Anomaly detection approach using hybrid algorithm of data mining technique

Abstract

The excessive use of the communication networks, rising of Internet of Things leads to increases the vulnerability to the important and secret information. advance attacking techniques and number of attackers are increasing radically. Intrusion is one of the main threats to the internet. Hence security issues had been big problem, so that various techniques and approaches have been presented to address the limitations of intrusion detection system such as low accuracy, high false alarm rate, and time consuming. This paper proposes a hybrid machine learning technique for network intrusion detection based on combination of K-means clustering and Sequential Minimal Optimization (SMO) classification. It introduces hybrid approach that able to reduce the rate of false positive alarm, false negative alarm rate, to improve the detection rate and detect zero-day attackers. The NSL-KDD dataset has been used in the proposed technique.. The classification has been performed by using Sequential Minimal Optimization. After training and testing the proposed hybrid machine learning technique, the results have shown that the proposed technique (K-mean + SMO) has achieved a positive detection rate of (94.48%) and reduce the false alarm rate to (1.2%) and achieved accuracy of (97.3695%).