Table 3 Results of the ablation study on PVAN model components, showing performance (CCC) after removing each component.

From: Interpretable deep learning reveals distinct spectral and temporal drivers of perceived musical emotion

Model configuration

Valence (CCC)

Arousal (CCC)

PVAN (Full Model)

0.67

0.73

Removed Psychological Constraint

0.58

0.62

Removed Temporal Attention (Transformer)

0.61

0.69

Removed Spectral Attention (CNN)

0.63

0.66

Back to article page