Batched Large-scale Bayesian Optimization in High-dimensional Spaces

TL;DR

This paper introduces ensemble Bayesian optimization (EBO), a scalable method that effectively handles high-dimensional, large-scale black-box optimization problems by using an ensemble of additive Gaussian process models, enabling rapid optimization with tens of thousands of observations.

Contribution

EBO is a novel approach that addresses large-scale observations, high-dimensional inputs, and batch query selection in Bayesian optimization simultaneously.

Findings

Scaled BO to tens of thousands of observations within minutes

Effectively handled high-dimensional input spaces

Balanced quality and diversity in batch query selection

Abstract

Bayesian optimization (BO) has become an effective approach for black-box function optimization problems when function evaluations are expensive and the optimum can be achieved within a relatively small number of queries. However, many cases, such as the ones with high-dimensional inputs, may require a much larger number of observations for optimization. Despite an abundance of observations thanks to parallel experiments, current BO techniques have been limited to merely a few thousand observations. In this paper, we propose ensemble Bayesian optimization (EBO) to address three current challenges in BO simultaneously: (1) large-scale observations; (2) high dimensional input spaces; and (3) selections of batch queries that balance quality and diversity. The key idea of EBO is to operate on an ensemble of additive Gaussian process models, each of which possesses a randomized strategy to divide and conquer. We show unprecedented, previously impossible results of scaling up BO to tens of thousands of observations within minutes of computation.