phoenixml/Maldataset-2021

Maldataset2021 is a malware dataset that consists of 28 classes of malware, in which each class represents a malware family, and each sample gives a RGB 224x224 PNG file. The PNG files are transformed from the original binary malware files. The motivation of image transformation is to identify malware on the raw bytes of entire executable files (i.e., image), so that deep learning technologies such as CNN can be applied to malware classification, since CNN model has been demonstrated with its outstanding capability on image classification. In this view, we provide here a new dataset that contains the latest malware samples. The entire PNG files are split as, 70% for training and the remaining 30% for testing.

MIT