Regret lower bound and optimal algorithm for high-dimensional contextual linear bandit

Research output: Contribution to journalArticlepeer-review

Abstract

In this paper, we consider the multi-armed bandit problem with high-dimensional features. First, we prove a minimax lower bound, O( (log d)α+1 2 T1−α 2 + log T), for the cumulative regret, in terms of horizon T, dimension d and a margin parameter α ∈ [0, 1], which controls the separation between the optimal and the sub-optimal arms. This new lower bound unifies existing regret bound results that have different dependencies on T due to the use of different values of margin parameter α explicitly implied by their assumptions. Second, we propose a simple and computationally efficient algorithm inspired by the general Upper Confidence Bound (UCB) strategy that achieves a regret upper bound matching the lower bound. The proposed algorithm uses a properly centered ℓ1-ball as the confidence set in contrast to the commonly used ellipsoid confidence set. In addition, the algorithm does not require any forced sampling step and is thereby adaptive to the practically unknown margin parameter. Simulations and a real data analysis are conducted to compare the proposed method with existing ones in the literature.

Original languageEnglish (US)
Pages (from-to)5652-5695
Number of pages44
JournalElectronic Journal of Statistics
Volume15
Issue number2
DOIs
StatePublished - 2021

Keywords

  • Contextual linear bandit
  • high-dimension
  • minimax regret
  • sparsity
  • upper confidence bound

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Fingerprint

Dive into the research topics of 'Regret lower bound and optimal algorithm for high-dimensional contextual linear bandit'. Together they form a unique fingerprint.

Cite this