Adding patient-clustering losses to a pretrained audio transformer improves ICBHI lung sound scores by about 1.3 points.
Performance on the ICBHI Dataset We evaluated our method on the ICBHI dataset using the of- ficial 60/40 (train/test) split
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Patient-Aware Feature Alignment for Robust Lung Sound Classification:Cohesion-Separation and Global Alignment Losses
Adding patient-clustering losses to a pretrained audio transformer improves ICBHI lung sound scores by about 1.3 points.