Ergodicity in reinforcement learning
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915854323548160 |
|---|---|
| author | Baumann, Dominik Noorani, Erfaun Mustafin, Arsenii Sheng, Xinyi Verbruggen, Bert Vanhoyweghen, Arne Ginis, Vincent Schön, Thomas B. |
| author_facet | Baumann, Dominik Noorani, Erfaun Mustafin, Arsenii Sheng, Xinyi Verbruggen, Bert Vanhoyweghen, Arne Ginis, Vincent Schön, Thomas B. |
| contents | In reinforcement learning, we typically aim to optimize the expected value of the sum of rewards an agent collects over a trajectory. However, if the process generating these rewards is non-ergodic, the expected value, i.e., the average over infinitely many trajectories with a given policy, is uninformative for the average over a single, but infinitely long trajectory. Thus, if we care about how the individual agent performs during deployment, the expected value is not a good optimization objective. In this paper, we discuss the impact of non-ergodic reward processes on reinforcement learning agents through an instructive example, relate the notion of ergodic reward processes to more widely used notions of ergodic Markov chains, and present existing solutions that optimize long-term performance of individual trajectories under non-ergodic reward dynamics. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_10895 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Ergodicity in reinforcement learning Baumann, Dominik Noorani, Erfaun Mustafin, Arsenii Sheng, Xinyi Verbruggen, Bert Vanhoyweghen, Arne Ginis, Vincent Schön, Thomas B. Machine Learning In reinforcement learning, we typically aim to optimize the expected value of the sum of rewards an agent collects over a trajectory. However, if the process generating these rewards is non-ergodic, the expected value, i.e., the average over infinitely many trajectories with a given policy, is uninformative for the average over a single, but infinitely long trajectory. Thus, if we care about how the individual agent performs during deployment, the expected value is not a good optimization objective. In this paper, we discuss the impact of non-ergodic reward processes on reinforcement learning agents through an instructive example, relate the notion of ergodic reward processes to more widely used notions of ergodic Markov chains, and present existing solutions that optimize long-term performance of individual trajectories under non-ergodic reward dynamics. |
| title | Ergodicity in reinforcement learning |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2603.10895 |