On the Malware Detection Problem: Challenges & Novel Approaches

Abstract

Many solutions to detect malware have been proposed over time, but effective and efficient malware detection still remains an open problem. In this work, I take a look at some malware detection challenges and pitfalls to contribute towards increasing system’s malware detection capabilities. I propose a new approach to tackle malware research in a practical but still scientific manner and leverage this approach to investigate four issues: (i) the need for understanding context to allow proper detection of localized threats; (ii) the need for developing better metrics for AntiVirus (AV) evaluation; (iii) the feasibility of leveraging hardware-software collaboration for efficient AV implementation, and (iv) the need for predicting future threats to allow faster incident responses.