Analytically computable neural networks perceptron type

The paper describes the main provisions of the developed technology for analytical calculation of the values of weights of neural networks perceptron type.Analytical formulas are given by which the values of the weights are calculated.The basic proposed architecture of a perceptron-type neural network is presented, on the basis of which the values of the perceptron weights can be calculated.The work also presents numerous experimental results that show the capabilities and advantages of analytical calculation of perceptron weights.In particular, experiments have shown that a perceptron with analytically calculated weight values learns much faster and is more resistant to a decrease in the size of the training database.