CNN Model to Classify Malware Using Image Feature
이미지 기능을 사용하여 맬웨어를 분류하는 CNN 모델

Espoir K. Kamundala, Chang Hoon Kim
2018 KIISE Transactions on Computing Practices  
Malware programs are common threats in the information and technology society. It has been proven that a number of developed malwares cripples the victim's computer as well as launching malicious attacks. Therefore, it is important to find a reasonable technical way to counter these attacks. Malware can be easily detected by checking whether a file has a malicious code inside the source code, if you detect a malicious code inside your content, then take an appropriate action by eliminating the
more » ... hreat. The first countermeasure to take is to delete the file or follow any other action defined by an Anti-malware software. After a file is infected, means it can be classified to its corresponding family based on its behavior in the infected system.. In this paper, we use Convolutional Neural Network to classify malware binaries using image features. Our work relies on the previously conducted research on malware visualization, whereby we used the dataset consisted of about 9,500 samples of 25 different malware familys. The built architecture achieved an accuracy of 98%.
doi:10.5626/ktcp.2018.24.5.256 fatcat:pemdhh62bne5vc6ckdufxcaiza