Malicious file detection method based on deep neural network and gray-scale graph feature fusion

The widespread spread and continuous evolution of malicious files pose a serious threat to the security of computer systems. Therefore, developing efficient and accurate malicious file detection methods has important research value. This paper presents a malicious file detection method based on gray graph features and deep neural network (DNN). First, the file is transformed into an image representation by extracting the grayscale graph features of the file. Then, the pre-trained DNN model is used to learn and classify the gray graph features and determine whether the file is malicious. To verify the effectiveness of this method, we performed an experimental evaluation using high-precision datasets. The experimental results show that the proposed method performs well in malicious file detection, with high accuracy and low false alarm rate. This study provides a new approach to the field of malicious file detection and can be applied in a real network environment.