Yifei Wang

Masked Auto-Regressive Variational Acceleration: Fast Inference Makes Practical Reinforcement Learning CVPR 2026 MIRA: Evaluating Multimodal AI on Complex Clinical Reasoning in Interventional Radiology AAAI 2026 Monet: Reasoning in Latent Visual Space Beyond Image and Language CVPR 2026 Nixie: Efficient, Transparent Temporal Multiplexing for Consumer GPUs OSDI 2026 R3: End-to-End Reasoning-based Planning for Multi-step Retrosynthesis via Reinforcement Learning ACL 2026 An Augmentation Overlap Theory of Contrastive Learning JMLR 2025 Beyond Interpretability: The Gains of Feature Monosemanticity on Model Robustness ICLR 2025 Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation ICML 2025 Can In-context Learning Really Generalize to Out-of-distribution Tasks? ICLR 2025 HS-STaR: Hierarchical Sampling for Self-Taught Reasoners via Difficulty Estimation and Budget Reallocation EMNLP 2025 Long-Short Alignment for Effective Long-Context Modeling in LLMs ICML 2025 Multi-Agent Collaboration via Cross-Team Orchestration ACL 2025 On the Emergence of Position Bias in Transformers ICML 2025 POSITION BIAS MITIGATES POSITION BIAS: Mitigate Position Bias Through Inter-Position Knowledge Distillation EMNLP 2025 Projection Head is Secretly an Information Bottleneck ICLR 2025 Rethinking Invariance in In-context Learning ICLR 2025 Scaling Large Language Model-based Multi-Agent Collaboration ICLR 2025 Semi-IIN: Semi-Supervised Intra-Inter Modal Interaction Learning Network for Multimodal Sentiment Analysis AAAI 2025 Uncertainty Unveiled: Can Exposure to More In-context Examples Mitigate Uncertainty for Large Language Models? ACL 2025 What is Wrong with Perplexity for Long-context Language Modeling? ICLR 2025 A Canonicalization Perspective on Invariant and Equivariant Learning NeurIPS 2024 An Expectation-Maximization Algorithm for Training Clean Diffusion Models from Corrupted Observations NeurIPS 2024 A Theoretical Understanding of Self-Correction through In-context Alignment NeurIPS 2024 Autonomous Agents for Collaborative Task under Information Asymmetry NeurIPS 2024 BadAgent: Inserting and Activating Backdoor Attacks in LLM Agents ACL 2024 Dissecting the Failure of Invariant Learning on Graphs NeurIPS 2024 Do Generated Data Always Help Contrastive Learning? ICLR 2024 Encourage or Inhibit Monosemanticity? Revisit Monosemanticity from a Feature Decorrelation Perspective EMNLP 2024 Experiential Co-Learning of Software-Developing Agents ACL 2024 Graph Neural Networks (with Proper Weights) Can Escape Oversmoothing ACML 2024 In-Context Symmetries: Self-Supervised Learning through Contextual World Models NeurIPS 2024 Look Ahead or Look Around? A Theoretical Comparison Between Autoregressive and Masked Pretraining ICML 2024 Non-negative Contrastive Learning ICLR 2024 On the Duality Between Sharpness-Aware Minimization and Adversarial Training ICML 2024 On the Role of Attention Masks and LayerNorm in Transformers NeurIPS 2024 On the Role of Discrete Tokenization in Visual Representation Learning ICLR 2024 OODRobustBench: a Benchmark and Large-Scale Analysis of Adversarial Robustness under Distribution Shift ICML 2024 Understanding the Role of Equivariance in Self-supervised Learning NeurIPS 2024 Unveiling Factual Recall Behaviors of Large Language Models through Knowledge Neurons EMNLP 2024 Adversarial Examples Are Not Real Features NeurIPS 2023 A Message Passing Perspective on Learning Dynamics of Contrastive Learning ICLR 2023 Architecture Matters: Uncovering Implicit Mechanisms in Graph Contrastive Learning NeurIPS 2023 Balance, Imbalance, and Rebalance: Understanding Robust Overfitting from a Minimax Game Perspective NeurIPS 2023 CFA: Class-Wise Calibrated Fair Adversarial Training CVPR 2023 ContraNorm: A Contrastive Learning Perspective on Oversmoothing and Beyond ICLR 2023 Contrastive Label Enhancement IJCAI 2023 Identifiable Contrastive Learning with Automatic Feature Importance Discovery NeurIPS 2023 Laplacian Canonization: A Minimalist Approach to Sign and Basis Invariant Spectral Embedding NeurIPS 2023 On the Connection between Invariant Learning and Adversarial Training for Out-of-Distribution Generalization AAAI 2023 On the Generalization of Multi-modal Contrastive Learning ICML 2023 Parallel Deep Neural Networks Have Zero Duality Gap ICLR 2023 Rethinking the Effect of Data Augmentation in Adversarial Contrastive Learning ICLR 2023 Rethinking Weak Supervision in Helping Contrastive Learning ICML 2023 Towards a Unified Theoretical Understanding of Non-contrastive Learning via Rank Differential Mechanism ICLR 2023 Unbiased Stochastic Proximal Solver for Graph Neural Networks with Equilibrium States ICLR 2023 USER: Unsupervised Structural Entropy-Based Robust Graph Neural Network AAAI 2023 A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training ICLR 2022 Beyond the Best: Distribution Functional Estimation in Infinite-Armed Bandits NeurIPS 2022 Chaos is a Ladder: A New Theoretical Understanding of Contrastive Learning via Augmentation Overlap ICLR 2022 G$^2$CN: Graph Gaussian Convolution Networks with Concentrated Graph Filters ICML 2022 How Mask Matters: Towards Theoretical Understandings of Masked Autoencoders NeurIPS 2022 Improving Out-of-Distribution Generalization by Adversarial Training with Structured Priors NeurIPS 2022 Optimization-Induced Graph Implicit Nonlinear Diffusion ICML 2022 The Convex Geometry of Backpropagation: Neural Network Gradient Flows Converge to Extreme Points of the Dual Convex Program ICLR 2022 The Hidden Convex Optimization Landscape of Regularized Two-Layer ReLU Networks: an Exact Characterization of Optimal Solutions ICLR 2022 When Adversarial Training Meets Vision Transformers: Recipes from Training to Architecture NeurIPS 2022 Adaptive Newton Sketch: Linear-time Optimization with Quadratic Convergence and Effective Hessian Dimensionality ICML 2021 Dissecting the Diffusion Process in Linear Graph Convolutional Networks NeurIPS 2021 Dynamic Context-Sensitive Filtering Network for Video Salient Object Detection ICCV 2021 On Dyadic Fairness: Exploring and Mitigating Bias in Graph Connections ICLR 2021 Residual Relaxation for Multi-view Representation Learning NeurIPS 2021 Train Once, and Decode As You Like COLING 2020