A network intrusion detection method based on deep learning with higher accuracy
Yuening Zhang, Yiming Zhang, Nan Zhang, Mingzhong Xiao
Abstract
The traditional network intrusion detection methods have the problem of long distance dependency. It is easy to ignore contextual information. Moreover, the current data dimension is too high and the feature extraction process is complex, which is not conducive to the requirements of real-time and accuracy of intrusion detection. For the above two problems, this paper presents a new network intrusion detection method based on Auto-Encoder network(AN) and long-term memory neural network (LSTM). First, KDDcup99 data set is used and pre-processed. And an Auto-Encoder network model is constructed by superimposing multiple auto-encoder networks to map high-dimensional data to low-dimensional space. Then the LSTM model optimized the cell structure was used to extract features, train data and predict intrusion detection types. The experimental results show that compared with several classical methods, the accuracy of network intrusion detection is improved by 2% on average, and the false alarm rates are lower.