapshot
R4NeFnapYQZ
Reinforcement Learning based Disease Progression Model for Alzheimer’s Disease
https://openreview.net/forum?id=R4NeFnapYQZ
[ "Krishnakant V. Saboo", "Anirudh Choudhary", "Yurui Cao", "Gregory Worrell", "David T Jones", "Ravi Iyer" ]
Poster
null
We model Alzheimer’s disease (AD) progression by combining differential equations (DEs) and reinforcement learning (RL) with domain knowledge. DEs provide relationships between some, but not all, factors relevant to AD. We assume that the missing relationships must satisfy general criteria about the working of the bra...
[ "disease progression model", "Alzheimer's disease", "cognition trajectory prediction", "reinforcement learning", "differential equations" ]
We model Alzheimer's disease progression by combining differential equation-based models with RL using domain knowledge to (i) predict individualized disease progression 10 years into the future, and (ii) understand processes underlying the disease.
7,684
2106.16187
title_snapshot
ngdcA1tlDvj
Scallop: From Probabilistic Deductive Databases to Scalable Differentiable Reasoning
https://openreview.net/forum?id=ngdcA1tlDvj
[ "Jiani Huang", "Ziyang Li", "Binghong Chen", "Karan Samel", "Mayur Naik", "Le Song", "Xujie Si" ]
Poster
null
Deep learning and symbolic reasoning are complementary techniques for an intelligent system. However, principled combinations of these techniques have limited scalability, rendering them ill-suited for real-world applications. We propose Scallop, a system that builds upon probabilistic deductive databases, to bridge th...
[ "probabilistic database", "logic reasoning", "neural symbolic" ]
Scalable Differentiable Reasoning Using A Probabilistic Deductive Database
7,681
null
null
yMf3SLah5-y
Optimal Uniform OPE and Model-based Offline Reinforcement Learning in Time-Homogeneous, Reward-Free and Task-Agnostic Settings
https://openreview.net/forum?id=yMf3SLah5-y
[ "Ming Yin", "Yu-Xiang Wang" ]
Poster
null
This work studies the statistical limits of uniform convergence for offline policy evaluation (OPE) problems with model-based methods (for episodic MDP) and provides a unified framework towards optimal learning for several well-motivated offline tasks. Uniform OPE $\sup_\Pi|Q^\pi-\hat{Q}^\pi|<\epsilon$ is a stronger me...
[ "Theory", "Reinforcement learning theory", "Markov decision process theory" ]
null
7,677
2105.06029
title_snapshot
VzuIzbRDrum
CSDI: Conditional Score-based Diffusion Models for Probabilistic Time Series Imputation
https://openreview.net/forum?id=VzuIzbRDrum
[ "YUSUKE TASHIRO", "Jiaming Song", "Yang Song", "Stefano Ermon" ]
Poster
null
The imputation of missing values in time series has many applications in healthcare and finance. While autoregressive models are natural candidates for time series imputation, score-based diffusion models have recently outperformed existing counterparts including autoregressive models in many tasks such as image genera...
[ "time series imputation", "generative modeling", "deep learning", "score-based diffusion models" ]
We propose a novel probabilistic time series imputation method which utilize conditional score-based diffusion models.
7,676
2107.03502
title_snapshot
2OqZZAqxnn
Unbiased Classification through Bias-Contrastive and Bias-Balanced Learning
https://openreview.net/forum?id=2OqZZAqxnn
[ "Youngkyu Hong", "Eunho Yang" ]
Poster
null
Datasets for training machine learning models tend to be biased unless the data is collected with complete care. In such a biased dataset, models are susceptible to making predictions based on the biased features of the data. The biased model fails to generalize to the case where correlations between biases and targets...
[ "Debiasing", "Fairness", "Contrastive Learning" ]
Debiased classification by bias-contrastive learning and bias-balanced regression.
7,662
null
null
hqDb8d65Vfh
Efficiently Identifying Task Groupings for Multi-Task Learning
https://openreview.net/forum?id=hqDb8d65Vfh
[ "Christopher Fifty", "Ehsan Amid", "Zhe Zhao", "Tianhe Yu", "Rohan Anil", "Chelsea Finn" ]
Spotlight
null
Multi-task learning can leverage information learned by one task to benefit the training of other tasks. Despite this capacity, naively training all tasks together in one model often degrades performance, and exhaustively searching through combinations of task groupings can be prohibitively expensive. As a result, effi...
