15
5

Best-of-Both-Worlds Algorithms for Linear Contextual Bandits

Abstract

We study best-of-both-worlds algorithms for KK-armed linear contextual bandits. Our algorithms deliver near-optimal regret bounds in both the adversarial and stochastic regimes, without prior knowledge about the environment. In the stochastic regime, we achieve the polylogarithmic rate (dK)2polylog(dKT)Δmin\frac{(dK)^2\mathrm{poly}\log(dKT)}{\Delta_{\min}}, where Δmin\Delta_{\min} is the minimum suboptimality gap over the dd-dimensional context space. In the adversarial regime, we obtain either the first-order O~(dKL)\widetilde{O}(dK\sqrt{L^*}) bound, or the second-order O~(dKΛ)\widetilde{O}(dK\sqrt{\Lambda^*}) bound, where LL^* is the cumulative loss of the best action and Λ\Lambda^* is a notion of the cumulative second moment for the losses incurred by the algorithm. Moreover, we develop an algorithm based on FTRL with Shannon entropy regularizer that does not require the knowledge of the inverse of the covariance matrix, and achieves a polylogarithmic regret in the stochastic regime while obtaining O~(dKT)\widetilde{O}\big(dK\sqrt{T}\big) regret bounds in the adversarial regime.

View on arXiv
Comments on this paper