Adversarial Graph Contrastive Learning with Topology-Aware Multimodal Fusion for Cross-Domain Emotion Recognition
Accepted
Research approach
Combined topology-aware modality fusion, graph contrastive learning and adversarial domain adaptation for subject and session shifts.
View full-size figure
View full-size figure
Results
SEED-IV accuracy: 76.1 ± 7.4% across subjects and 83.8 ± 8.3% across sessions. Evaluation is transductive and uses unlabeled target features.
| Accuracy (%) | Cross-subject | Cross-session |
|---|---|---|
| SEED-IV | 76.1 ± 7.4 | 83.8 ± 8.3 |
| SEED-V | 75.6 ± 6.3 | 82.5 ± 9.1 |
InterpretationTransductive domain adaptation uses unlabeled target features. Target emotion labels are excluded from training and model selection. SEED-IV cross-subject/session accuracy: 76.1 ± 7.4% / 83.8 ± 8.3%.
All publications