UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2021
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917210388168704 |
|---|---|
| author | Belomestny, Denis Levin, Ilya Naumov, Alexey Samsonov, Sergey |
| author_facet | Belomestny, Denis Levin, Ilya Naumov, Alexey Samsonov, Sergey |
| contents | Policy evaluation is an important instrument for the comparison of different algorithms in Reinforcement Learning (RL). However, even a precise knowledge of the value function $V^π$ corresponding to a policy $π$ does not provide reliable information on how far the policy $π$ is from the optimal one. We present a novel model-free upper value iteration procedure ({\sf UVIP}) that allows us to estimate the suboptimality gap $V^{\star}(x) - V^π(x)$ from above and to construct confidence intervals for \(V^\star\). Our approach relies on upper bounds to the solution of the Bellman optimality equation via the martingale approach. We provide theoretical guarantees for {\sf UVIP} under general assumptions and illustrate its performance on a number of benchmark RL problems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2105_02135 |
| institution | arXiv |
| publishDate | 2021 |
| record_format | arxiv |
| spellingShingle | UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms Belomestny, Denis Levin, Ilya Naumov, Alexey Samsonov, Sergey Machine Learning Optimization and Control Policy evaluation is an important instrument for the comparison of different algorithms in Reinforcement Learning (RL). However, even a precise knowledge of the value function $V^π$ corresponding to a policy $π$ does not provide reliable information on how far the policy $π$ is from the optimal one. We present a novel model-free upper value iteration procedure ({\sf UVIP}) that allows us to estimate the suboptimality gap $V^{\star}(x) - V^π(x)$ from above and to construct confidence intervals for \(V^\star\). Our approach relies on upper bounds to the solution of the Bellman optimality equation via the martingale approach. We provide theoretical guarantees for {\sf UVIP} under general assumptions and illustrate its performance on a number of benchmark RL problems. |
| title | UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms |
| topic | Machine Learning Optimization and Control |
| url | https://arxiv.org/abs/2105.02135 |