Malware Classification in Cloud Computing Using Transfer Learning

The adoption of cloud computing has revolutionized the way organizations handle their data and applications. However, this paradigm shift, with its shared infrastructure and remote accessibility, has also introduced big security concerns. One of the most important challenges is the growing threat of malware that may consume CPU, memory, and bandwidth of cloud resources. Consequently, there is an urgent need for malware detection and classification system. In this paper, we propose to use Convolutional Neural Network (CNN) based on three popular fine-tuning techniques to classify binary files of malwares in cloud computing. The experiments are applied on Malimg dataset, which contains grayscale images of 25 families of 9,339 malwares and ResNet50 transfer learning model reached a good accuracy with 98.29%.