High-Dimensional Bayesian Optimization with Constraints: Application to Powder Weighing

TL;DR

This paper introduces a novel high-dimensional Bayesian optimization method that incorporates constraints through disentangled representation learning, significantly reducing trial numbers in powder weighing applications.

Contribution

It presents a new approach combining parameter decomposition and nonlinear embedding to effectively handle constraints in high-dimensional Bayesian optimization.

Findings

Reduces number of trials by approximately 66% in powder weighing.

Effectively considers both equality and inequality constraints.

Improves optimization efficiency in high-dimensional black-box problems.

Abstract

Bayesian optimization works effectively optimizing parameters in black-box problems. However, this method did not work for high-dimensional parameters in limited trials. Parameters can be efficiently explored by nonlinearly embedding them into a low-dimensional space; however, the constraints cannot be considered. We proposed combining parameter decomposition by introducing disentangled representation learning into nonlinear embedding to consider both known equality and unknown inequality constraints in high-dimensional Bayesian optimization. We applied the proposed method to a powder weighing task as a usage scenario. Based on the experimental results, the proposed method considers the constraints and contributes to reducing the number of trials by approximately 66% compared to manual parameter tuning.