Contribution-Based Cooperative Co-Evolution for Nonseparable Large-Scale Problems With Overlapping Subcomponents.

Area of Science:

• Artificial Intelligence
• Computational Optimization

Background:

• Cooperative co-evolutionary algorithms excel at large-scale problems but struggle with nonseparable problems featuring overlapping subcomponents.
• Shared variables in these problems hinder decomposition, and existing frameworks fail to balance cooperation frequency and resource allocation for overlapping parts.

Purpose of the Study:

• To propose a novel contribution-based cooperative co-evolutionary algorithm for effective and efficient optimization of nonseparable large-scale problems with overlapping subcomponents.
• To address the challenges of shared variable decomposition and maintaining crucial optimization factors.

Main Methods:

• A contribution-based decomposition method assigns shared variables to the subcomponent with the highest overall problem contribution.
• A new contribution-based optimization framework utilizes a round-robin structure to prioritize important subcomponents, ensuring high cooperation frequency and efficient resource allocation.

Main Results:

• The proposed algorithm demonstrates significantly superior performance compared to state-of-the-art methods.
• Effective grouping structures and accelerated optimization speeds were key factors in the improved performance.

Conclusions:

• The novel contribution-based approach offers an effective solution for decomposing and optimizing nonseparable large-scale problems with overlapping subcomponents.
• The algorithm successfully balances cooperation and resource allocation, leading to enhanced optimization efficiency and effectiveness.