Evolutionary algorithms guided by network topologies

Abstract

This paper explores the integration of network topologies into the evolutionary algorithms. By examining various network structures, including small-world, scale-free, Internet, and tree topologies, we investigate how the arrangement and connectivity of individuals influence the selection and convergence speed of the mate in the classic algorithm. The study compares traditional evolutionary approaches with those guided by specific network topologies, revealing that certain configurations can significantly improve algorithm performance. Our findings indicate that the use of diverse network architectures leads to promising improvements in the optimization process. This research contributes to understanding how network theory could optimize evolutionary computation and shows the way for future studies to explore additional network models and optimization problems.