A method and system for optimizing network parameters of the simplest echo state based on particle swarm

Abstract

The invention relates to a method and a system for optimizing parameters of a minimum complexity echo state network based on a particle swarm. The method comprises the following steps of (1) establishing a minimum complexity echo state network model; (2) setting initialization parameters of the minimum complexity echo state network model; (3) establishing a fitness function; (4) computing a particle object function; (5) updating an individual optimal solution and structure parameters of each particle; (6) judging whether to reach an end condition; and (7) outputting an optimal solution of the particle swarm. The minimum complexity echo state network model used by the invention comprises an input layer, a reservoir and an output layer and is provided with a determined reservoir structure, and nerve cells in the reservoir are connected through an annular structure, so that the stability of the reservoir topology is enhanced and the calculated amount is simplified; and a plurality of parameters of the minimum complexity echo state network are optimized based on the characteristics of high speed of calculation and strong global searching ability of a particle swarm optimization algorithm, so that the optimal solution of the particle swarm is obtained, and the prediction precision is improved.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to a method and system for optimizing parameters of the simplest echo state network based on particle swarms. Background technique [0002] Echo State Network ESN (Echo State Network) is a new type of recurrent neural network. Its unique dynamic reserve pool structure enables the network to have good short-term memory capability. Compared with traditional recurrent neural networks, the biggest advantage of ESN is that it simplifies The training process of the network solves the problem that the structure of the traditional recurrent neural network is difficult to determine and the training algorithm is too complicated. It also overcomes the problem of memory fading in the recurrent network, but there are still some problems, such as: [0003] (1) The random topology inside the reserve pool is mapped to an unknown high-dimensional space; [0004] (2) The selection of ESN structure pa...

AI Technical Summary

Problems solved by technology

[0004] (2) The selection of ESN structure parameters relies on experience without the support of scientific theory

[0006] Compared with the classic echo state network, although MESN greatly simplifies the ESN structure and solves the first problem, that is, the random topology inside the reserve pool is mapped to an unknown high-dimensional space, but its parameter selection such as the number of neurons, spectral radius, Input weights, internal weights, and feedback weights are mostly realized by trial and error in a given parameter space, or selected based on experience, which has great blindness and uncertainty. Therefore, how to find the most suitable Parameters to solve the problem of parameter selection has become a hot research direction

Images