Publication Date

Fall 2011

Degree Type

Master's Project

Degree Name

Master of Science (MS)


Computer Science


Current anti-virus techniques include signature based detection, anomaly based detection, and machine learning based virus detection. Signature detection is the most widely used approach. Metamorphic malware changes its internal structure with each infection. Metamorphism provides one of the strong known methods for evading malware detection. In this project, we consider metamorphic virus detection based on a directed graph obtained from executable files. We compare our detection results with a previously developed and highly successful technique based on hidden Markov models.