Surrogate-Assisted Partial Order-based Evolutionary Optimisation

TL;DR

This paper introduces SAPEO, a surrogate-assisted evolutionary algorithm that dynamically selects individuals for evaluation based on model uncertainty and population diversity, improving multi-objective optimization performance.

Contribution

The paper presents a novel surrogate-assisted survival selection method (SAPEO) with variants, enhancing multi-objective evolutionary algorithms by integrating dynamic surrogate modeling.

Findings

SAPEO variants show competitive performance on BBOB benchmarks.

Surrogate models reduce the number of exact evaluations needed.

Results highlight conditions for effective surrogate-assisted optimization.

Abstract

In this paper, we propose a novel approach (SAPEO) to support the survival selection process in multi-objective evolutionary algorithms with surrogate models - it dynamically chooses individuals to evaluate exactly based on the model uncertainty and the distinctness of the population. We introduce variants that differ in terms of the risk they allow when doing survival selection. Here, the anytime performance of different SAPEO variants is evaluated in conjunction with an SMS-EMOA using the BBOB bi-objective benchmark. We compare the obtained results with the performance of the regular SMS-EMOA, as well as another surrogate-assisted approach. The results open up general questions about the applicability and required conditions for surrogate-assisted multi-objective evolutionary algorithms to be tackled in the future.