Data-Driven Upper Confidence Bounds with Near-Optimal Regret for Heavy-Tailed Bandits.
Ambrus TamásSzabolcs SzentpéteriBalázs Csanád CsájiPublished in: CoRR (2024)
Keyphrases
- heavy tailed
- regret bounds
- data driven
- reactive planning
- large deviations
- lower bound
- confidence bounds
- online learning
- linear regression
- upper bound
- heavy tails
- multi armed bandit
- generalized gaussian
- multi armed bandit problems
- control structure
- worst case
- machine learning
- prior distribution
- bregman divergences
- multiscale
- image processing