Principles and Applications of Learning in Large-Population Games

In this talk, we discuss the design, analysis, and application of learning models in large-population games. In population games, given a set of strategies, each agent in a population selects a strategy to engage in repeated strategic interactions with others. Rather than computing and adopting the best strategy selection based on a known cost function, the agents need to learn such strategy selection from instantaneous payoffs they receive at each stage of the repeated interactions. Unlike existing formulations in game theory literature, we consider that an underlying mechanism determining the payoffs has its own dynamics.

In the first part of this talk, leveraging passivity-based analysis for feedback control systems, I explain principled approaches to design learning models for the agent strategy selection that guarantee convergence to the Nash equilibrium of an underlying game, where no agent can be better off by changing its strategy unilaterally. I also talk about the design of a higher-order learning model that strengthens the convergence which is critical when the agents' strategy selection is subject to time delays. In the second part, we discuss two applications of the large-population games framework: 1) multi-robot task allocation where a decentralized decision-making model needs to be designed for a team of mobile robots to select and carry out a given set of tasks in dynamically changing environments, and 2) payoff mechanism design to minimize the endemic transmission rate in SIRS epidemics, where the agent's strategy selection is subject to random perturbations.