multi-scale-feature

This paper introduces the Multi-Scale Feature Attention Network (MSFAN), a deep learning architecture for classifying 12 types of polymers using THz Dual-Comb Spectroscopy, achieving 85.2% accuracy and outperforming state-of-the-art models.