Efficient Windows malware identification and classification scheme for plant protection information systems.

Area of Science:

• Agricultural Informatics
• Cybersecurity
• Computer Vision

Background:

• Plant protection information systems integrate technology for pest monitoring and control.
• Cyberattacks on these systems are increasing due to sophisticated malware variants.
• Existing malware classification methods using convolutional neural networks (CNNs) face challenges with image size imbalance.

Purpose of the Study:

• To propose a novel malware identification and classification scheme for plant protection information terminals.
• To address the image size imbalance issue in malware classification.
• To enhance the security and efficiency of plant protection information systems against evolving cyber threats.

Main Methods:

• Implemented bicubic interpolation to reconstruct and standardize malware binary images.
• Utilized the Cycle-GAN model for data augmentation to balance malware family samples.
• Developed a CNN-based model for efficient malware classification.

Main Results:

• The proposed scheme effectively resolves malware image size imbalance.
• Data augmentation balanced sample distribution across malware families.
• Achieved high classification accuracy: 99.76% for RGB images and 99.62% for grayscale images on the BIG2015 dataset.

Conclusions:

• The integrated approach significantly improves malware classification efficiency and accuracy.
• This method enhances the security posture of plant protection information terminal systems.
• The findings contribute to the development of robust cybersecurity measures in agricultural informatics.