ICML 2023

1,862 papers

Top Research Directions

Research directions most represented by ICML 2023 papers.

Clean Accuracy Degradation in Machine Learning Defenses 32
The Non-Linear Function Approximation Gap in Reinforcement Learning 30
Overcoming the Iterative Inference Latency of Diffusion Models 23
Bypassing the Computational Bottleneck of Adversarial Training 23
Overcoming the Homophily Assumption in Graph Neural Networks 20
The Function Approximation Gap in Reinforcement Learning Theory 20
The Weisfeiler Lehman Expressivity Bottleneck 19
Over-Smoothing in Graph Neural Networks 19
Scaling Gaussian Processes Beyond Cubic Complexity 17
Scaling Reinforcement Learning to High-Dimensional Visual Observations 16
The Mechanics of Benign Overfitting in High Dimensions 15
Exploration Strategies for Sparse Reward Reinforcement Learning 15
Conformal Prediction for Rigorous Uncertainty Quantification 14
Overcoming the Cubic Complexity of Kernel Methods 14
The Mechanics of Backdoor Attacks in Neural Networks 14

Top Authors

AuthorPapers
Quanquan Gu14
Yoshua Bengio12
Bernhard Schölkopf11
Remi Munos11
Yunhao Tang10
Bo Han9
Michal Valko9
Pieter Abbeel9
Tongliang Liu9
Zhangyang Wang9
Jun Zhu8
Sergey Levine8
Stefano Ermon8
Junchi Yan7
Percy Liang7
Tianbao Yang7
Will Dabney7
Clare Lyle6
Dacheng Tao6
Kun Zhang6

Impactful Papers

Papers from ICML 2023 that shaped research clusters.

Foundational

Influential

Papers

Title
Tensor Decompositions Meet Control Theory: Learning General Mixtures of Linear Dynamical Systems
Tensor Gaussian Process with Contraction for Multi-Channel Imaging Analysis
Test-time Adaptation with Slot-Centric Models
Test-Time Style Shifting: Handling Arbitrary Styles in Domain Generalization
Text Generation with Diffusion Language Models: A Pre-training Approach with Continuous Paragraph Denoise
Text-To-4D Dynamic Scene Generation
Text-To-Concept (and Back) via Cross-Model Alignment
TGRL: An Algorithm for Teacher Guided Reinforcement Learning
The Acquisition of Physical Knowledge in Generative Neural Networks
The Benefits of Mixup for Feature Learning
The Benefits of Model-Based Generalization in Reinforcement Learning
The Blessing of Heterogeneity in Federated Q-Learning: Linear Speedup and Beyond
The case for 4-bit precision: k-bit Inference Scaling Laws
The Catalog Problem: Clustering and Ordering Variable-Sized Sets
The Computational Complexity of Concise Hypersphere Classification
The Dormant Neuron Phenomenon in Deep Reinforcement Learning
The Edge of Orthogonality: A Simple View of What Makes BYOL Tick
The Fast Johnson-Lindenstrauss Transform Is Even Faster
The Flan Collection: Designing Data and Methods for Effective Instruction Tuning
The Hessian perspective into the Nature of Convolutional Neural Networks
The Ideal Continual Learner: An Agent That Never Forgets
The Impact of Exploration on Convergence and Performance of Multi-Agent Q-Learning Dynamics
The Implicit Regularization of Dynamical Stability in Stochastic Gradient Descent
The Monge Gap: A Regularizer to Learn All Transport Maps
The Numerical Stability of Hyperbolic Representation Learning
The Optimal Approximation Factors in Misspecified Off-Policy Value Function Estimation
Theoretical Behavior of XAI Methods in the Presence of Suppressor Variables
Theoretical Bounds on the Network Community Profile from Low-rank Semi-definite Programming
Theoretical Guarantees of Learning Ensembling Strategies with Applications to Time Series Forecasting
Theory on Forgetting and Generalization of Continual Learning
The Persistent Laplacian for Data Science: Evaluating Higher-Order Persistent Spectral Representations of Data
The Power of Learned Locally Linear Models for Nonlinear Policy Optimization
The Power of Preconditioning in Overparameterized Low-Rank Matrix Sensing
The Power of Uniform Sampling for k-Median
The Price of Differential Privacy under Continual Observation
The Regret of Exploration and the Control of Bad Episodes in Reinforcement Learning
The Role of Entropy and Reconstruction in Multi-View Self-Supervised Learning
The Saddle-Point Method in Differential Privacy
The SSL Interplay: Augmentations, Inductive Bias, and Generalization
The Statistical Benefits of Quantile Temporal-Difference Learning for Value Estimation
The Statistical Scope of Multicalibration
The Test of Tests: A Framework for Differentially Private Hypothesis Testing
The Unintended Consequences of Discount Regularization: Improving Regularization in Certainty Equivalence Reinforcement Learning
The Unreasonable Effectiveness of Few-shot Learning for Machine Translation
The Value of Out-of-Distribution Data
The Virtues of Laziness in Model-based RL: A Unified Objective and Algorithms
The Wisdom of Hindsight Makes Language Models Better Instruction Followers
Thompson Sampling for High-Dimensional Sparse Linear Contextual Bandits
Thompson Sampling with Diffusion Generative Prior
Thompson Sampling with Less Exploration is Fast and Optimal

« prev Page 34/38 next »