No abstract

People often use similar methods to invade network traffic, such as flood attacks and Ddos attacks. Early detection of malicious traffic usually uses manual filtering to establish a whitelist, and then uses some artificial intelligence to simply measure some characteristic parameters. The high predictive performance of artificial intelligence will inevitably introduce unknowable decision process into network intrusion detection. When it is applied to the extremely important network environment, human's doubt on its black box effect will hinder its advancement. Here, we make a propose to come up with a network traffic intrusion detection method (XAI-IDS) on account of interpretable artificial intelligence to detect malicious traffic intrusion in networks. XAI-IDS analyzes network traffic data to predict whether the traffic is malicious intrusion.

People often use similar methods to invade network traffic, such as flood attacks and Ddos attacks. Early detection of malicious traffic usually uses manual filtering to establish a whitelist, and then uses some artificial intelligence to simply measure some characteristic parameters. The high predictive performance of artificial intelligence will inevitably introduce unknowable decision process into network intrusion detection. When it is applied to the extremely important network environment, human’s doubt on its black box effect will hinder its advancement. Here, we make a propose to come up with a network traffic intrusion detection method (XAI-IDS) on account of interpretable artificial intelligence to detect malicious traffic intrusion in networks. XAI-IDS analyzes network traffic data to predict whether the traffic is malicious intrusion.