Research map / Vision-language and multimodal / Cross-modal learning
DHCM-CACL: Dynamic Hierarchical Cross-modal Mamba with Confidence-Adaptive Contrastive Learning for Multimodal Emotion Recognition
Baiqiang Wu, Yang Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| fuse facial expressions with EEG instead of EEG only (ablation) | DHCM-CACL · DEAP subject-independent valence | accuracy (%): 66.82 -> 72.89 | Improved |
| use DHCM instead of a standard transformer (ablation) | DHCM-CACL · DEAP subject-independent valence | accuracy (%): 67.15 -> 72.89 | Improved |
| use CACL instead of standard contrastive learning (ablation) | DHCM-CACL · DEAP subject-independent valence | accuracy (%): 69.82 -> 72.89 | Improved |
| use evidence layer instead of softmax classifier (ablation) | DHCM-CACL · DEAP subject-independent valence | accuracy (%): 71.05 -> 72.89 | Improved |
| add self-supervised pre-training (ablation) | DHCM-CACL · DEAP subject-independent valence | accuracy (%): 64.27 -> 72.89 | Improved |
| use dynamic state matrix instead of static fixed At (ablation) | DHCM · DEAP subject-independent valence | accuracy (%): 70.15 -> 72.89 | Improved |
| add forget gate (ablation) | DHCM · DEAP subject-independent valence | accuracy (%): 69.24 -> 72.89 | Improved |
| learned state matrices instead of random state matrices (ablation) | DHCM · DEAP subject-independent valence | accuracy (%): 67.58 -> 72.89 | Improved |
| set DHCM hierarchical levels L to 3 (ablation) | DHCM · DEAP valence and arousal | accuracy | Improved |
| use the method with only 20% annotations for fine-tuning | DHCM-CACL vs EEG2Rep · DEAP valence subject-independent | accuracy (%): 61.31 -> 65.72 | Improved |
| DHCM-CACL vs PhysioSync in subject-dependent valence | DHCM-CACL · DEAP subject-dependent valence | accuracy (%): 96.25 -> 95.21 | Hurt |
| DHCM-CACL vs EEG2Rep in subject-independent valence | DHCM-CACL · MAHNOB-HCI subject-independent valence | accuracy (%): 75.52 -> 75.12 | Hurt |
Closest papers by meaning
- Multi-to-Single: Reducing Multimodal Dependency in Emotion Recognition Through Contrastive Learning (AAAI 2025 · similarity 0.90)
- EMOE: Modality-Specific Enhanced Dynamic Emotion Experts (CVPR 2025 · similarity 0.89)
- Emotion-LLaMA: Multimodal Emotion Recognition and Reasoning with Instruction Tuning (NeurIPS 2024 · similarity 0.86)
- OV-MER: Towards Open-Vocabulary Multimodal Emotion Recognition (ICML 2025 · similarity 0.86)
- CARAT: Contrastive Feature Reconstruction and Aggregation for Multi-Modal Multi-Label Emotion Recognition (AAAI 2024 · similarity 0.86)
- MSAmba: Exploring Multimodal Sentiment Analysis with State Space Models (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
