Presentation | 2005/3/23 Quick construction of task model by combination of local environmental model and application to multiple tasks Yu OHIGASHI, Takashi OMORI, Satoru ISHIKAWA, Koji MORIKAWA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | The traditional Reinforcement Learning(RL) supposed a complex task but single. When the traditional RL agent faced a task similar to the learned one, the agent must re-learn the task from the beginning because of its unuse of the learned result. In this paper, we supposed a set of tasks that are some what similar each other. We propose a technique of action learning that is able to quickly learn similar tasks by the reuse of previously learned knowledge. It is the model-based RL method which uses the task model constracted by combining primitive local predictors for predicting environmental dynamics. To evaluate the proposed method, we performed a computer simulation using a simple ping pong game with variation. Then we discuss applicable range and type of tasks of our method. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Model based reinforcement learning / Multiple task / Task model / Reuse of knowledge |
Paper # | NC2004-219 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2005/3/23(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 | Neurocomputing (NC) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Quick construction of task model by combination of local environmental model and application to multiple tasks |
Sub Title (in English) | |
Keyword(1) | Model based reinforcement learning |
Keyword(2) | Multiple task |
Keyword(3) | Task model |
Keyword(4) | Reuse of knowledge |
1st Author's Name | Yu OHIGASHI |
1st Author's Affiliation | Graduate School of Information Science, Hokkaido Univ.() |
2nd Author's Name | Takashi OMORI |
2nd Author's Affiliation | Graduate School of Information Science, Hokkaido Univ. |
3rd Author's Name | Satoru ISHIKAWA |
3rd Author's Affiliation | Graduate School of Information Science, Hokkaido Univ. |
4th Author's Name | Koji MORIKAWA |
4th Author's Affiliation | Advanced Technology Research Laboratories, Matsushita Electric Industrial Co., Ltd. |
Date | 2005/3/23 |
Paper # | NC2004-219 |
Volume (vol) | vol.104 |
Number (no) | 760 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |