Multi-population Ensemble Genetic Programming via Cooperative Coevolution and Multi-view Learning for Classification

2025-09-15

AI Summary

To address classification in high-dimensional heterogeneous feature spaces, this paper proposes the Multi-Population Ensemble Genetic Programming (MEGP) framework, integrating coevolutionary optimization with multi-view learning to enable parallel subgroup evolution and dynamic ensemble interaction. Its key contributions are: (1) a two-level evolutionary mechanism balancing global exploration and local exploitation; (2) a differentiable Softmax-weighted fusion layer replacing conventional hard voting, enabling end-to-end differentiable integration of genetic program outputs—thereby enhancing both interpretability and diversity; and (3) a dynamic ensemble fitness evaluation mechanism that drives cooperative population optimization. Extensive experiments on eight benchmark datasets demonstrate that MEGP significantly outperforms state-of-the-art baselines across Log-Loss, F1-score, and AUC metrics, while achieving faster convergence and higher population diversity.

Technology Category

• Machine Learning: Ensemble Methods
• Search and Optimization: Evolutionary Computation
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

This paper introduces Multi-population Ensemble Genetic Programming (MEGP), a computational intelligence framework that integrates cooperative coevolution and the multiview learning paradigm to address classification challenges in high-dimensional and heterogeneous feature spaces. MEGP decomposes the input space into conditionally independent feature subsets, enabling multiple subpopulations to evolve in parallel while interacting through a dynamic ensemble-based fitness mechanism. Each individual encodes multiple genes whose outputs are aggregated via a differentiable softmax-based weighting layer, enhancing both model interpretability and adaptive decision fusion. A hybrid selection mechanism incorporating both isolated and ensemble-level fitness promotes inter-population cooperation while preserving intra-population diversity. This dual-level evolutionary dynamic facilitates structured search exploration and reduces premature convergence. Experimental evaluations across eight benchmark datasets demonstrate that MEGP consistently outperforms a baseline GP model in terms of convergence behavior and generalization performance. Comprehensive statistical analyses validate significant improvements in Log-Loss, Precision, Recall, F1 score, and AUC. MEGP also exhibits robust diversity retention and accelerated fitness gains throughout evolution, highlighting its effectiveness for scalable, ensemble-driven evolutionary learning. By unifying population-based optimization, multi-view representation learning, and cooperative coevolution, MEGP contributes a structurally adaptive and interpretable framework that advances emerging directions in evolutionary machine learning.

Problem

• Addresses classification challenges in high-dimensional heterogeneous feature spaces
• Decomposes input space into independent feature subsets for parallel evolution
• Enhances model interpretability and adaptive decision fusion through ensemble learning

Innovation

• Cooperative coevolution with multi-view learning integration
• Differentiable softmax weighting for decision fusion
• Hybrid fitness mechanism balancing diversity and cooperation