Research map / Vision-language and multimodal / Cross-modal learning
Decoupled Hierarchical Distillation for Multimodal Emotion Recognition
Yong Li, Yuanzhi Wang, Yi Ding, Shiqing Zhang, Ke Lu, Cuntai Guan
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 |
|---|---|---|---|
| add dictionary matching (DM) (ablation) | DHMD (FD+CA+GD) · CMU-MOSI | ACC7 (%): 45.6 -> 49.9 | Improved |
| add graph distillation (GD) (ablation) | DHMD (FD+CA+DM) · CMU-MOSI | ACC7 (%): 46.0 -> 49.9 | Improved |
| add GD without DM (ablation) | DHMD (FD+CA) · CMU-MOSI | ACC7 (%): 45.0 -> 45.6 | Improved |
| add DM without GD (ablation) | DHMD (FD+CA) · CMU-MOSI | ACC7 (%): 45.0 -> 46.0 | Improved |
| add cross-modal attention (CA) (ablation) | FD only · CMU-MOSI | ACC7 (%): 42.4 -> 45.0 | Improved |
| add feature decoupling (FD) (ablation) | CA+DM · CMU-MOSI | ACC7 (%): 44.8 -> 46.0 | Improved |
| add FD to plain baseline (ablation) | baseline (no FD/CA/GD/DM) · CMU-MOSI | ACC7 (%): 37.9 -> 42.4 | Improved |
| add DM (ablation) | DHMD (FD+CA+GD) · CMU-MOSEI | ACC7 (%): 54.5 -> 55.3 | Improved |
| add GD (ablation) | DHMD (FD+CA+DM) · CMU-MOSEI | ACC7 (%): 54.1 -> 55.3 | Improved |
| add feature decoupling to unimodal acoustic model (ablation) | DHMD baseline · CMU-MOSEI | A only ACC2 (%): 53.4 -> 64.9 | Improved |
| add GD and DM to MulT (ablation) | MulT · CMU-MOSI (unaligned) | ACC7 (%): 39.1 -> 40.7 | Improved |
| add feature decoupling on top of MulT w/ GD & DM (i.e. DHMD) (ablation) | MulT (w/ GD & DM) · CMU-MOSI (unaligned) | ACC7 (%): 40.7 -> 41.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- EMOE — EMOE: Modality-Specific Enhanced Dynamic Emotion Experts (CVPR 2025) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OV-MER: Towards Open-Vocabulary Multimodal Emotion Recognition (ICML 2025 · similarity 0.89)
- Multi-to-Single: Reducing Multimodal Dependency in Emotion Recognition Through Contrastive Learning (AAAI 2025 · similarity 0.89)
- EMOE: Modality-Specific Enhanced Dynamic Emotion Experts (CVPR 2025 · similarity 0.88)
- AffectGPT: A New Dataset, Model, and Benchmark for Emotion Understanding with Multimodal Large Language Models (ICML 2025 · similarity 0.88)
- Emotion-LLaMA: Multimodal Emotion Recognition and Reasoning with Instruction Tuning (NeurIPS 2024 · similarity 0.88)
- A Unified Self-Distillation Framework for Multimodal Sentiment Analysis with Uncertain Missing Modalities (AAAI 2024 · similarity 0.87)
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.
