The EIHW-GLAM Deep Attentive Multi-model Fusion System for Cough-based COVID-19 Recognition in the DiCOVA 2021 Challenge

TL;DR

This paper introduces a deep attentive multi-model fusion system that combines various audio representations to improve COVID-19 detection from cough sounds, achieving significant performance gains in the DiCOVA 2021 challenge.

Contribution

It presents a novel multi-model fusion approach using attention mechanisms at feature and decision levels for cough-based COVID-19 recognition.

Findings

Attention-based feature fusion achieves highest AUC of 77.96%.

System outperforms official baseline by 8.05%.

Multi-representation fusion enhances detection accuracy.

Abstract

Aiming to automatically detect COVID-19 from cough sounds, we propose a deep attentive multi-model fusion system evaluated on the Track-1 dataset of the DiCOVA 2021 challenge. Three kinds of representations are extracted, including hand-crafted features, image-from-audio-based deep representations, and audio-based deep representations. Afterwards, the best models on the three types of features are fused at both the feature level and the decision level. The experimental results demonstrate that the proposed attention-based fusion at the feature level achieves the best performance (AUC: 77.96%) on the test set, resulting in an 8.05% improvement over the official baseline.