A hypothetical-context estimator is claimed to make Thompson Sampling minimax optimal for linear contextual bandits under arbitrary contexts, though the proof has a critical gap.
Learning the Pareto Front Using Bootstrapped Observation Samples
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
We consider Pareto front identification (PFI) for linear bandits (PFILin), i.e., the goal is to identify a set of arms with undominated mean reward vectors when the mean reward vector is a linear function of the context. PFILin includes the best arm identification problem and multi-objective active learning as special cases. The sample complexity of our proposed algorithm is optimal up to a logarithmic factor. In addition, the regret incurred by our algorithm during the estimation is within a logarithmic factor of the optimal regret among all algorithms that identify the Pareto front. Our key contribution is a new estimator that in every round updates the estimate for the unknown parameter along multiple context directions -- in contrast to the conventional estimator that only updates the parameter estimate along the chosen context. This allows us to use low-regret arms to collect information about Pareto optimal arms. Our key innovation is to reuse the exploration samples multiple times; in contrast to conventional estimators that use each sample only once. Numerical experiments demonstrate that the proposed algorithm successfully identifies the Pareto front while controlling the regret.
fields
stat.ML 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Adaptive Data Augmentation for Thompson Sampling
A hypothetical-context estimator is claimed to make Thompson Sampling minimax optimal for linear contextual bandits under arbitrary contexts, though the proof has a critical gap.