55
3

A new integral loss function for Bayesian optimization

Abstract

We consider the problem of maximizing a real-valued continuous function ff using a Bayesian approach. Since the early work of Jonas Mockus and Antanas \v{Z}ilinskas in the 70's, the problem of optimization is usually formulated by considering the loss function maxfMn\max f - M_n (where MnM_n denotes the best function value observed after nn evaluations of ff). This loss function puts emphasis on the value of the maximum, at the expense of the location of the maximizer. In the special case of a one-step Bayes-optimal strategy, it leads to the classical Expected Improvement (EI) sampling criterion. This is a special case of a Stepwise Uncertainty Reduction (SUR) strategy, where the risk associated to a certain uncertainty measure (here, the expected loss) on the quantity of interest is minimized at each step of the algorithm. In this article, assuming that ff is defined over a measure space (X,λ)(\mathbb{X}, \lambda), we propose to consider instead the integral loss function X(fMn)+dλ\int_{\mathbb{X}} (f - M_n)_{+}\, d\lambda, and we show that this leads, in the case of a Gaussian process prior, to a new numerically tractable sampling criterion that we call EI2\rm EI^2 (for Expected Integrated Expected Improvement). A numerical experiment illustrates that a SUR strategy based on this new sampling criterion reduces the error on both the value and the location of the maximizer faster than the EI-based strategy.

View on arXiv
Comments on this paper