Detection of Network Intrusion based on Deep Learning
Chunli Li · Information Security and Communications Privacy · 2014
Aiming at how to apply deep learning to intrusion detection and improve detection accuracy rate,and combined with characteristics of network data,this paper proposes a design method of deep learning network,and with this,an intrusion detection method based on deep learning network is suggested. This method,with an auto- encoder network,extracts network characteristics and with a softmax classifier classifies the extracted data and thus acquires the results of network intrusion detection and analysis. Experiments based on KDD99 database show that this method is 40% lower than other algorithm in error detection rate while maintaining fairly high detection rate,and all this proves incontrovertibly,thereby verifies its effectiveness.