/Solving-TSP-with-Self-Organizing-Maps

An exploration on SOM algorithm to solve traveling salesman problem

Solving TSP with Self-Organizing Maps

This study aims to explore the performance of Self-Organizing Maps algorithm in solving Traveling Salesman Problem by implementing it with two different neigborhood functions:

  1. Gaussian Kernel
  2. Elastic Band

The algorithms have been run on the data set that includes the coordinates of Turkish provinces. The seventh trial (M = 3n, 100 iterations) yields the optimal solution for TSP with Turkish provinces.

Note: The code chunk "run" runs excessively slow. Thus, you can load pre-calculated variables and data frames from "Enivronment - Solving TSP with Self Organizing Maps.RData". Just place the RData file in your working directory and run the following code to skip that part:

load("Environment - Solving TSP with Self Organizing Maps.RData")

Feel free to make suggestions on performance :-)