MACSc Project Presentation - A Preliminary Analysis on Categorizing Malicious Android Applications
Who: Abdulazeez Sanni
Title: A Preliminary Analysis on Categorizing Malicious Android Applications
Dr. Nur Zincir-Heywood - Faculty of Computer Science (Supervisor)
Dr. Malcolm Heywood - Faculty of Computer Science (Reader)
In this research project, I have employed the use of different Machine Learning algorithms to classify / cluster malicious Android applications into predefined categories. To this end, I have employed J48 (C4.5) and Naive Bayes for classification and I have employed K-means and Expectation Maximization (EM) algorithms for clustering. Also, I evaluated the use of a combined classification and clustering approach for my analysis. In doing so, I aim to find the most effective approach in terms of analyzing different malwares into their respective behavioral categories. These approaches have been tested on approximately 4000 Android malicious applications, where the results are measured using different performance metrics.
Room 211, Goldberg Computer Science BuildingMACSc Project Presentation - A Preliminary Analysis on Categorizing Malicious Android Applications