Towards a Classifying Artificial Immune System for Web Server Attacks

Classic artificial immune systems for security provide only a simple binary classification of "attack" versus "normal". This work explores expanding an artificial immune system for Web server requests into a classifying system that can categorize the attack as one of several common attack categories. Classification can provide a system administrator with an indication of the severity of the attack and can help direct attack mitigation. This work shows promise at the task of classifying Web server attacks, but still requires some fine-tuning to get the best performance.