Towards Utilizing Fuzzy Self-Organizing Taxonomies to Identify Attacks on Computer Systems and Adaptively Respond

Several methods for doing intrusion detection have been developed over the years. However, most of these methods are based on crisp statistical techniques that measure deviation from a norm. Due to the wide range of attacks on computers, statistical methods are not always effective because they aggregate many system variables into a single mathematical measure. Instead, taxonomies of attack features based on the concepts of fuzzy logic can be utilized to classify attacks and build simple response rules based on local system variables. Taxonomies however require correct hierarchial construction from subtaxonomies of attack classifiers. An architecture that defines self organizing taxonomies based on fuzzy logic is therefore developed for future investigation.