ENHANCE DEEPLEARNING MODEL FOR MALWARE DETECTION WITH A NEW IMAGE REPRESENTATION METHOD
Vo Khuong Linh, Nguyễn Việt Hùng, Tran Ngoc Anh, Duong Do Nhuan, Dinh Cong Hien · Journal of Science and Technology on Information security · 2024
In recent years, there has been an explosion in the number of new malware created by hackers worldwide. The large number of malware families causes certain difficulties for traditional malware detection methods. One of the recent research directions of interest is the application of artificial intelligence to solve problems. In this paper, we proposed a new method of representing malicious code as an image by arranging highly correlated bytes in close pixels on the image. Deep learning models are trained on self-built datasets and compare the performance of different image representation methods. Experimental results show that the proposed "serpentine" pixel arrangement method provides better results than other methods.