Pattern Based Security Using Machine Learning Techniques
I worked on a project titled "Pattern Based Security Using Machine Learning Techniques" during my final year of computer engineering. The project was an attempt to reduce false alarm rate for Intrusion Detection System by using a Machine Learning algorithm. Our aim was to design an IDS by using machine learning which could meet the demands of Reducing False Alarm Rate (FAR) along with higher detection rate. Many types of IDS already exist in the world that provide assistance at different stages of project development. But one problem that is commonly observed is that of a high False Alarm Rate. This software should be able to assist the developers in this department greatly along with the individual stage support.
The project was implemented using a semi-supervised method and used the J48 algorithm.
This method achieved an accuracy rate of 98.78%
and reduced the false alarm rate to 1.22%
Due to the modified algorithm that achieved such good results, two papers were published on the methods in the International Journal of Computer Applications as well as the Journal of Harmonized Research.