Get our free extension to see links to code for papers anywhere online!

Add to Chrome

Add to Firefox

Picture for Jason D. Lee

Jason D. Lee

From Gradient Flow on Population Loss to Learning with Stochastic Gradient Descent

Oct 13, 2022
Satyen Kale, Jason D. Lee, Chris De Sa, Ayush Sekhari, Karthik Sridharan

Access Paper or Ask Questions

Self-Stabilization: The Implicit Bias of Gradient Descent at the Edge of Stability

Sep 30, 2022
Alex Damian, Eshaan Nichani, Jason D. Lee

Figure 1 for Self-Stabilization: The Implicit Bias of Gradient Descent at the Edge of Stability

Figure 2 for Self-Stabilization: The Implicit Bias of Gradient Descent at the Edge of Stability

Figure 3 for Self-Stabilization: The Implicit Bias of Gradient Descent at the Edge of Stability

Figure 4 for Self-Stabilization: The Implicit Bias of Gradient Descent at the Edge of Stability

* First two authors contributed equally

Access Paper or Ask Questions

PAC Reinforcement Learning for Predictive State Representations

Jul 15, 2022
Wenhao Zhan, Masatoshi Uehara, Wen Sun, Jason D. Lee

Figure 1 for PAC Reinforcement Learning for Predictive State Representations

Figure 2 for PAC Reinforcement Learning for Predictive State Representations

Figure 3 for PAC Reinforcement Learning for Predictive State Representations

Figure 4 for PAC Reinforcement Learning for Predictive State Representations

Access Paper or Ask Questions

Neural Networks can Learn Representations with Gradient Descent

Jun 30, 2022
Alex Damian, Jason D. Lee, Mahdi Soltanolkotabi

Figure 1 for Neural Networks can Learn Representations with Gradient Descent

Figure 2 for Neural Networks can Learn Representations with Gradient Descent

* COLT 2022

Access Paper or Ask Questions

Computationally Efficient PAC RL in POMDPs with Latent Determinism and Conditional Embeddings

Jun 24, 2022
Masatoshi Uehara, Ayush Sekhari, Jason D. Lee, Nathan Kallus, Wen Sun

Figure 1 for Computationally Efficient PAC RL in POMDPs with Latent Determinism and Conditional Embeddings

Figure 2 for Computationally Efficient PAC RL in POMDPs with Latent Determinism and Conditional Embeddings

Access Paper or Ask Questions

Provably Efficient Reinforcement Learning in Partially Observable Dynamical Systems

Jun 24, 2022
Masatoshi Uehara, Ayush Sekhari, Jason D. Lee, Nathan Kallus, Wen Sun

Figure 1 for Provably Efficient Reinforcement Learning in Partially Observable Dynamical Systems

Figure 2 for Provably Efficient Reinforcement Learning in Partially Observable Dynamical Systems

Access Paper or Ask Questions

Identifying good directions to escape the NTK regime and efficiently learn low-degree plus sparse polynomials

Jun 08, 2022
Eshaan Nichani, Yu Bai, Jason D. Lee

Figure 1 for Identifying good directions to escape the NTK regime and efficiently learn low-degree plus sparse polynomials

Figure 2 for Identifying good directions to escape the NTK regime and efficiently learn low-degree plus sparse polynomials

* 64 pages

Access Paper or Ask Questions

Decentralized Optimistic Hyperpolicy Mirror Descent: Provably No-Regret Learning in Markov Games

Jun 03, 2022
Wenhao Zhan, Jason D. Lee, Zhuoran Yang

Access Paper or Ask Questions

On the Effective Number of Linear Regions in Shallow Univariate ReLU Networks: Convergence Guarantees and Implicit Bias

May 18, 2022
Itay Safran, Gal Vardi, Jason D. Lee

Figure 1 for On the Effective Number of Linear Regions in Shallow Univariate ReLU Networks: Convergence Guarantees and Implicit Bias

Figure 2 for On the Effective Number of Linear Regions in Shallow Univariate ReLU Networks: Convergence Guarantees and Implicit Bias

Access Paper or Ask Questions

Nearly Minimax Algorithms for Linear Bandits with Shared Representation

Mar 29, 2022
Jiaqi Yang, Qi Lei, Jason D. Lee, Simon S. Du

Figure 1 for Nearly Minimax Algorithms for Linear Bandits with Shared Representation

Figure 2 for Nearly Minimax Algorithms for Linear Bandits with Shared Representation

Figure 3 for Nearly Minimax Algorithms for Linear Bandits with Shared Representation

* 19 pages, 3 figures

Access Paper or Ask Questions