Do Agents Dream of Electric Sheep?: Improving Generalization in Reinforcement Learning through Generative Learning
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929273902727168 |
|---|---|
| author | Franceschelli, Giorgio Musolesi, Mirco |
| author_facet | Franceschelli, Giorgio Musolesi, Mirco |
| contents | The Overfitted Brain hypothesis suggests dreams happen to allow generalization in the human brain. Here, we ask if the same is true for reinforcement learning agents as well. Given limited experience in a real environment, we use imagination-based reinforcement learning to train a policy on dream-like episodes, where non-imaginative, predicted trajectories are modified through generative augmentations. Experiments on four ProcGen environments show that, compared to classic imagination and offline training on collected experience, our method can reach a higher level of generalization when dealing with sparsely rewarded environments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_07979 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Do Agents Dream of Electric Sheep?: Improving Generalization in Reinforcement Learning through Generative Learning Franceschelli, Giorgio Musolesi, Mirco Machine Learning Artificial Intelligence The Overfitted Brain hypothesis suggests dreams happen to allow generalization in the human brain. Here, we ask if the same is true for reinforcement learning agents as well. Given limited experience in a real environment, we use imagination-based reinforcement learning to train a policy on dream-like episodes, where non-imaginative, predicted trajectories are modified through generative augmentations. Experiments on four ProcGen environments show that, compared to classic imagination and offline training on collected experience, our method can reach a higher level of generalization when dealing with sparsely rewarded environments. |
| title | Do Agents Dream of Electric Sheep?: Improving Generalization in Reinforcement Learning through Generative Learning |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2403.07979 |