[ "multi-task learning", "task groupings", "which tasks should train together" ]
An efficient approach to determine which tasks should train together in multi-task learning networks.
7,658
2109.04617
title_snapshot
CzVPfeqPOBu
Transfer Learning of Graph Neural Networks with Ego-graph Information Maximization
https://openreview.net/forum?id=CzVPfeqPOBu
[ "Qi Zhu", "Carl Yang", "Yidan Xu", "Haonan Wang", "Chao Zhang", "Jiawei Han" ]
Poster
null
Graph neural networks (GNNs) have achieved superior performance in various applications, but training dedicated GNNs can be costly for large-scale graphs. Some recent work started to study the pre-training of GNNs. However, none of them provide theoretical insights into the design of their frameworks, or clear requirem...
[ "graph neural network", "transfer learning", "pre-training", "domain adaptation", "theoretical guarantee" ]
We propose a theoretically guaranteed and practically effective GNN model for graph transfer learning
7,656
2009.05204
title_snapshot
UMrf6F4Tg9c
Fast Abductive Learning by Similarity-based Consistency Optimization
https://openreview.net/forum?id=UMrf6F4Tg9c
[ "Yu-Xuan Huang", "Wang-Zhou Dai", "Le-Wen Cai", "Stephen Muggleton", "Yuan Jiang" ]
Poster
null
To utilize the raw inputs and symbolic knowledge simultaneously, some recent neuro-symbolic learning methods use abduction, i.e., abductive reasoning, to integrate sub-symbolic perception and logical inference. While the perception model, e.g., a neural network, outputs some facts that are inconsistent with the symboli...
[ "abductive learning", "abduction", "symbolic learning", "neuro-symbolic learning" ]
We propose a fast neuro-symbolic learning method that can perform logical abduction while considering the similarity among training instances in feature space.
7,649
null
null
fU7-so5RRhW
Clockwork Variational Autoencoders
https://openreview.net/forum?id=fU7-so5RRhW
[ "Vaibhav Saxena", "Jimmy Ba", "Danijar Hafner" ]
Poster
null
Deep learning has enabled algorithms to generate realistic images. However, accurately predicting long video sequences requires understanding long-term dependencies and remains an open challenge. While existing video prediction models succeed at generating sharp images, they tend to fail at accurately predicting far in...
[ "temporal abstraction", "variational inference", "recurrent neural networks", "deep learning" ]
null
7,636
2102.09532
title_snapshot
C0bV8xGhsz
Last-iterate Convergence in Extensive-Form Games
https://openreview.net/forum?id=C0bV8xGhsz
[ "Chung-Wei Lee", "Christian Kroer", "Haipeng Luo" ]
Poster
null
Regret-based algorithms are highly efficient at finding approximate Nash equilibria in sequential games such as poker games. However, most regret-based algorithms, including counterfactual regret minimization (CFR) and its variants, rely on iterate averaging to achieve convergence. Inspired by recent advances on last-...
[ "last-iterate convergence", "optimistic algorithms", "extensive-form games", "no-regret learning", "two-player zero-sum games", "online learning", "game dynamics" ]
null
7,634
2106.14326
title_snapshot
e5RK939Zz1S
Adversarial Robustness with Semi-Infinite Constrained Learning
https://openreview.net/forum?id=e5RK939Zz1S
[ "Alexander Robey", "Luiz F. O. Chamon", "George J. Pappas", "Hamed Hassani", "Alejandro Ribeiro" ]
Poster
null
Despite strong performance in numerous applications, the fragility of deep learning to input perturbations has raised serious questions about its use in safety-critical domains. While adversarial training can mitigate this issue in practice, state-of-the-art methods are increasingly application-dependent, heuristic in...
[ "Robust learning", "adversarial robustness", "constrained learning", "semi-infinite programming" ]
null
7,631
2110.15767
title_snapshot
S2-j0ZegyrE
Profiling Pareto Front With Multi-Objective Stein Variational Gradient Descent
https://openreview.net/forum?id=S2-j0ZegyrE
[ "Xingchao Liu", "Xin Tong", "qiang liu" ]
Spotlight
null
Finding diverse and representative Pareto solutions from the Pareto front is a key challenge in multi-objective optimization (MOO). In this work, we propose a novel gradient-based algorithm for profiling Pareto front by using Stein variational gradient descent (SVGD). We also provide a counterpart of our method based o...
