A study on parameter setting methods for solving the traveling salesman problem using ring-shaped self-organizing maps

Abstract

In the Traveling Salesman Problem (TSP), when the number of cities increases from hundreds to thousands, the number of trials itself is limited due to the increase in calculation time, and it becomes impossible to even get the correct solution. In the SOM method, the calculation time increases on the order of N2. This N2 factor etermines the limit of SOM. In this study, we focused on the function of neighborhood radius, one of the parameters, and attempted to reduce the number of learning cycles.