Entropy-Based Malware Detection Using One Dimensional CNN
Sota Okubo, Tomotaka Kimura, Jun Cheng · 2024
In this paper, we propose an entropy-based malware detection method using a one-dimensional (1D) convolutional neural network (CNN). In the proposed method, we calculate the entropy of software and classify this software into two groups according to the value of its entropy. Software with high entropy is considered to be obfuscated and has significantly different characteristics from software with low entropy. Then, we apply a 1D CNN to each group, using 1D vectors as input. We conducted experiments on a dataset of Portable Executable files, which showed that our proposed detection method improved detection performance.