Design of a Hybrid Logic Based Adaboost Decision Tree Model for Identifying Web Attacks

Abstract

The foremost development in the volume and significance of web communication through the internet has enlarged the necessity of better security protection. The experts in providing security, while protecting the system maintains a record with containing marks of huge amount of support in recognizing attack revealing. Moreover, this limits the system capability as it can identify only the known attacks that are present in the database, in order to overcome this crisis, ensemble classifier to identify unknown attacks in the internet is proposed. This intrusion detection process involves elimination of redundant and irrelevant features using wrapper based and filter based approach. A hybrid Logic based Adaboost Decision tree model is employed here. The anticipated ensemble classifier was utilized in the online available NSL-KDD dataset which is an improved version of KDD cup dataset from 1999. The experimental outcomes demonstrates that the proposed method shows better trade off than the existing methods in terms of accuracy by 88.12 % in detecting the attacks than the traditional methods, while considering low false rejection rates. This proposed method is simulated in MATLAB environment to compute the accuracy.