Reinforcement Learning with Partial Parametric Model Knowledge
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2023
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917622276161536 |
|---|---|
| author | Wang, Shuyuan Loewen, Philip D. Lawrence, Nathan P. Forbes, Michael G. Gopaluni, R. Bhushan |
| author_facet | Wang, Shuyuan Loewen, Philip D. Lawrence, Nathan P. Forbes, Michael G. Gopaluni, R. Bhushan |
| contents | We adapt reinforcement learning (RL) methods for continuous control to bridge the gap between complete ignorance and perfect knowledge of the environment. Our method, Partial Knowledge Least Squares Policy Iteration (PLSPI), takes inspiration from both model-free RL and model-based control. It uses incomplete information from a partial model and retains RL's data-driven adaption towards optimal performance. The linear quadratic regulator provides a case study; numerical experiments demonstrate the effectiveness and resulting benefits of the proposed method. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2304_13223 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Reinforcement Learning with Partial Parametric Model Knowledge Wang, Shuyuan Loewen, Philip D. Lawrence, Nathan P. Forbes, Michael G. Gopaluni, R. Bhushan Systems and Control Artificial Intelligence Machine Learning We adapt reinforcement learning (RL) methods for continuous control to bridge the gap between complete ignorance and perfect knowledge of the environment. Our method, Partial Knowledge Least Squares Policy Iteration (PLSPI), takes inspiration from both model-free RL and model-based control. It uses incomplete information from a partial model and retains RL's data-driven adaption towards optimal performance. The linear quadratic regulator provides a case study; numerical experiments demonstrate the effectiveness and resulting benefits of the proposed method. |
| title | Reinforcement Learning with Partial Parametric Model Knowledge |
| topic | Systems and Control Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2304.13223 |