Research On Intrusion Detection Model Based On Artificial Neural Network Combined With Genetic Algorithm

Along with the high-speed development of Internet and computer technology, the Internet communication has penetrated into the political, economic, culture and life, and various fields of science, greatly affects the progress and development of all aspects of human society.At the same time, it is also greatly influencing and changing people’s life, work and study.However, with the development of Internet, various security problems have emerged.Relevant statistics show that in the worldwild, the network intrusion events ouucr every 20 seconds. The hacker can easily steal your private files, remove your bank account information, destroy your personal account information, to expose your private letters, optional correction, disturb and destroy your information in the database, or even directly destroy your disk or computer hardware implementation,lead to your network’ collapse.So, it has become a hot issue of academic and business research to do research on some practical and effective Internet security technology to protect the security of computer system and Internet system.Intrusion detection is a kind of active defense way of invasion, the main module is to maintain the network security, so the technology become the research hot spot.At present,the intrusion detection technology is in the direction of intelligent automation.The so-called intelligent intrusion detection is that the intelligent algorithms such as genetic algorithm and artificial neural network algorithm are used for intrusion detection.The rate of false positives and non-response rates of traditional IDS are high, it can not greatly identify new types of attacks.The traditional intrusion detection technology is not up to the requirement of real-time.In addition, the traditional genetic algorithm and neural network algorithm has the shortcomings such as slow convergence speed and easy to fall into local minimum values and so on.This article put great analysis and study in the characteristics of the intrusion detection algorithm and genetic algorithm neural network structure and do a series of improvements on the crossover and mutation and selection of traditional genetic algorithm(ga) based on rough set theory to adapt to the requirement of real-time. It also reset the fitness function to make it quickly delete redundant attributes and reduce the large knowledge system.Then it do a series of improvements and put my own ideas into the traditional neural network algorithm to make it identify intrusion data correctly when the reducted large knowledge system is input.Finally it set up two sets of experiment to prove the effectiveness of the algorithm proposed in this paper useing KDDCUP99 dataset in the Windows environment by Rosetta and MATLAB software.