Prof. Quanyan Zhu, an assistant professor in the Department of Electrical and Computer Engineering at New York University, visited CDSL and delivered a seminar titled ‘The Role of Information in Dynamic Games and Multi-Agent Learning’.
In his seminar, Prof. Zhu discussed the significant role that information structures play in dynamic games and multi-agent systems. He highlighted how, counter to intuition, more information does not always lead to improved social utility, and how this paradox challenges traditional frameworks.
Prof. Zhu introduced conjectural learning, an adaptive framework that allows agents to form and refine subjective conjectures about their environment and unknown elements. He demonstrated the effectiveness of this approach in addressing the complexities of incomplete information. The seminar also explored the applications of this framework in various domains, including misinformation dynamics in social networks, cyber deception strategies, and robotics. Furthermore, Prof. Zhu discussed the emerging role of large language models (LLMs) in enhancing multi-agent learning, providing new tools to improve decision-making and strategies in these challenging environments.
The seminar offered valuable insights into the evolving challenges in dynamic games and multi-agent learning. CDSL researchers and students found the discussion particularly engaging and thought-provoking.
We thank Prof. Zhu for his insightful presentation and look forward to future collaborations in this exciting area of research.