Machine Learning Meets Game Theory

  • Modern problems involve strategic agents, private information, unknown information, and opportunities to explore and interact with agents, etc.

  • ML and Game Theory are well investigated as individual problems. Interesting research questions arise when we try to meld them together

    • Currently we are looking at two aspects:

            • Use of MAB and Q-Learning with Game Theory

            • Use of Game Theory for deep learning or deep learning for Game Theory

              • Designing optimal mechanisms for environments involving strategic agents using data-driven approaches, to overcome the difficulty of solving it analytically

              • Demystifying convergence of certain machine learning algorithms using established results from Game theory