Abstract:

Malware classification and privacy-aware mobile application ("App" for short) recommendation are two critical problems in cyber security. Malware are responsible for a large number of malicious activities in the cyber space, such as spamming, identity theft, and DDoS (Distributed Denial of Service) attacks. The voluminous malware variants that appear in the Internet have posed severe threats to its security. Behind the sheer number of malware instances, however, lies the fact that a large number of them came from the same origins. New challenges come with the exponentially growing markets of mobile Apps. Public concerns about privacy issues with online activity and mobile phones are also elevating, demanding a mobile environment with more respect to user's privacy. In this talk, I will first talk about automated malware classification via discriminant distance learning on malware structural information. Our generic framework that exploits the rich structural information inherent in malware programs for accurate automated malware classification and the corresponding algorithms will be presented. Experimental results on real-world dataset demonstrate that our approach is able to classify malware instances with high accuracy.
Then, I will talk about privacy-aware personalized mobile App recommendation. Our Poisson matrix factorization model that captures the trade-off between mobile App functionality and user privacy preference for App recommendation will be presented. Experimental results on the crawled real-world dataset demonstrate that our method consistently and substantially outperforms the state-of-the-art approaches, which implies the importance of user privacy preference on personalized App recommendation.

Biography:

Host:

Dr. Xiaoming Liu