A Malware Detection Method Based on Machine Learning and Ensemble of Regression Trees

In the context of the current large number of malicious codes, the detection and protection of malicious codes is particularly important. In recent years, a method of using deep learning to detect malicious code has emerged. Thus, in this paper, we propose a new detection method that converts binary files of malicious code into decimal arrays and use 1-D CNN to perform classification and recognition. Aiming at the imbalance in the number of code families, we choose xgboost, which performs well in the classification prediction competition. We conduct experiments on 9,458 malware samples from 25 different malware families in the Vision Research Lab. The experimental results show that our classification prediction reaches 97% accuracy.