Feature set reduction for the detection of packed executables

Abstract

Emerging sophisticated malware utilises obfuscation to circumvent detection. This is achieved by using packers to disguise their malicious intent. In this paper a novel malware detection method for detecting packed executable files using entropy analysis is proposed. It utilises a reduced feature set of variables to calculate an entropy score from which classification can be performed. Competitive analysis with state-of-the-art reveals an increase in classification accuracy.