[ "Pareto Front", "Sampling", "Multi-objective Learning" ]
We design Multi-objective Stein Variational Gradient Descent that can profile the whole Pareto front with particles.
7,628
null
null
MT0pTKLyzkT
Self-Supervised GANs with Label Augmentation
https://openreview.net/forum?id=MT0pTKLyzkT
[ "Liang Hou", "Huawei Shen", "Qi Cao", "Xueqi Cheng" ]
Poster
null
Recently, transformation-based self-supervised learning has been applied to generative adversarial networks (GANs) to mitigate catastrophic forgetting in the discriminator by introducing a stationary learning environment. However, the separate self-supervised tasks in existing self-supervised GANs cause a goal inconsis...
[ "generative adversarial networks", "self-supervised learning", "label augmentation" ]
We propose a novel self-supervised GAN without causing any undesired goal to the generator.
7,625
2106.08601
title_snapshot
0DBYkHfkZlk
Parameter-free HE-friendly Logistic Regression
https://openreview.net/forum?id=0DBYkHfkZlk
[ "Junyoung Byun", "Woojin Lee", "Jaewook Lee" ]
Poster
null
Privacy in machine learning has been widely recognized as an essential ethical and legal issue, because the data used for machine learning may contain sensitive information. Homomorphic encryption has recently attracted attention as a key solution to preserve privacy in machine learning applications. However, current a...
[ "homomorphic encryption", "logistic regression", "parameter-free" ]
We propose an effective privacy-preserving logistic regression method which is free from hyperparameter selection.
7,623
null
null
824xC-SgWgU
Efficient Training of Retrieval Models using Negative Cache
https://openreview.net/forum?id=824xC-SgWgU
[ "Erik Lindgren", "Sashank J. Reddi", "Ruiqi Guo", "Sanjiv Kumar" ]
Poster
null
Factorized models, such as two tower neural network models, are widely used for scoring (query, document) pairs in information retrieval tasks. These models are typically trained by optimizing the model parameters to score relevant ``positive" pairs higher than the irrelevant ``negative" ones. While a large set of nega...
[ "dual encoders", "extreme classification", "negative sampling", "recommender systems" ]
We develop a streaming negative cache for efficient training of retrieval models.
7,622
null
null
5Ya8PbvpZ9
BARTScore: Evaluating Generated Text as Text Generation
https://openreview.net/forum?id=5Ya8PbvpZ9
[ "Weizhe Yuan", "Graham Neubig", "Pengfei Liu" ]
Poster
null
A wide variety of NLP applications, such as machine translation, summarization, and dialog, involve text generation. One major challenge for these applications is how to evaluate whether such generated texts are actually fluent, accurate, or effective. In this work, we conceptualize the evaluation of generated text as ...
[ "Pre-trained models", "Text Generation", "Sequence-to-Sequence" ]
Using text generation probability to evaluate the quality of generated text.
7,605
2106.11520
title_snapshot
0qnPBmvJSaf
Monte Carlo Tree Search With Iteratively Refining State Abstractions
https://openreview.net/forum?id=0qnPBmvJSaf
[ "Samuel Sokota", "Caleb Ho", "Zaheen Farraz Ahmad", "J Zico Kolter" ]
Poster
null
Decision-time planning is the process of constructing a transient, local policy with the intent of using it to make the immediate decision. Monte Carlo tree search (MCTS), which has been leveraged to great success in Go, chess, shogi, Hex, Atari, and other settings, is perhaps the most celebrated decision-time planning...
[ "decision-time planning", "stochasticity", "reinforcement learning", "tree search" ]
An augmentation of MCTS for stochastic settings
7,599
null
null
-646c8bpgPl
Visual Adversarial Imitation Learning using Variational Models
https://openreview.net/forum?id=-646c8bpgPl
[ "Rafael Rafailov", "Tianhe Yu", "Aravind Rajeswaran", "Chelsea Finn" ]
Poster
null
Reward function specification, which requires considerable human effort and iteration, remains a major impediment for learning behaviors through deep reinforcement learning. In contrast, providing visual demonstrations of desired behaviors presents an easier and more natural way to teach agents. We consider a setting w...
[ "imitation learning", "visual imitation learning", "deep reinforcement learning" ]
We train a distribution-matching imitation-learning algorithm using variational models of image-based environments.
