Estimation method of malware infection spreading with graph convolutional networks
Katsuki Uno, Tomotaka Kimura, Kouji Hirata · 2020
In this paper, we propose an estimation method of malware infection spreading with graph convolutional networks (GCN) for the epidemic model of future botnet malware. GCN is a machine learning technique that applies convolution operation for network structures. The infection spreading of malware depends on the network structures and infection source. The proposed method estimate the infection spreading level based on the network structures and the infection states of hosts, which are used as input data to GCN. In this paper, through numerical experiments, we show the effectiveness of the proposed method.