On March 4, 2025, the 73rd Zhi Xin Forum was held in Lecture Hall 117. Jointly organized by the College of Electronic and Information Engineering and the Shanghai Institute of Intelligent Science and Technology of Tongji University, Prof. Lacra Pavel from the University of Toronto was invited to give a lecture on“On System Theory for Learning in Games”.
First, Prof. Lacra Pavel introduced the key role of systems theory in the analysis and design of learning algorithms in games. She reviewed numerous algorithms/dynamics that have been proposed, including best-response play, (projected) gradient-play and proximal dynamics to fictitious-play, payoff-based play or Q-learning (reinforcement-learning). She pointed out that why certain algorithms work and others do not in certain game settings and how to relax the assumptions of these algorithms and generalize them in a systematic way is a topic that has received a lot of attention in recent years. In the presentation, Prof. Lacra Pavel elaborated on their team's contribution to this field. The team's approach is based on exploiting systems-theoretic principles and the connection to passivity/dissipation, showing how some popular game-theoretic algorithms can be viewed as feedback interconnections between dissipative/passive dynamical systems and certain game mappings. Once this is achieved, the convergence analysis of learning dynamics can be accomplished by concise arguments based on standard passivity theory. In addition, Prof. Lacra Pavel discussed how the ideas inspired by passivity can be used to design new algorithms and learning dynamics for solving Nash equilibrium and generalized Nash equilibrium problems. Finally, Prof. Lacra Pavel further introduced higher-order learning dynamics based on passivity and explores learning extensions for intelligences with intrinsic dynamic properties.
After the report, Prof. Lacra Pavel had a cordial exchange and discussion with the teachers and students, and she also encouraged them to actively broaden their horizons, explore, discover and solve new scientific problems based on her own experience. This report further expanded the vision of our teachers and students and enhanced their understanding and knowledge of the application of system theory in game learning.
