Classification Of Malicious Code Based On Grayscale Texture Fingerprint

Traditional malicious code classification technology faces a challenge to do with increasingly deformed malicious code, whose problems include having low detection rate and being time-consuming. In order to solve the above-mentioned problems, the traditional malicious code classification technology combines feature extraction of texture fingerprints on the image with the deep learning neural network classification method and proposes a malicious code classification method based on gray texture features. This method first converts the binary file of the malicious code into a grayscale image. It uses the grayscale co-occurrence matrix to extract the features in the malicious code, and the extracted features are in turn substituted into an improved neural network model. Classification training is then performed to reduce the probability of feature loss in the pooling layer. The experimental results indicate that the malicious code classification based on the above method have higher accuracy. Through the use of features extracted by the gray level co-occurrence matrix as output of the pooling layer, the learning time can also be effectively shortened hence making this method best for classification detection usage.