Certain investigations on the performance of swarm based classification algorithms for intrusion detection using feature selection methods

Abstract

newlineThere is a tremendous growth in the field of information technology due to which network security is also facing significant challenges The traditional Intrusion Detection System IDS is unable to handle the recent attacks and malwares Hence Intrusion Detection System IDS which is an indispensable component of the network needs to be protected IDS methodologies which are currently in use require human intervention to generate attack signatures or to determine effective models for normal behaviour In order to provide a potential alternative to expensive newlinehuman input we are in need of learning algorithms The predominant task of such learning algorithm is to discover appropriate behaviour of IDS as normal and abnormal system is under attack The algorithm should be accurate and it should process the information in quick successions which is one of the major drawbacks in IDS because of the large amount of features newlineData mining based network intrusion detection is widely used to identify how and where the intrusions occur Reducing the number of features by selecting the important features is critical to improve the accuracy and speed of classification algorithms Hence selecting the significant newlinefeatures and developing the best classifier model in terms of high accuracy and detection rates is the focus of the proposed method The ultimate goal is to choose an effective classification approach for developing an accurate intrusion detection model In order to improve the accuracy of an individual classifier we combine the classifiers which is the prevalent approach Recently application of swarm intelligence technique for intrusion detection has gained prominence among the research community newline