| Presentation | 2006-02-24 A Proposal of a Reinforcement Learning Method with accumulation and use of Examples as Experience Takuma OBATA, Yousuke SASAKI, Chiaki KUBOMURA, Hiroyuki KAMEDA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Reinforcement learning is a learning frame for machines to adapt themselves to environment with no teachers. With reinforcement learning methods, agents can learn to behave adapting to environment by trial and error. Many times of trials are inevitably made, when the number of actions of agents and that of kinds of situations are very large. In real world, the number of them would be much increased. Many trials would be also made in vain, while learning procedure converges, in spite that some of them could be effectively applied to learn. Moreover, similarity among situations can be possibly used to increase waste trials. In this paper, we focused on these waste trials, i. e., negative examples, so that agents can learn to reach goals more rapidly by accumulating and using them as experience. Concretely, we propose a method to reuse negative examples in learning process, and also another method for agents to take effective action in similar situations with accumulation and use of examples as experience. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | reinfocement learning / experience / situation / similarity / agent |
| Paper # | TL2005-97,PRMU2005-232 |
| Date of Issue |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2006/2/17(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 | Pattern Recognition and Media Understanding (PRMU) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | A Proposal of a Reinforcement Learning Method with accumulation and use of Examples as Experience |
| Sub Title (in English) | |
| Keyword(1) | reinfocement learning |
| Keyword(2) | experience |
| Keyword(3) | situation |
| Keyword(4) | similarity |
| Keyword(5) | agent |
| 1st Author's Name | Takuma OBATA |
| 1st Author's Affiliation | Graduage School of Bionics, Computer and Media Sciences, Tokyo University of Technology() |
| 2nd Author's Name | Yousuke SASAKI |
| 2nd Author's Affiliation | Graduage School of Bionics, Computer and Media Sciences, Tokyo University of Technology |
| 3rd Author's Name | Chiaki KUBOMURA |
| 3rd Author's Affiliation | Aesthetics and Health Science, Yamano College of Aesthetics |
| 4th Author's Name | Hiroyuki KAMEDA |
| 4th Author's Affiliation | School of Computer Science, Tokyo University of Technology |
| Date | 2006-02-24 |
| Paper # | TL2005-97,PRMU2005-232 |
| Volume (vol) | vol.105 |
| Number (no) | 615 |
| Page | pp.pp.- |
| #Pages | 5 |
| Date of Issue | |