Image Recognition of Human Faces based on BP Neural Network and Particle Swarm Optimization

Abstract

With the development of artificial intelligence and machine learning, BP neural network has been widely studied in the realm of face recognition. To address the problems that it is sensitive to initial weights and thresholds, easily fall into local minimum, and have slow learning rates. This paper uses an adaptive mutation particle swarm optimization to improve BP networks, and the network used is compared with a single BP network in the ORL database for comparison experiments. Finally, the experimental results demonstrate that the algorithm has faster learning rate and higher recognition rate.