Malware Detection Framework Using PCA Based ANN

Rami, Khyati; Desai, Vinod

Different kinds of computer threats exist to damage the computer system, and Malicious programs is one of them. Internet can be the main source to spread some threats. Experts continuously detect those which can slow down the system, or totally damage it. Malware creators have always been a step ahead. To detect malware threat, there are two basic approaches, based on signature and heuristic. For accurate and efficient result of malware detection there are detection techniques based on heuristic method. Polymorphic malwares are growing day by day and heuristic method is combined with machine learning to get more precise and effective detection. Malware detection system using data mining and machine learning methods have been proposed by many researchers to detect known and unknown malware. In this paper we present the ideas behind our malware detection framework by PCA based ANN to detect known and unknown malware. To design the proposed framework we have used MATLAB GUI.ANN is used to detect the presence of malware in CSDMC2019 API dataset. The computational time for ANN classifier is less than 0.2 s compared to NB classifier which has a computational time of 0.82 s.