Towards explainable metaheuristics: PCA for trajectory mining in evolutionary algorithms.

Fyvie, Martin; McCall, John A.W.; Christie, Lee A.

Abstract

The generation of explanations regarding decisions made by population-based meta-heuristics is often a difficult task due to the nature of the mechanisms employed by these approaches. With the increase in use of these methods for optimisation in industries that require end-user confirmation, the need for explanations has also grown. We present a novel approach to the extraction of features capable of supporting an explanation through the use of trajectory mining - extracting key features from the populations of NDAs. We apply Principal Components Analysis techniques to identify new methods of population diversity tracking post-runtime after projection into a lower dimensional space. These methods are applied to a set of benchmark problems solved by a Genetic Algorithm and a Univariate Estimation of Distribution Algorithm. We show that the new sub-space derived metrics can capture key learning steps in the algorithm run and how solution variable patterns that explain the fitness function may be captured in the principal component coefficients.

Citation

FYVIE, M., MCCALL, J.A.W. and CHRISTIE, L.A. 2021. Towards explainable metaheuristics: PCA for trajectory mining in evolutionary algorithms. In Bramer, M. and Ellis, R (eds.) Artificial intelligence XXXVIII: proceedings of 41st British Computer Society's Specialist Group on Artificial Intelligence (SGAI) Artificial intelligence international conference 2021 (AI-2021) (SGAI-AI 2021), 14-16 December 2021, [virtual conference]. Lecture notes in computer science, 13101. Cham: Springer [online], pages 89-102. Available from: https://doi.org/10.1007/978-3-030-91100-3_7

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 41st British Computer Society's Specialist Group on Artificial Intelligence (SGAI) Artificial intelligence international conference 2021 (AI-2021)
• Start Date: Dec 14, 2021
• End Date: Dec 16, 2021
• Acceptance Date: Aug 31, 2021
• Online Publication Date: Dec 6, 2021
• Publication Date: Dec 31, 2021
• Deposit Date: Sep 16, 2021
• Publicly Available Date: Dec 6, 2021
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Volume: 13101
• Pages: 89-102
• Series Title: Lecture notes in artificial intelligence
• Series Number: 13101
• Series ISSN: 0302-9743
• Book Title: Artificial intelligence XXXVIII: proceedings of 41st British Computer Society's Specialist Group on Artificial Intelligence (SGAI) Artificial intelligence international conference 2021 (AI-2021) (SGAI-AI 2021), 14-16 December 2021, [virtual conference]
• ISBN: 9783030910990
• DOI: https://doi.org/10.1007/978-3-030-91100-3_7
• Keywords: Evolutionary algorithms; PCA; Explainability; Population diversity
• Public URL: https://rgu-repository.worktribe.com/output/1457054