Research And Implementation Of BP Neural Network Classifier Optimization Algorithm

The research of neural network originated from the 1940 s to 1950 s.The mathematical model of information processing based on the structure similar to the synaptic connection of brain is called artificial neural network.Among them,BP neural network has superior performance,so it is widely used in life.However,BP neural network still has defects.In order to overcome the defects and further improve the performance of BP neural network classifier,the particle swarm optimization algorithm is applied to optimize the BP neural network classifier,and a new network classifier is designed.In this paper,through the research and improvement of the traditional particle swarm optimization algorithm PSO algorithm,the improved algorithm is combined with BP neural network classifier to optimize the classifier.Traditional particle swarm optimization(PSO)algorithm is easy to fall into local extremum,slow operation speed and other shortcomings,so the optimized particle swarm optimization algorithm SCPSO(self-Changing particle swarm optimization algorithm improves the shortcomings of traditional particle swarm optimization algorithm by dynamically adjusting the parameters and dispersing and mutating the particles,which makes the global optimization ability enhanced and the operation speed faster.The dynamic adjustment of parameters leads to the contraction factor through the learning factor,and adjusts the convergence speed through the contraction factor to speed up the operation speed of PSO.When the particle is trapped in the local extremum and cannot get rid of it,the particle mutation is carried out to get rid of the local extremum and get the global optimal solution quickly.The internal structure of BP neural network classifier is optimized by applying the corresponding relationship between the particle and BP neural network connection weight in the optimized particle swarm optimization algorithm SCPSO algorithm,so as to optimize BP neural network classifier.Compared with a variety of algorithms for BP neural network classifier,the optimization effect of SCPSO algorithm is the best.The optimized BP neural network classifier is used to detect malicious web pages.The results show that the optimized classifier has better performance,and its convergence speed and stability are improved.