Malware Classification Based on Multilayer Perception and Word2Vec for IoT Security

With the construction of smart cities, number Internet Things (IoT) devices is growing rapidly, leading to an explosive growth malware designed for IoT devices. These pose a serious threat security The traditional classification methods mainly rely on feature engineering. To improve accuracy, large different types features will be extracted from files in these methods. That brings high complexity classification. solve issues, method based Word2Vec and Multilayer Perception (MLP) proposed this article. First, one sample, used calculate word vector all bytes binary file instructions assembly file. Second, we combine vectors into 256x256x2-dimensional matrix. Finally, deep learning network structure MLP train model. Then model classify testing samples. experimental results prove that has accuracy 99.54%.