Our paper on developing a new fast two-time-scale stochastic gradient method with an optimal convergence rate, joint work with Sihan Zeng (JP Morgan AI Research), has been accepted to the Conference on Learning Theory (COLT) and is available on arXiv.
Our paper on studying the complexity of policy gradient and actor-critic methods for solving multi-task reinforcement learning, joint work with Sihan Zeng and Justin Romberg, is available on arXiv.
I'm honored to receive an NSF CAREER Award.
My newest paper on fast nonlinear two-time-scale SA can be found here, where I propose a new variant to achieve an optimal convergence rate.
We are pleased to welcome Zhenyuan Yuan as a postdoc. Zhenyuan received his Ph.D. in ECE at Pennsylvania State University in 2023.
We are honored to receive the AFOSR Young Investigator Program (YIP) award. We thank AFOSR/AFRL for supporting our research in resilient resource allocation over dynamic networks. Details can be found here.
Our paper on a two-time-scale optimization framework, joint work with Sihan Zeng and Justin Romberg, has been accepted to the SIAM Optimization.
Our paper on the connected superlevel set in (deep) reinforcement learning, joint work with Sihan Zeng and Justin Romberg, has been accepted to the NeurIPS 2023.
Our paper on Byzantine fault tolerance in federated learning for non-convex optimization, written by my student A. Dutta, has been accepted to the IEEE CDC Conference 2023.
Our paper on federated reinforcement learning for dynamic spectrum access, co-authored with Hao-Hsuan Chang, Yifei Song, and Lingjia Liu, has been accepted to the IEEE Transactions on Wireless Communications as a full paper.
Our paper on Byzantine fault tolerance in federated learning under 2-f redundancy, co-authored with Nirupam Gupta and Nitin Vaidya, has been accepted to the IEEE Transactions on Control of Network Systems as a full paper.
Our paper on gradient descent-ascent for two-player zero-sum Markov games, co-authored with Sihan Zeng and Justin Romberg, has been accepted to NeurIPS 2022.
Our paper on decentralized stochastic approximation, co-authored with Sihan Zeng and Justin Romberg, has been accepted to the IEEE Transactions on Automatic Control - Special Issue on Learning and Control.
We are pleased to receive two funding grants, REP and SFP, from Commonwealth Cyber Initiative (CCI), Southwest node. Thank you for your support.
My student, Yitao Bai, receives a fellowship from Bradley ECE Department for his PhD study. Congrats Yitao!
Our former undergraduate, Caleb Zhang, will start his master degree at University of Michigan in Fall 2022. Congrats and best of luck, Caleb.
My paper, the convergence rates of two-time-scale gradient descent-ascent dynamics for solving nonconvex min-max problems (available here), was accepted at the 4th Annual Learning for Dynamics & Control Conference.
We are excited to have Dr. Sarnaduti Brahma joining us as a postdoc in our group. Welcome aboard, Sarnaduti!
My student, Amit Dutta, is selected to have an oral presentation at Southeast Control Conference (SECC) 2021, where he will talk about our work on convergence rates of decentralized optimization over cluster networks. Congrats Amit!
Our new paper, co-authored with Sihan Zeng and Justin Romberg, on analyzing the finite-time complexity of online primal-dual natural actor-critic methods for solving constrained MDP, is available here.