Design and Verification of an Adversarial Training Defense Algorithm for Network Traffic

At present, there are more and more attacks targeting network traffic, among which adversarial sample attack method is a popular attack method in recent years, which brings relatively high Security threats. This paper studies the basic principles of network traffic adversarial attacks and adversarial defense, designs a network traffic defense algorithm based on adversarial training to resist adversarial attacks, and verifies its effectiveness and accuracy on the ISCXVPN-2016 dataset.