Performance of Improved SOM-TSP Algorithm for Traveling Salesman Problem of Many Cities
Kikuo Fujimura, Tokutaka Heizo, Masumi Ishikawa · IEEJ Transactions on Electronics Information and Systems · 1999
Angeniol et al. applied Kohonen's Self-Organizing Maps (SOM) to solve the traveling salesman problem (TSP) and showed to have a practically enough solution obtained for a short time in comparison to a method depending on a conventional neural network in 1988. We confirmed already that a calculation time is shortened further about TSP of 500 cities, by introducing a momentum effect to the renewal coefficient of original method (Angeniol's, method; SOM-TSP). Here, we report that we evaluated a performance of our improved method about TSP of more large 1000, 2000, 10000 cities.