New hybrid CNN-ViT framework enhances malware classification accuracy

Researchers have developed ThreatVisionAI, a novel hybrid framework for classifying malware families using image-based analysis. This system combines a Convolutional Neural Network (CNN) with a Vision Transformer (ViT) to extract a comprehensive set of features, including spatial, frequency-domain, and global relational information. The framework demonstrated strong performance on the Malimg dataset, achieving 98.01% accuracy and a weighted F1 score of 0.9742, with particular effectiveness in distinguishing visually similar or minority malware families. AI

IMPACT This hybrid approach could improve the detection of sophisticated and novel malware by leveraging advanced AI architectures.

RANK_REASON The cluster contains an academic paper detailing a new technical framework for malware classification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• CNN
• Malimg dataset
• Prashanth BusiReddyGari
• ThreatVisionAI
• ViT

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →