Project Title: "Elitism Enhancements for Genetic Algorithm Based Network Intrusion Detection System" (AceGA)
Please kindly obtain the published academic paper here: http://www.aicit.org/jcit/global/paper_detail.html?jname=JCIT&q=4159 or directly here: http://www.aicit.org/JCIT/ppl/JCIT4159PPL.pdf
To run this program, you need to install deap (it's a python library). optional: install pypy, it's a flavor of python intepreter (Much faster and better performance).
Network Security and Artificial intelligence are the two sub-fields in Computer Science that I fascinate and have much interest in. Thus, I would like to make a contribution to them by doing this project, as well as to learn and acquire knowledge as much as possible along the way.
By integrating Genetic Algorithms into Network Intrusion Detection Systems, my purpose could be realized.
Signature-based Network Intrusion Detection Systems (i.e. Profile-based, Rule-based) is able to detect and alert network administrators about the attack connections. But this NIDS approach suffers high false-negative result rates, because a signature-based Intrusion Detection System is only as good as its records of rules in its database. This is due to its inability to learn and identify new types of attacks.
A field in Artificial Intelligence: Evolutionary Algorithms, specifically Genetic Algorithms (GA), has an excellent approach to solve this problem. There are two general steps to apply GA to an NIDS, training stage and testing stage:
Training Stage: Firstly, GA can learn to make new rules (attack patterns) from existing identified attack records in a dataset.
Testing Stage: Secondly, the newly made rules can then be stored in an IDS database in order to strengthen the security level.