Profiles2021-2025
Song Mei
UC Berkeley
19Papers
8Last author
19Unique co-authors
1First author
1Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers19
- Generalization or Hallucination? Understanding Out-of-Context Reasoning in TransformersAuthor 9 of 9NeurIPS 2025 · 10 citations
- A Statistical Theory of Contrastive Learning via Approximate Sufficient StatisticsAuthor 3 of 3NeurIPS 2025 · 4 citations
- U-Nets as Belief Propagation: Efficient Classification, Denoising, and Diffusion in Generative Hierarchical ModelsAuthor 1 of 1ICLR 2025
- How Do Transformers Learn In-Context Beyond Simple Functions? A Case Study on Learning with RepresentationsAuthor 3 of 7ICLR 2024 · 80 citations
- Transformers as Decision Makers: Provable In-Context Reinforcement Learning via Supervised PretrainingAuthor 3 of 3ICLR 2024 · 74 citations
- Large Stepsize Gradient Descent for Non-Homogeneous Two-Layer Networks: Margin Improvement and Fast OptimizationAuthor 3 of 5NeurIPS 2024 · 20 citations
- Statistical Estimation in the Spiked Tensor Model via the Quantum Approximate Optimization AlgorithmAuthor 3 of 3NeurIPS 2024 · 7 citations
- Transformers as Statisticians: Provable In-Context Learning with In-Context Algorithm SelectionAuthor 5 of 5NeurIPS 2023 · 356 citations
- What can a Single Attention Layer Learn? A Study Through the Random Features LensAuthor 4 of 4NeurIPS 2023 · 47 citations
- Lower Bounds for Learning in Revealing POMDPsAuthor 4 of 5ICML 2023 · 18 citations
- Partially Observable RL with B-Stability: Unified Structural Condition and Sharp Sample-Efficient AlgorithmsAuthor 3 of 3ICLR 2023 · 2 citations
- When Can We Learn General-Sum Markov Games with a Large Number of Players Sample-Efficiently?Author 2 of 3ICLR 2022 · 83 citations
- Near-Optimal Learning of Extensive-Form Games with Imperfect InformationAuthor 3 of 4ICML 2022 · 31 citations
- Efficient and Differentiable Conformal Prediction with General Function ClassesAuthor 2 of 5ICLR 2022 · 29 citations
- Efficient Phi-Regret Minimization in Extensive-Form Games via Online Mirror DescentAuthor 3 of 5NeurIPS 2022 · 24 citations
- Sample-Efficient Learning of Correlated Equilibria in Extensive-Form GamesAuthor 2 of 3NeurIPS 2022 · 11 citations
- Don't Just Blame Over-parametrization for Over-confidence: Theoretical Analysis of Calibration in Binary ClassificationAuthor 2 of 4ICML 2021 · 47 citations
- Exact Gap between Generalization Error and Uniform Convergence in Random Feature ModelsAuthor 3 of 3ICML 2021 · 19 citations
- Understanding the Under-Coverage Bias in Uncertainty EstimationAuthor 2 of 4NeurIPS 2021 · 18 citations
