Explaining evolutionary feature selection via local optima networks

Adair, Jason; Thomson, Sarah L; Brownlee, Alexander E. I.

Abstract

We analyse fitness landscapes of evolutionary feature selection to obtain information about feature importance in supervised machine learning. Local optima networks (LONs) are a compact representation of a landscape, and can potentially be adapted for use in explainable artifcial intelligence (XAI). This work examines their applicability for discerning feature importance in supervised machine learning datasets. We visualise aspects of feature selection LONs for a breast cancer prediction dataset as case study, and this process reveals information about the composition of feature sets for the underlying ML models. The estimations of feature importance obtained from LONs are compared with the coeffcients extracted from logistic regression models (interpretable AI), and also against feature importances obtained through an established XAI technique: SHAP (explainable AI). We find that the features present in the LON are not strongly correlated with the model coeffcients and SHAP values derived from a model trained prior to feature selection, nor are they strongly correlated within similar groups of local optima after feature selection, calling into question the effects of constraining the feature space for wrapper-based techniques based on such ranking metrics.

Citation

Adair, J., Thomson, S. L., & Brownlee, A. E. I. (2024, July). Explaining evolutionary feature selection via local optima networks. Presented at ACM Genetic and Evolutionary Computation Conference (GECCO) 2024, Melbourne, Australia

• Presentation Conference Type: Conference Paper (published)
• Conference Name: ACM Genetic and Evolutionary Computation Conference (GECCO) 2024
• Start Date: Jul 14, 2024
• End Date: Jul 18, 2024
• Acceptance Date: May 3, 2024
• Online Publication Date: Aug 1, 2024
• Publication Date: 2024-07
• Deposit Date: May 13, 2024
• Publicly Available Date: Jul 31, 2024
• Peer Reviewed: Peer Reviewed
• Book Title: Genetic and Evolutionary Computation Conference (GECCO ’24), July 14–18, 2024, Melbourne, VIC, Australia
• ISBN: 9798400704956
• DOI: https://doi.org/10.1145/3638530.3664183
• Keywords: Fitness Landscapes, Explainable AI, Local Optima Networks (LONs)
• Publisher URL: https://dl.acm.org/conference/gecco
• External URL: https://gecco-2024.sigevo.org/