A neural network face recognition algorithm based on surface-simplex swarm evolution

Abstract

In the BP neural network training algorithm, the optimization of weights is easy to fall into the local extreme points and the convergence rate is slow. Many researches introduce the intelligent optimization algorithm to improve them, but the traditional intelligent optimization algorithm usually has many Control parameters, if not correctly selected parameters, or did not properly select the initial point position, it is difficult to search for the optimal neural network weights. In this paper, a BP neural network learning algorithm based on surface-simplex swarm evolution is proposed. It reduces the control parameters of the algorithm by all random search. The diversity of the particles is maintained by the polygonal state of the population, and the algorithm is prevented from falling into the local extreme points. Dependence on initial values. In the application, the algorithm is applied to the training algorithm of face recognition neural network. The experimental results show that the trained neural network effectively improves the recognition rate.