Profiles2020-2025
Peter L Bartlett
University of California, Berkeley
14Papers
4Last author
24Unique co-authors
0First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers14
- Large Stepsizes Accelerate Gradient Descent for Regularized Logistic RegressionAuthor 4 of 5NeurIPS 2025 · 12 citations
- Improved Scaling Laws in Linear Regression via Data ReuseAuthor 4 of 5NeurIPS 2025 · 7 citations
- Benefits of Early Stopping in Gradient Descent for Overparameterized Logistic RegressionAuthor 2 of 4ICML 2025
- Implicit Bias of Gradient Descent for Non-Homogeneous Deep NetworksAuthor 6 of 6ICML 2025
- Fast Best-of-N Decoding via Speculative RejectionAuthor 8 of 9NeurIPS 2024 · 144 citations
- How Many Pretraining Tasks Are Needed for In-Context Learning of Linear Regression?Author 7 of 8ICLR 2024 · 94 citations
- Scaling Laws in Linear Regression: Compute, Parameters, and DataAuthor 4 of 7NeurIPS 2024 · 57 citations
- In-Context Learning of a Linear Transformer Block: Benefits of the MLP Component and One-Step GD InitializationAuthor 3 of 3NeurIPS 2024 · 37 citations
- Large Stepsize Gradient Descent for Non-Homogeneous Two-Layer Networks: Margin Improvement and Fast OptimizationAuthor 5 of 5NeurIPS 2024 · 20 citations
- A Statistical Analysis of Wasserstein Autoencoders for Intrinsically Low-dimensional DataAuthor 2 of 2ICLR 2024 · 3 citations
- The Double-Edged Sword of Implicit Bias: Generalization vs. Robustness in ReLU NetworksAuthor 3 of 6NeurIPS 2023 · 25 citations
- Implicit Bias in Leaky ReLU Networks Trained on High-Dimensional DataAuthor 4 of 8ICLR 2023 · 5 citations
- On the Theory of Reinforcement Learning with Once-per-Episode FeedbackAuthor 3 of 6NeurIPS 2021 · 37 citations
- On Approximate Thompson Sampling with Langevin AlgorithmsAuthor 4 of 5ICML 2020 · 34 citations
