PB
Peter L. Bartlett
cs.LGstat.MLmath.STstat.THcs.AImath.OCcs.DSstat.COcs.NEcs.CL
On Valency
published · living versionsW_d9p2aje4·v1 · currentpublished
Transformers Learn Two Competing Sorting Algorithms, and Only One Length-Generalizes
with Khashayar Gatmiry, Nathan W. Henry, Chloe H. Su, Emmanuel Abbe +3
1 version
W_4q8shnze·v1published
Preprints & journals
140 papers in the corpus · 1998–2026Internal Guidance as a Sampling Method for Tilted Diffusion Models2610.29086v1 · Curtis McDonald, Peter Bartlett2026 · 0 citationsarXivon Valency
Transformers Learn Two Competing Sorting Algorithms, and Only One Length-Generalizes2610.26974v1 · Khashayar Gatmiry, Nathan W. Henry, Chloe H. Su et al.2026 · 0 citationsarXivon Valency
Improving score-based sampling via affine post-processing2610.04773v1 · Curtis McDonald, Kevin Rojas, Molei Tao et al.2026 · 0 citationsarXiv
Generalization Bounds for Flow-matching Generative Models for Intrinsically Low-dimensional Data2610.02663v2 · Saptarshi Chakraborty, Quentin Berthet, Peter L. Bartlett2026 · 0 citationsarXiv
Principal Component Regression Dominates all Monotone Spectral Filters for Linear Regression2609.39440v1 · Juno Kim, Hengyu Fu, Peter Bartlett et al.2026 · 0 citationsarXiv
Learn Your Own Thoughts: Abstract Token Curriculum2609.19717v2 · Khashayar Gatmiry, Avrajit Ghosh, Parsa Mirtaheri et al.2026 · 0 citationsarXiv
Hilbert Operator for Progressive Encoding (HOPE): A Mathematical Framework for Deconstructing Learned Representations in Deep Networks2607.21366v1 · Hossein Mobahi, Peter L. Bartlett2026 · 0 citationsarXiv
RubiConv -- Efficient Boundary-Respecting Convolutions2605.08451v1 · Linda Friso, Annie Marsden, Xinyi Chen et al.2026 · 0 citationsarXiv
Scale-free adaptive planning for deterministic dynamics & discounted rewards2604.18312v1 · Peter L. Bartlett, Victor Gabillon, Jennifer Healey et al.2026 · 2 citationsProceedings of the 36th International Conference on Machine Learning (ICML 2019)
Best of both worlds: Stochastic & adversarial best-arm identification2604.14860v2 · Yasin Abbasi-Yadkori, Peter L. Bartlett, Victor Gabillon et al.2026 · 59 citationsarXiv
Hard labels sampled from sparse targets mislead rotation invariant algorithms2603.20967v2 · Avrajit Ghosh, Bin Yu, Manfred Warmuth et al.2026 · 0 citationsICML-2026
Generalization Properties of Score-matching Diffusion Models for Intrinsically Low-dimensional Data2603.03700v2 · Saptarshi Chakraborty, Quentin Berthet, Peter L. Bartlett2026 · 0 citationsarXiv
Training Dynamics of Softmax Self-Attention: Fast Global Convergence via Preconditioning2603.01514v1 · Gautam Goel, Mahdi Soltanolkotabi, Peter Bartlett2026 · 0 citationsarXiv
A result relating convex n-widths to covering numbers with some applications to neural networks2512.04912v1 · Jonathan Baxter, Peter Bartlett2025 · 1 citationarXiv
Reinforcement Learning in POMDP's via Direct Gradient Ascent2512.02383v1 · Jonathan Baxter, Peter L. Bartlett2025 · 86 citationsProceedings of the 17th International Conference on Machine Learning (ICML 2000), 2000, pp. 41--48
Risk Comparisons in Linear Regression: Implicit Regularization Dominates Explicit Regularization2509.17251v2 · Jingfeng Wu, Peter L. Bartlett, Sham M. Kakade et al.2025 · 1 citationarXiv
Improved Scaling Laws in Linear Regression via Data Reuse2506.08415v2 · Licong Lin, Jingfeng Wu, Peter L. Bartlett2025 · 0 citationsarXiv
Large Stepsizes Accelerate Gradient Descent for Regularized Logistic Regression2506.02336v2 · Jingfeng Wu, Pierre Marion, Peter Bartlett2025 · 0 citationsarXiv
Minimax Optimal Convergence of Gradient Descent in Logistic Regression via Large and Adaptive Stepsizes2504.04105v2 · Ruiqi Zhang, Jingfeng Wu, Licong Lin et al.2025 · 0 citationsarXiv
Benign Overfitting and the Geometry of the Ridge Regression Solution in Binary Classification2503.07966v1 · Alexander Tsigler, Luiz F. O. Chamon, Spencer Frei et al.2025 · 0 citationsarXiv
Implicit Bias of Gradient Descent for Non-Homogeneous Deep Networks2502.16075v2 · Yuhang Cai, Kangjie Zhou, Jingfeng Wu et al.2025 · 0 citationsICML 2025
Benefits of Early Stopping in Gradient Descent for Overparameterized Logistic Regression2502.13283v2 · Jingfeng Wu, Peter Bartlett, Matus Telgarsky et al.2025 · 0 citationsarXiv
A Statistical Analysis for Supervised Deep Learning with Exponential Families for Intrinsically Low-dimensional Data2412.09779v1 · Saptarshi Chakraborty, Peter L. Bartlett2024 · 0 citationsarXiv
