Intrusion Detection Research Based On Deep Belief Network

Recently, with the development of network technology and computer intrusion means is becoming more and more complex and diverse. Computer network security has become an important problem. Intrusion detection as a new network security technologies, it can search and find the information on the key nodes which are in the network. However, the traditional intrusion detection technology has been unable to meet the current network environment because of the invasion method has changed. therefore intelligent technologies must be used. Deep learning has been a great success in the field of image recognition and text classification. It provides a new method to solve the problem of data quantity and high dimension feature of intrusion dataBased on the existing intrusion detection technology and the deep belief network learning. This paper presents an intrusion detection model based on deep belief network. This model consists of 3 layers of restricted RBM and a layer of BP neural network. The training algorithm uses the contrast divergence algorithm. BP neural network to play the parameters of the callback and the third layer RBM output of the low dimensional characteristics of the classification. And an intrusion detection model based on the deep belief network is designed. Then, this paper uses KDDCUP99 10% data set to simulate the proposed model. In the experiment, the influence of the iteration times of RBM, the number of BP iterations and the setting of the RBM hidden layer nodes on the final classification accuracy is discussed. And the parameters are set to the optimal and TANN、FC-ANN、SA-DT-SVMS、BP for comparison. In general, intrusion detection method based on depth confidence network is more accurate.