A direct fully adaptive strategy for contextual bandits with latent HMM states achieves high-probability regret bounds independent of reward functions and depending only on online HMM estimation.
up to some permutation ρ of [H]
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Direct Approach for Handling Contextual Bandits with Latent State Dynamics
A direct fully adaptive strategy for contextual bandits with latent HMM states achieves high-probability regret bounds independent of reward functions and depending only on online HMM estimation.