The EIHW-GLAM Deep Attentive Multi-model Fusion System for Cough-based COVID-19 Recognition in the DiCOVA 2021 Challenge
This work addresses COVID-19 diagnosis using cough sounds, but it is incremental as it builds on existing methods for audio-based medical recognition.
The authors tackled COVID-19 detection from cough sounds by proposing a deep attentive multi-model fusion system, which achieved an AUC of 77.96% on the test set, improving over the baseline by 8.05%.
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.