Review of swarm intelligence algorithm optimization of BP neural network

Abstract

As a model used for parameter estimation, BP neural network has a remarkable effect in many prediction algorithms. However, there are situations that will fall into the local optimal solution and the learning speed is slow. To solve these two problems, this paper combines the dynamic adaptive strategy in the genetic algorithm, the method of eliminating honey sources in the bee colony algorithm, the adaptive greedy strategy in the ant colony algorithm, and the introduction of variation operator in the particle swarm algorithm to improve the two defects of the BP neural network, It has made optimization research on the use of BP neural network and prediction, and also laid the groundwork for the future research on optimization of BP neural network.