Malware Detection Based on Image Conversion

Due to different malware and their variants appear every year, it is difficult to identify the virus. In traditional malware analysis methods, both static analysis methods and dynamic analysis methods may be limited due to related detection methods. The rise of artificial intelligence has allowed the classification of malware to be detected by artificial intelligence. Therefore, this paper uses artificial intelligence to create a classification model for malware images. We first use Discrete Cosine Transform (DCT) and Discrete Wavelet Transform (DWT) to further feature extraction of the visualized malware image, and use Generative Adversarial Network (GAN) to generate malware of the same family, which can increase the number of samples. Finally, a convolutional neural network (CNN) is used to create a classification model of malware images to achieve the purpose of classifying malware. The results show that the accuracy of the classification results after discrete cosine transform (DCT) can reach 99.46%. After the discrete wavelet transform (DWT), the accuracy of the classification results can reach up to 99.84%.