RMIX: Learning risk-sensitive policies for cooperative reinforcement learning agents

• Main Authors: QIU, Wei, WANG, Xinrun, YU, Runsheng, HE, Xu, WANG, Rundong, AN, Bo, OBRAZTSOVA, Svetlana, RABINOVICH, Zinovi
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2021
• Subjects: Artificial Intelligence and Robotics; Theory and Algorithms
• Online Access: https://ink.library.smu.edu.sg/sis_research/9137; https://ink.library.smu.edu.sg/context/sis_research/article/10140/viewcontent/NeurIPS_2021_rmix__pvoa.pdf
• Institution: Singapore Management University

Current value-based multi-agent reinforcement learning methods optimize individual Q values to guide individuals' behaviours via centralized training with decentralized execution (CTDE). However, such expected, i.e., risk-neutral, Q value is not sufficient even with CTDE due to the randomness o...