Intrusion Detection Results Analysis Based on Variational Auto-Encoder

In the face of increasingly complex network environment, network security plays an increasingly important role in today's era. Therefore, analyze and detect network intrusion based on existing collected data in the complex network environment is one of the current research hotspots in the field of network security. Data mining and knowledge discovery based on the big data analysis is also a major trend of current research on intrusion detection. In the analysis and detection of network intrusion based on big data, the key step is to cluster and extract features of a large amount of collected data in advance. The model introduced in this paper mainly improves the structural model of variational auto-encoder[1] and adds the function of sample clustering, which integrates feature extraction, sample clustering and sample generation[1][2][3], laying a foundation for the detection and analysis of network intrusion behavior based on big data.