Date of Publication :9th November 2017

Abstract: An Intrusion Detection System (IDS) gathers and evaluates information from different locations, and finds potential security risks that include exterior as well as inside of the organization. It contains an enormous volume of data with irrelevant and redundant features which result in longer processing time and poor detection rate. So, feature selection should be empowered as an important characteristic for better performance on massive datasets. Feature selection refines the high dimensional data sets by removing over fitting and curse of dimensionality problems mainly in the domain of machine learning. The perceptive of feature selection lies in increasing the accurateness. In this paper, Fuzzy_Chi_Euc algorithm was given for selecting best features in KDD Cup 99 data set. In this algorithm integration of two filtering methods is done. The fuzzy inference rules are applied for selecting the features. The classification is carried out for finding intrusion and normal data using Support Vector Machines (SVMs). From the experiments conducted it is shown, most significant and relevant features are thus helpful for classification, which, in turn, reduce the time of training with better classification accuracy

Reference :