7,596
2107.08829
title_snapshot
_61Qh8tULj_
Conflict-Averse Gradient Descent for Multi-task learning
https://openreview.net/forum?id=_61Qh8tULj_
[ "Bo Liu", "Xingchao Liu", "Xiaojie Jin", "Peter Stone", "qiang liu" ]
Poster
null
The goal of multi-task learning is to enable more efficient learning than single task learning by sharing model structures for a diverse set of tasks. A standard multi-task learning objective is to minimize the average loss across all tasks. While straightforward, using this objective often results in much worse final ...
[ "multi-task learning" ]
we propose a new gradient manipulation method for multi-task learning that mitigates the conflicting gradient problem while still provably converging to an optimum point of the average loss.
7,590
2110.14048
title_snapshot
lHvy0DLYWm
When Is Generalizable Reinforcement Learning Tractable?
https://openreview.net/forum?id=lHvy0DLYWm
[ "Dhruv Malik", "Yuanzhi Li", "Pradeep Kumar Ravikumar" ]
Poster
null
Agents trained by reinforcement learning (RL) often fail to generalize beyond the environment they were trained in, even when presented with new scenarios that seem similar to the training environment. We study the query complexity required to train RL agents that generalize to multiple environments. Intuitively, tract...
[ "reinforcement learning", "generalization" ]
We prove lower and upper bounds on the sample complexity required by RL to generalize to multiple environments.
7,570
2101.00300
title_snapshot
NWYlZ5z8Q-R
See More for Scene: Pairwise Consistency Learning for Scene Classification
https://openreview.net/forum?id=NWYlZ5z8Q-R
[ "Gongwei Chen", "Xinhang Song", "Bohan Wang", "Shuqiang Jiang" ]
Poster
null
Scene classification is a valuable classification subtask and has its own characteristics which still needs more in-depth studies. Basically, scene characteristics are distributed over the whole image, which cause the need of “seeing” comprehensive and informative regions. Previous works mainly focus on region discover...
[ "pairwise learning", "scene classification", "convolution neural networks" ]
From the perspective of the focus area, we explore a new way to understand scene classification and propose a new learning scheme with a tailored loss in accordance with the empirical knowledge about scenes.
7,555
null
null
4CRpaV4pYp
Medical Dead-ends and Learning to Identify High-Risk States and Treatments
https://openreview.net/forum?id=4CRpaV4pYp
[ "Mehdi Fatemi", "Taylor W. Killian", "Jayakumar Subramanian", "Marzyeh Ghassemi" ]
Poster
null
Machine learning has successfully framed many sequential decision making problems as either supervised prediction, or optimal decision-making policy identification via reinforcement learning. In data-constrained offline settings, both approaches may fail as they assume fully optimal behavior or rely on exploring altern...
[ "medical deadends", "treatment security", "reinforcement learning", "offline rl", "sepsis", "healthcare" ]
We present a methodology with formal proof to learn security with respect to dead-end states, where all future treatments lead to death.
7,546
2110.04186
title_snapshot
ySFGlFjgIfN
Towards Robust Bisimulation Metric Learning
https://openreview.net/forum?id=ySFGlFjgIfN
[ "Mete Kemertas", "Tristan Ty Aumentado-Armstrong" ]
Poster
null
Learned representations in deep reinforcement learning (DRL) have to extract task-relevant information from complex observations, balancing between robustness to distraction and informativeness to the policy. Such stable and rich representations, often learned via modern function approximation techniques, can enable pr...
[ "reinforcement learning", "bisimulation", "state abstraction", "sparse rewards", "state similarity metrics", "representation learning", "continuous control" ]
We theoretically analyze on-policy bisimulation metric learning, and improve its performance in distracting and sparse reward environments.
7,537
2110.14096
title_snapshot
VvUldGZ3izR
ELLA: Exploration through Learned Language Abstraction
https://openreview.net/forum?id=VvUldGZ3izR
[ "Suvir Mirchandani", "Siddharth Karamcheti", "Dorsa Sadigh" ]
Poster
null
Building agents capable of understanding language instructions is critical to effective and robust human-AI collaboration. Recent work focuses on training these agents via reinforcement learning in environments with synthetic language; however, instructions often define long-horizon, sparse-reward tasks, and learning p...
[ "instruction following", "reward shaping", "reinforcement learning" ]
We introduce a reward shaping approach for guiding exploration in instruction following settings by correlating complex, high-level instructions with simple, low-level behaviors.
