Focus of Attention in Reinforcement Learning
| dc.creator | Li,Lihong | |
| dc.creator | Bulitko,Vadim | |
| dc.creator | Greiner,Russell | |
| dc.date | 2007 | |
| dc.date.accessioned | 2024-02-06T12:55:44Z | |
| dc.date.available | 2024-02-06T12:55:44Z | |
| dc.description | Classification-based reinforcement learning (RL) methods have recently been pro-posed as an alternative to the traditional value-function based methods. These methods use a classifier to represent a policy, where the input (features) to the classifier is the state and theoutput (class label) for that state is the desired action. The reinforcement-learning community knows that focusing on more important states can lead to improved performance. In this paper,we investigate the idea of focused learning in the context of classification-based RL. Specifically, we define a useful notation of state importance, which we use to prove rigorous bounds on policyloss. Furthermore, we show that a classification-based RL agent may behave arbitrarily poorly if it treats all states as equally important. | |
| dc.format | text/html | |
| dc.identifier | https://doi.org/10.3217/jucs-013-09-1246 | |
| dc.identifier | https://lib.jucs.org/article/28849/ | |
| dc.identifier.uri | https://openrepository.mephi.ru/handle/123456789/9472 | |
| dc.language | en | |
| dc.publisher | Journal of Universal Computer Science | |
| dc.relation | info:eu-repo/semantics/altIdentifier/eissn/0948-6968 | |
| dc.relation | info:eu-repo/semantics/altIdentifier/pissn/0948-695X | |
| dc.rights | info:eu-repo/semantics/openAccess | |
| dc.rights | J.UCS License | |
| dc.source | JUCS - Journal of Universal Computer Science 13(9): 1246-1269 | |
| dc.subject | reinforcement learning | |
| dc.subject | function approximation | |
| dc.subject | generalization | |
| dc.subject | attention | |
| dc.title | Focus of Attention in Reinforcement Learning | |
| dc.type | Research Article |