Machine Learning Approach to Network Intrusion Detection

Network Intrusion (forcefully entering a network) Detection Systems are now a critical means to provide security against malicious activities. In network intrusion detection research, one of the popular strategies for finding attacks is monitoring a network’s activity for anomalies which are deviations from normal profiles previously learned from benign traffic, identified typically using tools borrowed from the machine learning community. This project uses data mining techniques to extract data and features required for machine learning tools. Tools that are used are Anomaly Detection and Neural Networks. Anomaly Detection is used to classify data as either malicious or normal. Malicious data and normal data are passed on to Neural Networks to remove all the false predictions. This is a research project to use Machine Learning tools to provide a robust system that is flexible, adaptive, simple and deployable in real-time systems.