A Study on Credit Assignment among Reinforcement Learning Agents

Please enable javascript in your browser.

Rahaie, Zahra | 2010

520 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 40309 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Beigy, Hamid
Abstract:
Nowadays, multi-agent systems as part of the distributed artificial intelligence play an important role in modeling and solving complex industrial and commercial problems. They have distinguishing characteristics such as distributiveness (spatial, temporal, semantic, or functional distribution), robustness, parallel processing, etc. One of the capabilities that can be added to this system is the learning capability. It can help the system to adapt itself to the new environment. This paper proposed a method for the problem of credit assignment in multi-agent domain. Solving the multi-agent credit assignment problem, one can expect individual learning for a single agent in systems of more-than-one agent and in the presence of interaction among coopering agents. Proposed method tries to investigate the behavior of the environment when encountered with different actions on behalf of the agents. Reported results show the satisfactory performance of the method in a multi-agent single-step domain.
Keywords:
Multiagent System ; Reinforcement Learning ; Credit Assignment

No TOC