Data Expansion Using GAN to Improve Malware Classification by CNN

IR IR (HANDLE)

Bibliographic Information

Other Title
  • CNNによるマルウェア分類を改善するためのGANを用いたデータ拡張

Search this article

Description

In recent years, the spread of malware has become a threat to computer security. The existence of malware variants is a factor that has a significant impact on the increase in the number of malware discoveries. Research has been conducted to automatically and efficiently classify these variants of malware. With the development of deep learning, it is now used to classify subspecies of malware. A typical research is to convert malware into grayscale images and classify them using CNN (Convolutional neural network). In deep learning, a large amount of training data is used. However, when a new type of malware appears, it is difficult to collect a large amount of samples. In this research, we investigated whether it is possible to solve the problem of insufficient samples by generating training data for deep learning using GAN (Generative Adversarial Network) and extending the data. We conducted an experiment to see if the classification accuracy could be improved by expanding the data for training using GAN. We used datasets that consisted of 25 different malware families. It was confirmed that the classification accuracy was improved compared to that before the data expansion. From the results, it was found that the data expansion for malware classification using GAN was effective.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top