Byzantine-Resilient Reinforcement Learning for Multi-UAV Systems
Overview
Abstract
Multi-Unmanned Aerial Vehicle (multi-UAV) systems have become crucial in applications such as aerial surveillance and search and rescue missions. State-of-the-art multi-agent reinforcement learning (MARL) algorithms offer promising methods to train multi-UAV systems in both collaborative and competitive scenarios. However, the inherent vulnerabilities in multi-agent systems present substantial privacy and security challenges, particularly when deploying conventional MARL algorithms. The introduction of even a single Byzantine adversary within the system can drastically compromise the learning performance of UAV agents. In this talk, Dr. Chen will introduce a Byzantine-resilient MARL algorithm designed to mitigate, and potentially eliminate, the impact of Byzantine attacks, enhancing the reliability and robustness of multi-UAV deployments. Finally, Dr. Chen will introduce some of her ongoing research projects focused on safe and reliable AI, as well as the application of generative AI in education.
Speaker Biography
Dr. Xuhui (Tracy) Chen is an Assistant Professor in the Department of Computer Science at San Francisco State University. She earned her Ph.D. in Computer Engineering from Case Western Reserve University in 2019, her M.S. in Electrical and Computer Engineering from Mississippi State University in 2015, and her B.E. in Information Engineering from Xidian University, Xi'an, China, in 2012. Dr. Chen’s research interests focus on Artificial Intelligence and cybersecurity, particularly addressing security and privacy challenges in distributed machine learning, with applications in bioinformatics and cyber-physical systems, including UAVs, IoT, and power systems. She is also engaged in education research, exploring the use of generative AI to facilitate communication in interdisciplinary teams. Dr. Chen has authored over 25 peer-reviewed articles in leading journals and conferences, with her research supported by federal programs, industry, and university funding.