Presentation | 2003/1/24 Policy Gradient Method in Multi-Agent Systems : Pursuit Problem Seiji ISHIHARA, Harukazu IGARASHI, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | We propose a method using the policy gradient for reinforcement learning in multi-agent systems. In our approach, motion planning problems in multi-agent systems are formulated as problems that each agent selects its actions to minimize each objective function independently. The objective function can be defined by a state-value function, the sum of weight parameters of state-action rules, and heuristic potentials. The functions include some parameters. The parameters are updated stochastically in order to maximize the expectation of the reward based on a history of states and actions in each episode. The results of experiments for the pursuit problem showed that our method can make short episode plans as Q-learning does, and can easily deal with limitations such as time-window restrictions imposed on the episode length and heuristic knowledge such as an attractive potential to the target. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | reinforcement learning / policy gradient method / pursuit problem / multi-agent system |
Paper # | AI2002-58 |
Date of Issue |
Conference Information | |
Committee | AI |
---|---|
Conference Date | 2003/1/24(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Artificial Intelligence and Knowledge-Based Processing (AI) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Policy Gradient Method in Multi-Agent Systems : Pursuit Problem |
Sub Title (in English) | |
Keyword(1) | reinforcement learning |
Keyword(2) | policy gradient method |
Keyword(3) | pursuit problem |
Keyword(4) | multi-agent system |
1st Author's Name | Seiji ISHIHARA |
1st Author's Affiliation | School of Engineering,Kinki University() |
2nd Author's Name | Harukazu IGARASHI |
2nd Author's Affiliation | School of Engineering,Kinki University |
Date | 2003/1/24 |
Paper # | AI2002-58 |
Volume (vol) | vol.102 |
Number (no) | 615 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |