2020/06/20 by Kiyeob Lee, Lee, Kiyeob, Desik Rengarajan +5
Business, Management and Accounting · Decision Sciences · Social Sciences · #Advanced Bandit Algorithms Research #Computer Science and Game Theory (cs.GT) #Consumer Market Behavior and Pricing #Experimental Behavioral Economics Studies #FOS: Computer and information sciences #FOS: Mathematics #Game Theory and Applications #Machine Learning (cs.LG) #Multiagent Systems (cs.MA) #Optimization and Control (math.OC)
paper · pdf · doi:10.48550/arxiv.2006.11683
openalex publication_date 2020/06/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Mean Field Games (MFG) are the class of games with a very large number of\nagents and the standard equilibrium concept is a Mean Field Equilibrium (MFE).\nAlgorithms for learning MFE in dynamic MFGs are unknown in general. Our focus\nis on an important subclass that possess a monotonicity property called\nStrategic Complementarities (MFG-SC). We introduce a natural refinement to the\nequilibrium concept that we call Trembling-Hand-Perfect MFE (T-MFE), which\nallows agents to employ a measure of randomization while accounting for the\nimpact of such randomization on their payoffs. We propose a simple algorithm\nfor computing T-MFE under a known model. We also introduce a model-free and a\nmodel-based approach to learning T-MFE and provide sample complexities of both\nalgorithms. We also develop a fully online learning scheme that obviates the\nneed for a simulator. Finally, we empirically evaluate the performance of the\nproposed algorithms via examples motivated by real-world applications.\n