Evolutionary algorithms guided by Erdős-Rényi complex networks.

Area of Science:

• Computational Intelligence
• Network Science
• Optimization Algorithms

Background:

• Evolutionary algorithms (EAs) are widely used for complex optimization problems.
• Traditional EAs can face challenges in solution refinement and convergence speed.
• Complex networks offer a framework to model relationships and dynamics within populations.

Purpose of the Study:

• To propose a novel evolutionary algorithm that integrates Erdős-Rényi complex networks.
• To enhance the regulation of population crossovers and candidate solution refinement.
• To investigate the impact of network structure on optimization performance.

Main Methods:

• Conceptualizing the algorithm's population as an interrelated complex network.
• Utilizing Erdős-Rényi random graphs to model dynamic connections between solutions.
• Comparing the proposed algorithm against traditional EAs and other network-based algorithms on Traveling Salesman Problem instances.

Main Results:

• The Erdős-Rényi dynamic network-guided algorithm outperformed traditional and alternative network-based algorithms.
• The proposed model achieved improved convergence rates.
• Shorter execution times were observed with the novel approach.

Conclusions:

• Complex network characteristics provide valuable insights for solving optimization problems.
• Network structures can effectively regulate decision-making processes in optimization.
• The proposed integration of Erdős-Rényi networks offers a promising strategy for enhancing evolutionary computation.