Intrusion Detection Method based on Deep Learning
Zhijun Wu, Liang Cheng, Yuqi Li · 2021
In response to the issue that the classification accuracy of small proportion attack data is low due to the imbalance of classification in existing intrusion detection methods, this paper mainly focuses on two aspects. First, we put forward a hybrid intrusion detection method founded on AdaBoost algorithm and convolutional neural network (CNN). It utilizes CNN to extract network traffic data, and utilizes AdaBoost algorithm to classify network attack data; Second, the improved sparse autoencoder is combined with softmax classifier to increase the classification preciseness of the model by continuously adjusting the parameters of the sparse autoencoder. This method uses KDD99 data set to test the code in pycharm environment. The test results show that this method has advantages over the existing methods in the detection rate (DR) and false positive rate (FPR) of small proportion attack data. This method is suitable for the analysis of continuous and classified attack data, improves the detection effect of imbalanced data classification, and solves the defects of existing intrusion detection models with large classification error and slow calculation speed.