7,528
2103.05825
title_snapshot
gISH-80g05u
BlendGAN: Implicitly GAN Blending for Arbitrary Stylized Face Generation
https://openreview.net/forum?id=gISH-80g05u
[ "Mingcong Liu", "Qiang Li", "Zekui Qin", "Guoxin Zhang", "Pengfei Wan", "Wen Zheng" ]
Poster
null
Generative Adversarial Networks (GANs) have made a dramatic leap in high-fidelity image synthesis and stylized face generation. Recently, a layer-swapping mechanism has been developed to improve the stylization performance. However, this method is incapable of fitting arbitrary styles in a single model and requires hun...
[ "GAN", "image generation", "stylized face synthesis", "arbitrary style transfer" ]
We present BlendGAN for arbitrary stylized face generation, which provides both latent-guided and reference-guided synthesis to generate high-quality natural and stylized face image pairs.
7,526
2110.11728
title_snapshot
G1jmxFOtY_
Learning with User-Level Privacy
https://openreview.net/forum?id=G1jmxFOtY_
[ "Daniel Asher Nathan Levy", "Ziteng Sun", "Kareem Amin", "Satyen Kale", "Alex Kulesza", "Mehryar Mohri", "Ananda Theertha Suresh" ]
Poster
null
We propose and analyze algorithms to solve a range of learning tasks under user-level differential privacy constraints. Rather than guaranteeing only the privacy of individual samples, user-level DP protects a user's entire contribution ($m \ge 1$ samples), providing more stringent but more realistic protection against...
[ "privacy", "user-level", "optimization", "sco", "learning", "minimax", "lower bounds", "differential privacy" ]
How to solve learning tasks under a more stringent but more realistic notion of privacy where each user contributes $m\ge 1$ samples and we wish to protect information leaks at the user-level.
7,522
2102.11845
title_snapshot
hMY6nm9lld
Predicting Molecular Conformation via Dynamic Graph Score Matching
https://openreview.net/forum?id=hMY6nm9lld
[ "Shitong Luo", "Chence Shi", "Minkai Xu", "Jian Tang" ]
Poster
null
Predicting stable 3D conformations from 2D molecular graphs has been a long-standing challenge in computational chemistry. Recently, machine learning approaches have demonstrated very promising results compared to traditional experimental and physics-based simulation methods. These approaches mainly focus on modeling t...
[ "Molecular conformation generation", "score matching", "generative model" ]
we propose Dynamic Graph Score Matching (DGSM) for molecular conformation prediction, which models both the local and long-range interactions within molecules.
7,517
null
null
VMAfyuC3uXP
Sparse Deep Learning: A New Framework Immune to Local Traps and Miscalibration
https://openreview.net/forum?id=VMAfyuC3uXP
[ "Yan Sun", "Wenjun Xiong", "Faming Liang" ]
Poster
null
Deep learning has powered recent successes of artificial intelligence (AI). However, the deep neural network, as the basic model of deep learning, has suffered from issues such as local traps and miscalibration. In this paper, we provide a new framework for sparse deep learning, which has the above issues addressed in...
[ "Asymptotic Normality", "Posterior Consistency", "Prior Annealing", "Structure Selection", "Uncertainty Quantification" ]
null
7,516
2110.00653
title_snapshot
rJwDMui8DI
Neural Regression, Representational Similarity, Model Zoology & Neural Taskonomy at Scale in Rodent Visual Cortex
https://openreview.net/forum?id=rJwDMui8DI
[ "Colin Conwell", "David Mayo", "Andrei Barbu", "Michael A Buice", "George A. Alvarez", "Boris Katz" ]
Poster
null
How well do deep neural networks fare as models of mouse visual cortex? A majority of research to date suggests results far more mixed than those produced in the modeling of primate visual cortex. Here, we perform a large-scale benchmarking of dozens of deep neural network models in mouse visual cortex with both repres...
[ "neuro_ai", "deep neural networks", "rodent visual cortex", "mouse brains", "optical physiology" ]
we offer a large-scale benchmarking of hundreds of deep neural network models in mouse visual cortex using multiple methods of comparison (encoding models, RSA)
7,514
null
null
2lBhfVPYOM
(Almost) Free Incentivized Exploration from Decentralized Learning Agents
https://openreview.net/forum?id=2lBhfVPYOM
[ "Chengshuai Shi", "Haifeng Xu", "Wei Xiong", "Cong Shen" ]
Poster
null
Incentivized exploration in multi-armed bandits (MAB) has witnessed increasing interests and many progresses in recent years, where a principal offers bonuses to agents to do explorations on her behalf. However, almost all existing studies are confined to temporary myopic agents. In this work, we break this barrier and...
