Research On Optimized BP Neural Network Intrusion Detection Based On Improved GA-TS Algorithm

Through the research in recent years, many experts and scholars will BP network is applied to the intrusion detection field, greatly improve the intrusion detection system of quality and performance, but it still has many problems, such as BP neural network depends on the initial value,in view of the present situation, this paper on BP neural network were studied. Specific research work are as follows:(1) Crossover operator of genetic algorithm. Before genetic algorithm cross the similarity between individuals is compared, if the similarity between the two individuals is too large, then do not allow cross operation between them.(2) The whole taboo method of tabu search is improved.. The processing method for the solution of a variation, if the original is 0, the mutations of 1, otherwise to 0, the current solution variation obtained solution as a solution, which is an iterative process of the current solution.After follow the normal steps into operation.(3) The improved GA-TS algorithm for optimizing BP network is proposed.. According to the analysis of BP neural network, genetic algorithm and tabu search algorithm and BP neural network dependent initial weights, if the initial value is not well, will cause the network for a long time does not converge, forming local minima. Ga-ts algorithm in this paper improved the BP neural network optimization, combines the advantages of two algorithms, avoiding the shortage of the two, of the proposed method and other methods are compared. Experimental results show this method can get better effect.