Preference-Aware Constrained Multi-Objective Bayesian Optimization

TL;DR

This paper introduces PAC-MOO, a Bayesian optimization method that efficiently finds optimal solutions in complex, constrained, multi-objective problems with preferences, demonstrated on analog circuit design tasks.

Contribution

The paper presents a novel preference-aware constrained multi-objective Bayesian optimization approach called PAC-MOO, addressing challenges in large, constrained design spaces with preferences.

Findings

PAC-MOO outperforms prior methods in real-world analog circuit design problems.

It effectively learns surrogate models for objectives and constraints.

The approach efficiently identifies the Pareto front considering preferences.

Abstract

This paper addresses the problem of constrained multi-objective optimization over black-box objective functions with practitioner-specified preferences over the objectives when a large fraction of the input space is infeasible (i.e., violates constraints). This problem arises in many engineering design problems including analog circuits and electric power system design. Our overall goal is to approximate the optimal Pareto set over the small fraction of feasible input designs. The key challenges include the huge size of the design space, multiple objectives and large number of constraints, and the small fraction of feasible input designs which can be identified only after performing expensive simulations. We propose a novel and efficient preference-aware constrained multi-objective Bayesian optimization approach referred to as PAC-MOO to address these challenges. The key idea is to learn surrogate models for both output objectives and constraints, and select the candidate input for evaluation in each iteration that maximizes the information gained about the optimal constrained Pareto front while factoring in the preferences over objectives. Our experiments on two real-world analog circuit design optimization problems demonstrate the efficacy of PAC-MOO over prior methods.