http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
악성코드의 이미지화 방법이 딥러닝 기반의 악성코드 분류 성능에 미치는 영향 분석
송무준(Mujun Song),이종관(Jongkwan Lee) 육군사관학교 화랑대연구소 2021 한국군사학논집 Vol.77 No.1
In this paper, we present the effect of the interpolation method and image size, which are an essential process for pre-processing in deep learning, on the performance of deep learning-based malware classification. The classifier uses a deep learning model to classify malware, and the visualized malware samples are used as train and test datasets. The malware images should be the same size to be used as input data in the deep learning model. Then, the malware size is not the same, so that resizing of images is necessary. When the image is resized, the image features can be distorted depending on the interpolation methods and the target size of the image. We conduct experiments to understand the impact on the deep learning model’s classification performance under various conditions. The results of the experiments can guide for selecting interpolation method and image size for the deep learning-based malware classifier.