| With the globalization of the world economy and the development of the information society, the Internet has played an important role. But it has also brought various kinds of problems, especially computer security, which has become a serious practical trouble.Security audit system works as a third line of defense protecting the system from the security issues can't be detected by the firewall and the intrusion detection systems, it can find and take precautions against intrusion behavior by checking the system trails. Most traditional security audit system is rule-based, which requires constantly update of the security rules. But the attacker's methods and hardware and software of the system attacked computer will continue to change. Therefore, the abnormal behavior can never be accurately identified no matter how the rules are updated.To find a solution to the problem mentioned above, this article proposes a security audit system model based on back propagation algorithm optimized by adaptive genetic algorithm. The paper assimilates both advantages of strong convergence of genetic algorithm in full and fast convergence of BP algorithm in part and adopts the adaptive probability crossover and the adaptive probability mutation in order to guarantee the diversity as well as the stability of groups in fine mode and ensure that no damages occur to them while generating of a new entity, which can provide the best probability of the crossover and mutation.Via optimizing the initial weight. of the BP neural network by the adaptive genetic algorithm, the optimized neural network implemented learning step, find the optional network structure and then can analyze the audit trails and response to the abnormal action after learning the work mode. The comparison and analysis of the algorithm experimental results show that the back propagation algorithm optimized by adaptive genetic algorithm can be used in security audit system effectively and validly. |
PDF Full Text Request