[ "Incentivized Exploration", "Principal Agent Models", "Multi-armed Bandits" ]
This work studied incentivized exploration in multi-armed bandits with decentralized self-interested agents, and found that when there are sufficiently many learning agents involved, the exploration process of the principal can be (almost) free.
7,509
2110.14628
title_snapshot
WL7pr00_fnJ
Sanity Checks for Lottery Tickets: Does Your Winning Ticket Really Win the Jackpot?
https://openreview.net/forum?id=WL7pr00_fnJ
[ "Xiaolong Ma", "Geng Yuan", "Xuan Shen", "Tianlong Chen", "Xuxi Chen", "Xiaohan Chen", "Ning Liu", "Minghai Qin", "Sijia Liu", "Zhangyang Wang", "Yanzhi Wang" ]
Poster
null
There have been long-standing controversies and inconsistencies over the experiment setup and criteria for identifying the "winning ticket" in literature. To reconcile such, we revisit the definition of lottery ticket hypothesis, with comprehensive and more rigorous conditions. Under our new definition, we show concret...
[ "Lottery ticket hypothesis" ]
null
7,496
2107.00166
title_snapshot
vLvsnP64VC0
Dynamic Trace Estimation
https://openreview.net/forum?id=vLvsnP64VC0
[ "Prathamesh Dharangutte", "Christopher P Musco" ]
Poster
null
We study a dynamic version of the implicit trace estimation problem. Given access to an oracle for computing matrix-vector multiplications with a dynamically changing matrix A, our goal is to maintain an accurate approximation to A's trace using as few multiplications as possible. We present a practical algorithm for s...
[ "randomized numerical linear algebra", "trace estimation", "dynamic algorithms" ]
null
7,490
2110.13752
title_snapshot
FYDE3I9fev0
Influence Patterns for Explaining Information Flow in BERT
https://openreview.net/forum?id=FYDE3I9fev0
[ "Kaiji Lu", "Zifan Wang", "Piotr Mardziel", "Anupam Datta" ]
Poster
null
While attention is all you need may be proving true, we do not know why: attention-based transformer models such as BERT are superior but how information flows from input tokens to output predictions are unclear. We introduce influence patterns, abstractions of sets of paths through a transformer model. Patterns qua...
[ "BERT", "interpretability", "attention", "information flow", "Transformers" ]
This paper introduces a new technique for explaining the information flow in BERT's computational graph using gradient-based method.
7,478
2011.00740
title_snapshot
CRFSrgYtV7m
MERLOT: Multimodal Neural Script Knowledge Models
https://openreview.net/forum?id=CRFSrgYtV7m
[ "Rowan Zellers", "Ximing Lu", "Jack Hessel", "Youngjae Yu", "Jae Sung Park", "Jize Cao", "Ali Farhadi", "Yejin Choi" ]
Oral
null
As humans, we understand events in the visual world contextually, performing multimodal reasoning across time to make inferences about the past, present, and future. We introduce MERLOT, a model that learns multimodal script knowledge by watching millions of YouTube videos with transcribed speech -- in an entirely labe...
[ "commonsense reasoning", "representation learning", "vision and language" ]
We learn multimodal, temporal representations about “how the world works” through videos.
7,477
2106.02636
title_snapshot
z-l1kpDXs88
TokenLearner: Adaptive Space-Time Tokenization for Videos
https://openreview.net/forum?id=z-l1kpDXs88
[ "Michael S Ryoo", "AJ Piergiovanni", "Anurag Arnab", "Mostafa Dehghani", "Anelia Angelova" ]
Poster
null
In this paper, we introduce a novel visual representation learning which relies on a handful of adaptively learned tokens, and which is applicable to both image and video understanding tasks. Instead of relying on hand-designed splitting strategies to obtain visual tokens and processing a large number of densely sample...
[ "visual representation", "transformers", "video understanding" ]
Instead of relying on hand-designed splitting strategies to obtain visual tokens and processing a large number of densely sampled patches for attention, TokenLearner learns to mine important tokens in visual data.
