Evolutionary Approaches to Expensive Optimisation
This work provides a review and guidelines for researchers and practitioners using evolutionary algorithms in expensive optimization scenarios, but it is incremental as it synthesizes existing knowledge rather than introducing novel methods.
The paper addresses the challenge of expensive fitness evaluations in evolutionary algorithms for complex optimization problems by discussing surrogate-assisted approaches, focusing on key issues like model selection, integration, and reliability without presenting new experimental results.
Surrogate assisted evolutionary algorithms (EA) are rapidly gaining popularity where applications of EA in complex real world problem domains are concerned. Although EAs are powerful global optimizers, finding optimal solution to complex high dimensional, multimodal problems often require very expensive fitness function evaluations. Needless to say, this could brand any population-based iterative optimization technique to be the most crippling choice to handle such problems. Use of approximate model or surrogates provides a much cheaper option. However, naturally this cheaper option comes with its own price. This paper discusses some of the key issues involved with use of approximation in evolutionary algorithm, possible best practices and solutions. Answers to the following questions have been sought: what type of fitness approximation to be used; which approximation model to use; how to integrate the approximation model in EA; how much approximation to use; and how to ensure reliable approximation.