The Spontaneous Evolution Genetic Algorithm for solving the traveling salesman problem
William W.Y. Hsu, Ching‐Chi Hsu · 2001
In this paper, we present a new model of genetic algorithms, the Spontaneous Evolution Genetic Algorithm (SEGA) that incorporates both the concept of spontaneous generation and winner-take-all competition. It employs the idea of adding potential members during the search process, which keeps the evolution process boiling and thus preventing premature convergence. The winner-take-all competition preserves elitism during the search process. We used a learning scheme to preserve past results for future use in creating new candidates, which are the potential members, during the spontaneous generation phase. Experiments conducted on TSP with this new SEGA model proved that it produces better quality solution than bare GLS, LK local search, and simple SEGA (without learning memory).