AudioFuse combines ViT and 1D CNN for robust phonocardiogram classification

Researchers have developed AudioFuse, a novel architecture that combines Vision Transformer (ViT) and 1D Convolutional Neural Network (CNN) models to classify phonocardiograms (PCGs). This hybrid approach simultaneously learns from spectral and temporal data representations of biomedical audio signals. AudioFuse achieved a competitive ROC-AUC of 0.8608 on the PhysioNet 2016 dataset, outperforming its individual baseline models. The architecture also demonstrated superior robustness to domain shift on the PASCAL dataset, maintaining a high ROC-AUC while a spectrogram-only baseline significantly degraded. AI

IMPACT Introduces a novel hybrid architecture for improved biomedical audio signal classification, potentially enhancing diagnostic tools.

RANK_REASON Research paper detailing a new hybrid AI architecture for biomedical audio classification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• 1d Cnn
• AudioFuse
• GitHub
• Md. Saiful Bari Siddiqui
• PASCAL dataset
• PhysioNet 2016 dataset
• Vít

• paper
• model release
• other

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