Malware Classification using Deep Learning Technique

The increasing dependency of humans on computers for data storage and the internet for connectivity has paved way for cybercriminals, who are leveraging on this growing number for their personal benefits. This has resulted in the creation of countless malwares with the sole aim of malicious attack. In this study, focus is on the use of deep learning technique for the classification of malware into their respective family or author of origin. The approach used in this study involves transforming malwares, obtained as Portable Executables, into their corresponding image representation. The images generated are then used as dataset for our model which uses transfer learning approach. Our implemented model was deemed successful as we were able to obtain a higher average classification accuracy of 98.8% when evaluated with other techniques from previous literature.