Project translated automatically. Log in or register, to view the original The method is attached, the code is commented.1 .Neural networks.Training of Perseptron.and str.8) The Set random points in space - 10 in each of the eight areas formed by eight coordinates.Training a neural network of the type per10 septron, consisting of 3 entrances and 8 outlets so that it can recognize which of the 8 areas of space is a point (similar to an example for flatness).2ndNeural networks.Learning without a teacher.(page 6 ) Apply the cohonen neuronal network with self-learning for the clustering task.In the first stage generate random points on a flat around 2 cluster centers (approximately 20-30 points).Furthermore, consider that the network has two entrances (point coordinates) and two outputs - one of them is equal to 1, the other 0 (by which cluster the point belongs).By submitting consistently to the input (variable) point, set the network by applying the described learning procedure so that it acquires the ability to determine which cluster the point belongs to.The coefficient to choose by reducing it from step to step by the rule = (50-i)/100, and for each neuron it will be its value, and adapting on each step will be the weight of only one (winning) neuron.