Feature Selection and Performance Improvement of Malware Detection System using Cuckoo Search Optimization and Rough Sets

The proliferation of malware is a severe threat to host and network-based systems. Design and evaluation of efficient malware detection methods is the need of the hour. Windows Portable Executable (PE) files are a primary source of windows based malware. Static malware detection involves an analysis of several PE header file features and can be done with the help of machine learning tools. In the design of efficient machine learning models for malware detection, feature reduction plays a crucial role. Rough set dependency degree is a proven tool for feature reduction. However, quick reduct using rough sets is an NP-hard problem. This paper proposes a hybrid Rough Set Feature Selection using Cuckoo Search Optimization, RSFSCSO, in finding the best collection of reduced features for malware detection. Random forest classifier is used to evaluate the proposed algorithm; the analysis of results proves that the proposed method is highly efficient.