7,474
null
null
KzYIEQ_B1BX
Continuous Latent Process Flows
https://openreview.net/forum?id=KzYIEQ_B1BX
[ "Ruizhi Deng", "Marcus A Brubaker", "Greg Mori", "Andreas Lehrmann" ]
Poster
null
Partial observations of continuous time-series dynamics at arbitrary time stamps exist in many disciplines. Fitting this type of data using statistical models with continuous dynamics is not only promising at an intuitive level but also has practical benefits, including the ability to generate continuous trajectories a...
[ "continuous dynamics", "time series", "stochastic differential equation", "normalizing flow" ]
null
7,470
2106.15580
title_snapshot
kLWGdQYsmC5
Are My Deep Learning Systems Fair? An Empirical Study of Fixed-Seed Training
https://openreview.net/forum?id=kLWGdQYsmC5
[ "Shangshu Qian", "Hung Viet Pham", "Thibaud Lutellier", "Zeou Hu", "Jungwon Kim", "Lin Tan", "Yaoliang Yu", "Jiahao Chen", "Sameena Shah" ]
Poster
null
Deep learning (DL) systems have been gaining popularity in critical tasks such as credit evaluation and crime prediction. Such systems demand fairness. Recent work shows that DL software implementations introduce variance: identical DL training runs (i.e., identical network, data, configuration, software, and hardware)...
[ "DL software implementation", "fairness variance", "empirical study" ]
We perform a large-scale empirical study on DL software implementations' impact on the fairness of DL systems.
7,459
null
null
HbaQ4FEh-6
Adapting to function difficulty and growth conditions in private optimization
https://openreview.net/forum?id=HbaQ4FEh-6
[ "Hilal Asi", "Daniel Asher Nathan Levy", "John Duchi" ]
Poster
null
We develop algorithms for private stochastic convex optimization that adapt to the hardness of the specific function we wish to optimize. While previous work provide worst-case bounds for arbitrary convex functions, it is often the case that the function at hand belongs to a smaller class that enjoys faster rates. Conc...
[ "privacy", "differential privacy", "adaptivity", "optimization", "convex", "learning", "sco", "lower bounds", "minimax", "growth" ]
We provide private adaptive algorithms that adapt to the growth of the actual instance and improves upon standard worst-case rates and show that on easier functions, the extra error incurred by privacy shrinks significantly.
7,457
2108.02391
title_snapshot
djbC2A4uTHP
Consistency Regularization for Variational Auto-Encoders
https://openreview.net/forum?id=djbC2A4uTHP
[ "Samarth Sinha", "Adji Bousso Dieng" ]
Poster
null
Variational Auto-Encoders (VAEs) are a powerful approach to unsupervised learning. They enable scalable approximate posterior inference in latent-variable models using variational inference. A VAE posits a variational family parameterized by a deep neural network---called an encoder---that takes data as input. This enc...
[ "Variational Auto-Encoders" ]
A simple and general regularization technique to improve representations learned by VAEs and generalize better.
7,454
2105.14859
title_snapshot
W-agFo22-TS
TestRank: Bringing Order into Unlabeled Test Instances for Deep Learning Tasks
https://openreview.net/forum?id=W-agFo22-TS
[ "YU LI", "Min Li", "Qiuxia LAI", "Yannan Liu", "Qiang Xu" ]
Poster
null
Deep learning (DL) systems are notoriously difficult to test and debug due to the lack of correctness proof and the huge test input space to cover. Given the ubiquitous unlabeled test data and high labeling cost, in this paper, we propose a novel test prioritization technique, namely TestRank, which aims at revealing m...
[ "Deep Neural Networks", "Debugging", "Testing", "Graph Neural Network", "Test input prioritization" ]
A novel test input prioritization technique for efficient testing and debugging of deep neural network
7,453
2105.10113
title_snapshot
10anajdGZm
Provably Faster Algorithms for Bilevel Optimization
https://openreview.net/forum?id=10anajdGZm
[ "Junjie Yang", "Kaiyi Ji", "Yingbin Liang" ]
Spotlight
null
Bilevel optimization has been widely applied in many important machine learning applications such as hyperparameter optimization and meta-learning. Recently, several momentum-based algorithms have been proposed to solve bilevel optimization problems faster. However, those momentum-based algorithms do not achieve provab...
[ "Bilevel Optimization", "Momentum", "Recursive Gradient Estimator", "Hessian Vector Computation" ]
This paper proposes two bilevel optimizers that provably outperform all existing algorithms by the order of magnitude.
7,452
2106.04692
title_snapshot