| Presentation | 2007-12-22 The Policy Gradient On The Average Reward Manifold Takamitsu MATSUBARA, Jun MORIMOTO, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper we propose a novel policy gradient type reinforcement learning method on the average reward manifold, in which a metric to measure the effect of change in policy parameters on the average reward is introduced. In our method, the derivative of the average reward with respect to the policy improvement can be fixed as a constant. Moreover, around a (sub-) optimal policy, the policy gradient method is equivalent to the Newton method. Simple simulation results with comparison to previously proposed natural policy gradient methods demonstrate the effectiveness of our policy gradient method. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Reinforcement Learning / Policy Gradient / Natural Policy Gradient |
| Paper # | NC2007-85 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2007/12/15(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 | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | The Policy Gradient On The Average Reward Manifold |
| Sub Title (in English) | |
| Keyword(1) | Reinforcement Learning |
| Keyword(2) | Policy Gradient |
| Keyword(3) | Natural Policy Gradient |
| 1st Author's Name | Takamitsu MATSUBARA |
| 1st Author's Affiliation | Nara Institute of Science and Technology:ATR, CNS() |
| 2nd Author's Name | Jun MORIMOTO |
| 2nd Author's Affiliation | ATR, CNS:JST, ICORP, Computational Brain Project |
| Date | 2007-12-22 |
| Paper # | NC2007-85 |
| Volume (vol) | vol.107 |
| Number (no) | 410 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |