Neural network construction method

Abstract

The invention provides a neural network construction method, which comprises the following steps of obtaining neural network generation parameters, wherein the generation parameters comprise a neuronclustering number, a neuron concentration degree parameter, a distribution space size parameter and a neuron total number; generating a neural network according to the neural network generation parameters, wherein the neural network satisfies the following formula: x(n+ 1) = W1u (n + 1) + W2x (n) + W3y (n); Calculating a state transformation matrix W2 of the neural network, wherein the state transformation matrix is used for obtaining the next internal state of the neural network according to the current internal state of the neural network; Using a preset training set to train the neural network, obtaining an input and output mapping matrix in the training process, and enabling the input and output mapping matrix to determine output uniquely according to input. The complete neural networkis constructed from the perspective of framework generation of the neural network and state transformation matrix setting, and the connection structure of the neural network is more similar to that of a biological network compared with the neural network obtained in the prior art.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a neural network construction method. Background technique [0002] The construction of the neural network is based on the premise of the application of the neural network. In the past ten years, the biological neural network system that simulates the biological neural network has excellent performance in the fields of identification, decision and prediction. The biological neural network has better intelligence and adaptability by simulating the biological neural network, but usually the completely random connection of each neuron in the neural network leads to a high degree of coupling inside the neural network and insufficient dynamic characteristics, resulting in The adaptiveness of the neural network is difficult to improve and the output error is difficult to reduce. Contents of the invention [0003] In order to solve the above technical problems, the present invention pr...

AI Technical Summary

Problems solved by technology

The biological neural network has better intelligence and adaptability by simulating the biological neural network, but usually the completely random connection of each neuron in the neural network leads to a high degree of coupling inside the neural network and insufficient dynamic characteristics, resulting in The adaptability of the neural network is difficult to improve and the output error is difficult to reduce

Images