Composition of feature space and state space dynamics models for model-based reinforcement learning (ニューロコンピューティング)
スポンサーリンク
概要
- 論文の詳細を見る
Learning a dynamics model and a reward model during reinforcement learning is a useful way, since the agent can also update its value function by using the models. In this paper, we propose a general dynamics model that is a composition of the feature space dynamics model and the state space dynamics model. This way enables to obtain a good generalization from a small number of samples because of the linearity of the state space dynamics, while it does not lose the accuracy. We demonstrate the simulation comparison of some dynamics models used together with a Dyna algorithm.
- 2009-07-06
著者
-
OGASAWARA Tsukasa
Graduate School of Information Science, Nara Institute of Science and Technology
-
Ogasawara Tsukasa
Graduate School Of Information Science Nara Institute Of Science And Technology
-
YAMAGUCHI Akihiko
Graduate School of Information Science, Nara Institute of Science and Technology
-
TAKAMATSU Jun
Graduate School of Information Science, Nara Institute of Science and Technology
-
Takamatsu Jun
Graduate School Of Information Science Nara Institute Of Science And Technology
-
Yamaguchi Akihiko
Graduate School Of Information Science Nara Institute Of Science And Technology
関連論文
- Composition of feature space and state space dynamics models for model-based reinforcement learning (ニューロコンピューティング)
- Composition of feature space and state space dynamics models for model-based reinforcement learning (非線形問題)
- Fusing learning strategies to learn various tasks with single configuration (ニューロコンピューティング)
- Enhanced continuous valued Q-learning for real autonomous robots
- A 1-d.o.f. Dynamic Pitching Robot That Independently Controls Velocity, Angular Velocity and Direction of a Ball