악성코드 작성자 그룹 분류를 위한 특징 선택

초록

One of the greatest threats to modern cybersecurity is the existence of malware. The authors of malware are avoiding law enforcement and continuously producing new malware. Classification of malware by author groups can provide useful information for digital forensics. In this paper, we propose feature selection methods to identify more useful features for author group classification among a great number of features extracted from malware. As a result of the feature selection process, we confirm that the feature extraction time, classification model learning time, and classification time were significantly shortened. We also verify that decrease in accuracy is also minimized.

키워드