24
51

Optimal Order Simple Regret for Gaussian Process Bandits

Abstract

Consider the sequential optimization of a continuous, possibly non-convex, and expensive to evaluate objective function ff. The problem can be cast as a Gaussian Process (GP) bandit where ff lives in a reproducing kernel Hilbert space (RKHS). The state of the art analysis of several learning algorithms shows a significant gap between the lower and upper bounds on the simple regret performance. When NN is the number of exploration trials and γN\gamma_N is the maximal information gain, we prove an O~(γN/N)\tilde{\mathcal{O}}(\sqrt{\gamma_N/N}) bound on the simple regret performance of a pure exploration algorithm that is significantly tighter than the existing bounds. We show that this bound is order optimal up to logarithmic factors for the cases where a lower bound on regret is known. To establish these results, we prove novel and sharp confidence intervals for GP models applicable to RKHS elements which may be of broader interest.

View on arXiv
Comments on this paper