Research And Implementation Of Intrusion Detection System Based On Neural Network

With the rapid development and popularization of network technology,network security is becoming more and more prominent.The existing security measures have been unable to meet the actual needs.Intrusion detection as a proactive security technology,has attracted more and more attention of network security personnel.The current intrusion detection system in the practical application still exist many defects,such as high and low detection efficiency and intelligent degree of lack of false positive rate and false negative rate.In order to improve the detection accuracy and reduce false positive rate and false negative rate,most people will study focus on the proper selection of data sources and data attributes,improvement of existing detection algorithm,detection algorithm for discovering new and improved intrusion detection system architecture method.The application of BP neural network in intrusion detection system is studied in this paper,and the problems in the application of BP neural network are discussed,and a new method for improving the BP neural network is presented.According to the advantages and disadvantages of BP neural network algorithm adaptive learning rate,the method of variable weight and threshold learning rate is put forward,which makes full use of the direction information of the local gradient,and realizes the change of the learning rate according to the needs of the algorithm.This system takes WindowsXP as the platform,takes vc++6.0 as the main development tool,and uses MATLAB to carry on the training and the examination to the neural network.Finally,the results of the experiments and optimization of the BP neural network identification results were compared to draw the conclusion.Finally draws the conclusion through learning rate for adjustment and can adjusting factor of the transfer function,the improved algorithm can better the ease of learning rate in the transfer function tends to saturation interval regulation ineffective problem,error function early to get rid of the flat region,greatly improve the network convergence precision,reduces the training time and accelerate the convergence speed and the improved algorithm can achieve higher detection rate and lower false positive rate and false negative rate,make the intrusion detection system has a better performance.