Multi-Scale Feature Attention Network for Polymer Classification using THz Dual-Comb Spectroscopy
Researchers have developed a new deep learning model called the Multi-Scale Feature Attention Network (MSFAN) specifically for classifying polymers using Terahertz Dual-Comb Spectroscopy (THz-DCS). This novel architecture incorporates feature gating, multi-scale convolutions, and attention mechanisms to effectively analyze the complex spectral data. MSFAN achieved an 85.2% classification accuracy, outperforming existing models and demonstrating the potential of AI in conjunction with THz-DCS for robust polymer identification. AI
IMPACT This research demonstrates a novel deep learning approach for improving accuracy in material classification tasks.