The changing pattern of cyber-attacks has increased the difficulty of detecting network traffic anomalies. Based on the enabling effect of artificial intelligence, cyberspace security faces new risks, which include increasingly intelligent cyber-attacks, more frequent large-scale attacks, more covert cyber-attacks, more adversarial gaming of cyber-attacks, and increasing vulnerability to the theft of important data. Maintaining network security is both an offensive and defensive game. Network traffic anomaly detection, as a prerequisite for ensuring network security, is receiving more attention because it can identify unknown network attacks. Therefore, it is key to understand how to build an intelligent and efficient network anomaly traffic detection model.