Thompson Sampling for High-Dimensional Sparse Linear Contextual Bandits
Machine Learning
2023-01-31 v2 Machine Learning
Statistics Theory
Methodology
Statistics Theory
Abstract
We consider the stochastic linear contextual bandit problem with high-dimensional features. We analyze the Thompson sampling algorithm using special classes of sparsity-inducing priors (e.g., spike-and-slab) to model the unknown parameter and provide a nearly optimal upper bound on the expected cumulative regret. To the best of our knowledge, this is the first work that provides theoretical guarantees of Thompson sampling in high-dimensional and sparse contextual bandits. For faster computation, we use variational inference instead of Markov Chain Monte Carlo (MCMC) to approximate the posterior distribution. Extensive simulations demonstrate the improved performance of our proposed algorithm over existing ones.
Cite
@article{arxiv.2211.05964,
title = {Thompson Sampling for High-Dimensional Sparse Linear Contextual Bandits},
author = {Sunrit Chakraborty and Saptarshi Roy and Ambuj Tewari},
journal= {arXiv preprint arXiv:2211.05964},
year = {2023}
}
Comments
38 pages, 4 figures