Partition Tree Weighting for Non-Stationary Stochastic Bandits
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913708368723968 |
|---|---|
| author | Veness, Joel Hutter, Marcus Gyorgy, Andras Grau-Moya, Jordi |
| author_facet | Veness, Joel Hutter, Marcus Gyorgy, Andras Grau-Moya, Jordi |
| contents | This paper considers a generalisation of universal source coding for interaction data, namely data streams that have actions interleaved with observations. Our goal will be to construct a coding distribution that is both universal \emph{and} can be used as a control policy. Allowing for action generation needs careful treatment, as naive approaches which do not distinguish between actions and observations run into the self-delusion problem in universal settings. We showcase our perspective in the context of the challenging non-stationary stochastic Bernoulli bandit problem. Our main contribution is an efficient and high performing algorithm for this problem that generalises the Partition Tree Weighting universal source coding technique for passive prediction to the control setting. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_19325 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Partition Tree Weighting for Non-Stationary Stochastic Bandits Veness, Joel Hutter, Marcus Gyorgy, Andras Grau-Moya, Jordi Machine Learning Artificial Intelligence This paper considers a generalisation of universal source coding for interaction data, namely data streams that have actions interleaved with observations. Our goal will be to construct a coding distribution that is both universal \emph{and} can be used as a control policy. Allowing for action generation needs careful treatment, as naive approaches which do not distinguish between actions and observations run into the self-delusion problem in universal settings. We showcase our perspective in the context of the challenging non-stationary stochastic Bernoulli bandit problem. Our main contribution is an efficient and high performing algorithm for this problem that generalises the Partition Tree Weighting universal source coding technique for passive prediction to the control setting. |
| title | Partition Tree Weighting for Non-Stationary Stochastic Bandits |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2502.19325 |