Machine learning is quickly becoming a popular concept throughout the computer science world. Amazon and Netflix are both well known for using machine learning to provide a better experience for their customers. Also numerous spam filters are beginning to use machine learning to better detect unwanted emails. However, despite all of its success in these other fields, machine learning is rarely used in network intrusion detection systems. 

This project will be focussed around the propositions set forward by Robin Sommer and Vern Paxson in their paper "Outsite the Closed World: On Using Machine Learnign For Network Intrusion Detection". Sommer and Paxson suggest that the reason for limited "real world" implementations of anomaly detection systems using machine learning is due to a special set of complications that network intrusion detection systems must deal with. These complications will be explored then reaffirmed or rejected through the process of creating a simple machine learning anomaly detection system at the Internet Layer of the TCP/IP Stack.

Paper to be used for topic: http://www.icir.org/robin/papers/oakland10-ml.pdf
Paper to be used for creating anomaly detection system: https://www.usenix.org/legacy/event/sysml07/tech/full_papers/ahmed/ahmed.pdf