Multi-agent Reinforcement Learning for Decentralized Stable Matching

In the real world, people/entities usually find matches independently and autonomously, such as finding jobs, partners, roommates, etc. It is possible that this search for starts with no initial knowledge of environment. We propose use a multi-agent reinforcement learning (MARL) paradigm spatially formulated decentralized two-sided matching market independent autonomous agents. Having agents acting makes our environment very dynamic uncertain. Moreover, lack preferences other have to explore interact discover their own through noisy rewards. think setting better approximates world we study usefulness MARL approach it. Along conventional stable case where strictly ordered preferences, check applicability incomplete lists ties. investigate results stability, level instability (for unstable results), fairness. Our mostly yields fair outcomes.