A Corpus of Encoded Malware Byte Information as Images for Efficient Classification
Ryan Frederick, Joseph Shapiro, Ricardo A. Calix · 2022
Malware has become an increasing problem in today’s world. In an attempt to combat this ever-changing problem, this research presents a machine-learning approach for analyzing malware using image classification. This paper discusses the idea of using image-based machine learning (ML) in malware analysis. Ways of applying this technique into the real-world are also presented and discussed. The highest accuracy achieved with the dataset used for training was 98.21%. Other datasets were also tested and achieved up to 93.22%. This shows that not only is this an effective method, but also that it can be used on other datasets and that a pre-trained model is more effective. The pre-trained model, dataset, and application are made available here to the wider research community.