Affect-Diff achieves 0.384 balanced accuracy on imbalanced CMU-MOSEI data by integrating causal re-weighting, beta-VAE compression, and a DDPM prior, detecting all six emotions where baselines fail on minorities.
Unbiased Missing-Modality Multimodal Learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Multimodal Emotion Recognition via Causal-Diffusion Bridge (Affect-Diff)
Affect-Diff achieves 0.384 balanced accuracy on imbalanced CMU-MOSEI data by integrating causal re-weighting, beta-VAE compression, and a DDPM prior, detecting all six emotions where baselines fail on minorities.