The Application of Convolutional Neural Network in Malware Images Classification
Shiyu Wang, Zehao Li, Xiaotian Zhao · Advances in Social Science, Education and Humanities Research/Advances in social science, education and humanities research · 2022
Malicious software is a fundamental challenge to information security, which can hijack browsers, force software installation, automatically pop-up ads on web pages, and even support intelligence gathering and destructive cyberattacks.There are always various malicious software and malicious programs on both computers and mobile phones, which have a bad influence on society and people's life.It is important to find a way to recognize them and clean them up.Most new malware is a variant of known malware samples, which can be divided into different types so that each of the same types of malwares has highly similar behaviour characteristics.Therefore, these shared characteristics between malicious samples belonging to the same type can be used to detect and classify unknown programs.Deep learning has achieved good effect in malware classification assignment that converts malware into grayscale images and facilitated the improvement of classification tasks, because models using deep learning convolutional neural network (CNN) can embrace images as input simply.Based on these conditions and combined with the related documents, this paper analyses the nature and mechanism of CNN to classify the current malwares and proposes some possible prospects of it.Finally, it is concluded that compared with ordinary machine learning, the convolutional neural network in malware images classification improves the accuracy of malware classification and reduces the time needed for classification.