Detecting User-To-Root (U2R) Attacks Based on Various Machine Learning Techniques

Abstract: Intrusion detection mainly focused on four major attack category such as denial of service, probe, user-to- root, and remote-to-local. This paper focused on user-to-root attack, which the attacker tries to access normal user account and gains root access information of the system. The U2R attacks leads to several vulnerability such as sniffing password, a dictionary attack and social engineering attacks. This paper makes a comparative study analyses for U2R attacks based on several popular machine learning techniques such as navie bayes, random forest, J48, random tree, JRIP and Multilayer perceptron to achieve better accuracy and to reduce mean square error for individual attacks that belongs to user to root category.

Keyword: Intrusion detection, User-to-Root attack, Random Forest, Multilayer perception, J48