Research On Malicious Code Classification Technology Based On Deep Learning

In today’s society,with the rapid development of network technology,there are more and more network security-related events.Nowadays,the number of malicious code is increasing with the continuous development of society,which seriously threatens the security of the Internet environment.With the attention paid by the society to network security,it is a key work to classify malicious code accurately and quickly.There are many methods to detect and classify malicious code,but usually the method of feature extraction is used.The binary file,ASM file or operation code of malicious code are extracted by feature extraction,and then some methods in deep learning and machine learning are used to classify and study it.This paper first describes the basic knowledge related to malicious code,and then describes the basic types of malicious code,as well as the characteristics of malicious code and its acquisition methods.In addition,it also introduces four feature classification methods,which pave the way for the following comparative experiments using these methods.Next is the main research on malicious code classification,which is divided into two parts.The first part is to extract the features of malicious code.The method used is GLCM method based on grayscale image.First,the malicious code is compressed into a grayscale image,then the slightly improved GLCM method is used to extract the features of the malicious code grayscale image,and then the feature training set is used to train the four classification models mentioned above.The second part is the classification stage.About the classification method,this paper gives a classification method based on the improved deep forest,which improves the accuracy of classification by introducing the fitting quality feature and XGBoost into the deep forest.Finally,through the experimental comparison,the results show that the feature extraction method proposed in this paper can effectively improve the accuracy of classification,and the improved deep forest classification model also has better classification effect compared with other classification models.