Malware Detection and Classification by Graph Neural Network
Hsiao-Chung Lin, Ping Wang, Wen‐Hui Lin, Yu-Hsiang Lin, Jiahong Chen · 2023
In recent years, information security has received increasing attention. Every year, there are tens of millions of computers in the world infected by malware, which causes huge losses. Malware uses many methods, including viruses, worms, and trojans to infiltrate computer systems. Malware accesses important and personal data, damages the operating and network system, or performs various improper activities. It causes huge losses to businesses or individuals. Thus, we developed a malware detection and classification method using a graph neural network (GNN) and deep learning. Important behavioral features of the malware and malware variants were determined using a developed model with GCN, and the malware and their variants were classified. The Cuckoo Sandbox logs were also employed. Ransomware downloaded from the Bazaar Database was detected and analyzed with the GCN-based model. The evaluation metrics were accurate indicating that the GCN-based model detected ransomware effectively.