New ECE/ISR assistant professor wins award for dissertation

news story image

ISR and ECE’s newest joint appointment faculty member, Kaiqing Zhang, has received the University of Illinois at Urbana-Champaign’s (UIUC) Coordinated Science Laboratory Ph.D. Thesis Award for his 2021 dissertation. In a UIUC awards ceremony on Oct. 12, Zhang spoke about this thesis research, “Reinforcement Learning for Multi-Agent and Robust Control Systems: Towards Large-scale and Reliable Autonomy.”

Recent years have witnessed tremendous successes of AI and machine learning, especially reinforcement learning (RL), in solving many decision-making and control tasks. However, many RL algorithms are still miles away from being applied to practical autonomous systems, which usually involve more complicated scenarios with model uncertainty and multiple decision-makers by nature. 

Zhang’s work studied RL for control and sequential decision-making with provable guarantees, especially with robustness and multi-agent interaction considerations. He showed that policy optimization, one of the main drivers of many empirical successes of RL, can solve a fundamental class of robust control tasks with global optimality guarantees, despite nonconvexity. Importantly, he also showed that certain policy optimization approaches automatically preserve some “robustness” during learning, a property he calls “implicit regularization.” This interesting phenomenon has been observed in other different machine learning contexts.

Before coming to Maryland, Zhang was a postdoctoral scholar affiliated with the LIDS Lab and the CSAIL robotics lab at the Massachusetts Institute of Technology. At MIT, he worked with Professors Asu Ozdaglar, Russ Tedrake, and Constantinos Daskalakis.

Zhang received his Ph.D. in Electrical and Computer Engineering in 2021 from the Department of Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign. He was advised by Professor Tamer Başar. Prior to his Ph.D., Zhang  received two M.S. degrees from UIUC in 2017, one in Electrical and Computer Engineering and the other in Applied Math. Zhang holds a B.E. in Automation and a B.S. in Economics from Tsinghua University in China.

Zhang's research interests lie in the intersection of control theory, game theory, and machine/reinforcement learning, especially in multi-agent and safety-critical systems; with applications in intelligent and distributed cyber-physical systems, e.g., robotics, smart grid, and transportation systems. He uses mathematical tools from control theory, game theory, operations research, and probability theory to develop provably convergent and efficient algorithms. His broad goal is to lay theoretical foundations for  learning algorithms and systems that address (data-driven) sequential-decision-making problems in game theory and control theory, particularly in the presence of multiple decision-makers, towards large-scale and reliable autonomy.


Published October 13, 2022