Approximate difference rewards for scalable multigent reinforcement learning

• Main Authors: SINGH, Arambam James, KUMAR, Akshat
• 格式: text
• 語言: English
• 出版: Institutional Knowledge at Singapore Management University 2021
• 主題: Reinforcement learning; Multiagent systems; Databases and Information Systems
• 在線閱讀: https://ink.library.smu.edu.sg/sis_research/6901; https://ink.library.smu.edu.sg/context/sis_research/article/7904/viewcontent/Approximate_Difference_Rewards_for_Scalable_Multiagent.pdf

We address the problem of multiagent credit assignment in a large scale multiagent system. Difference rewards (DRs) are an effective tool to tackle this problem, but their exact computation is known to be challenging even for small number of agents. We propose a scalable method to compute difference...