Dimension Selection in Axis-Parallel Brent-STEP Method for Black-Box Optimization of Separable Continuous Functions

Abstract

The recently proposed Brent-STEP algorithm was gener alized for separable functions by performing axis-parallel searches, interleaving the steps in individual dimensions in a round-robin fashion. This article explores the possibility to choose the dimension for the next step in a more “intel ligent way”, i.e. to optimize first along dimensions which are believed to bring the highest profit. We present here the results for the epsilon-greedy strategy, and for a method based on the internals of the Brent-STEP algorithm. Al though the proposed methods work better than the round robin strategy in some situations, due to the marginal im provement they bring we suggest the round robin strategy to be used, thanks to its simplicity.