3d TSP and Kohonen networksThe Traveling Salesman Problem (TSP) consists of finding the shortest (closed!) path connecting a number of points (e.g. cities). This kind of problem can be solved by finding an optimal solution (that usually isn't the perfect one) through a series of classical algorithms. This program proves the ability of a neural network to adjust itself to the given inputs (cities), creating a path close enough to the perfect one. The cities (in a 3D-environment) are shown in green, the network neurons are shown in red, the links between neurons are white. The various options on the left allow to shape the network's behaviour and to change its evolution speed and the number of cities showed to the network. Simply by dragging the cursor on the points, the model will rotate for a better view. |
||