Abstract:
AI Models have been increasingly deployed to develop “Autonomous Agents” for decision-making, with prominent application examples including playing Go and video games, robotics, autonomous driving, healthcare, human-assistant, etc. Most such success stories naturally involve multiple AI-agents interacting dynamically with each other and humans. More importantly, these agents oftentimes operate with asymmetric information in practice, both across different agents and across the training-testing phases. In this talk, we will share some of our recent explorations in understanding (multi-)AI-agents decision-making with such decentralized and asymmetric information. First, we will focus on Reinforcement Learning (RL)-Agents, in partially observable environments: we will analyze the pitfalls and efficiency of RL in partially observable Markov decision processes when there is privileged information in training, a common practice in robot learning and deep RL, and in partially observable stochastic games, when information-sharing is allowed among decentralized agents. We will show the provable benefits of privileged information and information sharing in these cases. Second, we will examine Large-Language-Model (LLM)-(powered-)Agents, which use LLM as the main controller for decision-making, by understanding and enhancing their decision-making capability in canonical decentralized and multi-agent scenarios. In particular, we use the metric of Regret, commonly studied in Online Learning and RL, to understand LLM-agents’ decision-making limits in context and in controlled experiments. Motivated by the observed pitfalls of existing LLM agents, we also proposed a new fine-tuning loss to promote the no-regret behaviors of the models, both provably and experimentally. Time permitting, we will conclude with some additional thoughts on building principled AI-Agents for decision-making with information constraints.
Bio:
Kaiqing Zhang is currently an Assistant Professor at the Department of Electrical and Computer Engineering (ECE) and the Institute for Systems Research (ISR), at the University of Maryland, College Park. He is also a member of the Center for Machine Learning, Maryland Robotics Center, and Artificial Intelligence Interdisciplinary Institute at Maryland (AIM). Prior to joining Maryland, he was a postdoctoral scholar affiliated with LIDS and CSAIL at MIT, and a Research Fellow at the Simons Institute for the Theory of Computing at Berkeley. He finished his Ph.D. from the Department of ECE at the University of Illinois at Urbana-Champaign (UIUC). He also received M.S. in both ECE and Applied Math from UIUC, and B.E. from Tsinghua University. His research interests lie in Control and Decision Theory, Game Theory, Robotics, Reinforcement/Machine Learning, Computation, and their intersections. His works have been recognized by several awards, including Simons-Berkeley Research Fellowship, CSL Thesis Award, IEEE Robotics and Automation Society TC Best-Paper Award, ICML Outstanding Paper Award, AAAI New Faculty Highlights, and NSF CAREER Award.
Orchard View Room
Kaiqing Zhang, University of Maryland