A new weighted attribute fairness metric for SER quantifies each demographic attribute's absolute contribution to model bias and detects gender bias in HuBERT and WavLM on CREMA-D.
Inherent tradeoffs in learning fair representations
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Explainable Speech Emotion Recognition: Weighted Attribute Fairness to Model Demographic Contributions to Social Bias
A new weighted attribute fairness metric for SER quantifies each demographic attribute's absolute contribution to model bias and detects gender bias in HuBERT and WavLM on CREMA-D.