Cyber-Attack Classification using Improved Ensemble Technique based on Support Vector Machine and Neural Network

Abstract

Cyber-attack classification and detection process is based on the fact that intrusive activities are different from normal system activities .Its detection is a very complex process in network security. In current network security scenario various types of cyber-attack family exist, some are known family and some are unknown one . The detection of known attack is not very difficult it generally uses either signature base approach or rule based approach, but to find out the unknown one is a challenging task. Intrusion detection is a process for this .One of the major developments in machine learning in the past decade is the ensemble method, which finds highly accurate classifier by combining many moderately accurate component classifiers. This paper addresses using of an ensemble classification methods for intrusion detection. The paper proposes a cascaded support vector machine classifier or an improved ensemble classifier using multiple kernel function. The multiple kernel is Gaussian in nature. The graph based /neural network technique used for feature collection of different types of cyber-attack data. The proposed algorithm is very efficient in comparison of pervious method. General Terms SVM, Gaussian hyper plane, Euclidean distance KeywordsCyber-attack Classification, Ensemble Technique NN, KDDCUP99, k-Nearest Neighbour.

An Empirical Comparison on Malicious Activity Detection Using Different Neural Network-Based Models

“…Govindarajan and Chandrashekaran [6] which also put forth an anomaly intrusion detection system to assess the efficiency of the RBF-SVM hybrid system by carrying out various experiments on the NSL-KDD dataset. The authors in [7] suggested an improved ensemble classifier or a cascaded SVM classifier employing more than one kernel function, which is Gaussian in nature. The harnessing of graph-based neural network techniques eases the process of features' collection of different cyber-attack data.…”

An Intelligent Model for Online Recruitment Fraud Detection

“…It total number is important to observe the total number of faults to assess the classifiers works [36]. The evaluation parameters represented on the following [37]: 1) Accuracy is the quality measurement for classifier, which can be calculated through mathematically finding the ratio of correct classified attempts number according to the total number of all classification attempts.…”