Optimization Research Of BP Neural Network Based On Artificial Fish Swarm Algorithm

Artificial neural network is an operation model that simulates the composition,function structure and working mode of biological nervous system.Artificial neural network is one of the major achievements in the development of science and technology in the 20 th century,and it is another important milestone in the history of human society.BP(Back-Propagation)neural network is the fastest growing and most widely used artificial neural network.However,with the continuous deepening of the research on BP algorithm,it is found that there are some defects in the algorithm,which are mainly reflected in the problems such as the algorithm's tendency to fall into the local extreme point and the possibility of nonconvergence.For the defects of BP network algorithm,In this paper,artificial fish school algorithm is used to optimize BP neural network.However,in the process of using Artificial Fish Swarm Algorithm to optimize the BP neural network,the problem that the artificial fish swarm optimization algorithm is prone to turbulence and slow optimization is found.Therefore,this paper first improves the shortcomings of artificial fish swarm algorithm.A layered artificial fish school algorithm(LAFSA)is used.Then the BP neural network is optimized using the LAFSA algorithm,and the optimized algorithm is applied to Web security.The work of this paper mainly has the following three aspects.First: Aiming at the problem that the artificial fish swarm algorithm is easy oscillation and slow speed of optimization,this paper presents a layered artificial fish school algorithm(LAFSA).This algorithm introduces the core ideas of layering and grouping of shuffled frog leaping algorithm,and changes the step length,visual field and congestion factor of artificial fish nonlinearly,which can make the algorithm perform refined search in the later period.Experiments show that the optimization accuracy and speed of the algorithm have been greatly improved.Second: Using the improved artificial fish swarm algorithm to optimize the initial weights and thresholds of the BP network algorithm,and the LAFSA-BPNN algorithm is proposed.The improved algorithm can be used to solve the classification problem.The algorithm first uses LAFSA algorithm to optimize the BP network weights and thresholds,and then uses the optimized weights and thresholds as the initial parameters of the BP algorithm to train the network model.Experiments show that the performance of the BP algorithm has been greatly improved.Third: Apply the LAFSA-BPNN algorithm to Web security.Use this algorithm to detect XSS(cross-site scripting)security vulnerabilities.Add XSS filtering layer to Web applications to filter data submitted by users to the server.To prevent XSS attacks,the process and method for detecting XSS vulnerabilities using the LAFSA-BPNN algorithm are specified.Experiments show that the algorithm has a high recognition rate in detecting XSS attacks.