A Statistical Analysis of Deep Federated Learning for Intrinsically Low-dimensional Data2410.20659v2 · Saptarshi Chakraborty, Peter L. Bartlett2024 · 0 citationsarXiv
Fast Best-of-N Decoding via Speculative Rejection2410.20290v2 · Hanshi Sun, Momin Haider, Ruiqi Zhang et al.2024 · 1 citationarXiv
FutureFill: Fast Generation from Convolutional Sequence Models2410.03766v3 · Naman Agarwal, Xinyi Chen, Evan Dogariu et al.2024 · 0 citationsarXiv
Scaling Laws in Linear Regression: Compute, Parameters, and Data2406.08466v3 · Licong Lin, Jingfeng Wu, Sham M. Kakade et al.2024 · 2 citationsarXiv
Large Stepsize Gradient Descent for Non-Homogeneous Two-Layer Networks: Margin Improvement and Fast Optimization2406.08654v2 · Yuhang Cai, Jingfeng Wu, Song Mei et al.2024 · 1 citationarXiv
A Statistical Analysis of Wasserstein Autoencoders for Intrinsically Low-dimensional Data2402.15710v1 · Saptarshi Chakraborty, Peter L. Bartlett2024 · 0 citationsarXiv
Large Stepsize Gradient Descent for Logistic Loss: Non-Monotonicity of the Loss Improves Optimization Efficiency2402.15926v2 · Jingfeng Wu, Peter L. Bartlett, Matus Telgarsky et al.2024 · 1 citationarXiv
In-Context Learning of a Linear Transformer Block: Benefits of the MLP Component and One-Step GD Initialization2402.14951v1 · Ruiqi Zhang, Jingfeng Wu, Peter L. Bartlett2024 · 1 citationarXiv
Implicit Diffusion: Efficient Optimization through Stochastic Sampling2402.05468v3 · Pierre Marion, Anna Korba, Peter Bartlett et al.2024 · 0 citationsarXiv
On the Statistical Properties of Generative Adversarial Models for Low Intrinsic Data Dimension2401.15801v2 · Saptarshi Chakraborty, Peter L. Bartlett2024 · 1 citationarXiv
Contextual Bandits with Stage-wise Constraints2401.08016v2 · Aldo Pacchiano, Mohammad Ghavamzadeh, Peter Bartlett2024 · 0 citationsarXiv
Can a Transformer Represent a Kalman Filter?2312.06937v3 · Gautam Goel, Peter Bartlett2023 · 1 citationarXiv
How Many Pretraining Tasks Are Needed for In-Context Learning of Linear Regression?2310.08391v2 · Jingfeng Wu, Difan Zou, Zixiang Chen et al.2023 · 1 citationarXiv
Sharpness-Aware Minimization and the Edge of Stability2309.12488v6 · Philip M. Long, Peter L. Bartlett2023 · 0 citationsarXiv
Trained Transformers Learn Linear Models In-Context2306.09927v3 · Ruiqi Zhang, Spencer Frei, Peter L. Bartlett2023 · 15 citationsarXiv
Prediction, Learning, Uniform Convergence, and Scale-sensitive Dimensions2304.11059v2 · Peter L. Bartlett, Philip M. Long2023 · 48 citationsarXiv
The Double-Edged Sword of Implicit Bias: Generalization vs. Robustness in ReLU Networks2303.01456v2 · Spencer Frei, Gal Vardi, Peter L. Bartlett et al.2023 · 2 citationsarXiv
Benign Overfitting in Linear Classifiers and Leaky ReLU Networks from KKT Conditions for Margin Maximization2303.01462v1 · Spencer Frei, Gal Vardi, Peter L. Bartlett et al.2023 · 3 citationsarXiv
Kernel-based off-policy estimation without overlap: Instance optimality beyond semiparametric efficiency2301.06240v1 · Wenlong Mou, Peng Ding, Martin J. Wainwright et al.2023 · 0 citationsarXiv
Implicit Bias in Leaky ReLU Networks Trained on High-Dimensional Data2210.07082v1 · Spencer Frei, Gal Vardi, Peter L. Bartlett et al.2022 · 5 citationsarXiv
The Dynamics of Sharpness-Aware Minimization: Bouncing Across Ravines and Drifting Towards Wide Minima2210.01513v2 · Peter L. Bartlett, Philip M. Long, Olivier Bousquet2022 · 2 citationsarXiv
Off-policy estimation of linear functionals: Non-asymptotic theory for semi-parametric efficiency2209.13075v1 · Wenlong Mou, Martin J. Wainwright, Peter L. Bartlett2022 · 0 citationsarXiv
Joint Representation Training in Sequential Tasks with Shared Structure2206.12441v1 · Aldo Pacchiano, Ofir Nachum, Nilseh Tripuraneni et al.2022 · 0 citationsarXiv
Generalization Bounds for Data-Driven Numerical Linear Algebra2206.07886v1 · Peter Bartlett, Piotr Indyk, Tal Wagner2022 · 2 citationsarXiv
A Complete Characterization of Linear Estimators for Offline Policy Evaluation2203.04236v2 · Juan C. Perdomo, Akshay Krishnamurthy, Peter Bartlett et al.2022 · 0 citationsarXiv
Random Feature Amplification: Feature Learning and Generalization in Neural Networks2202.07626v4 · Spencer Frei, Niladri S. Chatterji, Peter L. Bartlett2022 · 2 citationsarXiv
Benign Overfitting without Linearity: Neural Network Classifiers Trained by Gradient Descent for Noisy Linear Data2202.05928v5 · Spencer Frei, Niladri S. Chatterji, Peter L. Bartlett2022 · 6 citationsarXiv
Career total: 472 works. 140 are in this corpus.Showing the 50 most recent.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.