Application Of Based Semi-adaptive Genetic Neural Network In Intrusion Detection

Along with the computer technology with wide range of application, the government, enterprises and individuals to computer dependence is getting higher and higher. Meanwhile, computer network security problem more and more serious. At present, the computer network security technology, mainly including firewall, routers packet filtering, all sorts of antivirus software and holes patches, these all belong to the static security defense technology, while the illegal intrusion activities play a certain role, but they are unpredictable difficult in the system of the loss of intercepting behavior before the attack.Intrusion Detection System is a kind of active network, it can completely defense technology to make timely response invasion of events. However, most of the intrusion detection system has higher rate and efficiency is not ideal. Genetic neural network self-learning, lenovo has advantages such as memory and distributed processing, genetic neural network to the intrusion detection, can make up for the deficiencies of traditional intrusion detection technology. Research the neural network in intrusion detection application, the main work is as follows:(1) On the intrusion detection system, analyzes the research of intrusion detection technology, mainly classification method of intrusion detection technology, and mainly, expounds the development direction of the intrusion detection system.(2) The study of artificial neural network, and the basic principle and function of artificial neural network in the BP algorithm is studied and analyzed its existence with slow convergence speed, easy to fall into the local minimum value.(3) Genetic algorithm is studied on the basic principle of genetic algorithm, the application of neural network to carry on the detailed analysis. A comprehensive comparison of traditional genetic algorithm with adaptive genetic algorithm, and analyzed their performance and deficiencies.(4) In adaptive genetic algorithm is proposed, and the problem of time-consuming and adaptive genetic algorithm. In half an adaptive genetic algorithm, the mutation probability is no longer through calculating the fitness of each individual in the population, but through the generation of mutation probability directly. Combining the LM algorithm to further improve the local search properties of genetic neural network is applied to the intrusion detection of detection efficiency. (5) Completed two groups of simulation experiment, the experimental results show that the method proposed is training time is short, the advantages of false positives, achieve the expected effect.