Detecting Intrusions and Attacks in the Network Traffic using Anomaly based Techniques

Technology has become the backbone of today's Information and Communication Technology. Today a large number of transactions are carried out online and thus possess a risk of information stealing and misuse. Intrusion Detection techniques are thus developed and applied to the networks and individual host computers to detect such intrusive actions and alarm the administrators. A secure machine-controlled anomaly detection system is a more practical procedure to help in network analysis. An anomaly detection system (ADS) examine the network flow and focuses on detecting uncommon network behaviour, and classify them into attacks. The propsoed research work present a comparison of some of these intrusion detection techniques using different machine learning algorithms and ensemble learning. A subset of attributes (significant) is chosen from the primitive set of attributes using a random forest regressor technique, and then, the chosen set of significant features is used to train different classifiers. Moreover, the proposed work has been analyzed by experimenting on one of the publically available dataset CICIDS2017, available from the Canadian University of Cybersecurity.