An Self-organizing Neural Network with Convex-hull Expanding Property for TSP

The self-organizing map (SOM) has been explored to handle the Euclidean traveling salesman problem (TSP). By combining its neighborhood preserving property and the convex-hull property of the TSP, a renewed self-organizing neural network for TSP is introduced. In each learning iteration, the method not only draws the excited neurons close to the input vector (city), but also pushes them towards the convex-hull of cities. A new learning rule is proposed and justified. Experimental results demonstrate the advantage of the new method over its previous versions