Research map / Probabilistic and scientific modeling / Causal discovery
Towards Context-Aware Emotion Recognition Debiasing From a Causal Demystification Perspective via De-Confounded Training
Dingkang Yang, Kun Yang, Haopeng Kuang, Zhaoyu Chen, Yuzheng Wang, Lihua Zhang
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 CCIM causal intervention module | EmotiCon · EMOTIC (discrete, mAP) | mAP (%): 35.28 -> 39.13 | Improved |
| add CCIM | EMOT-Net, CAER-Net, GNN-CNN, CD-Net, EmotiCon · EMOTIC (discrete, mAP) | mAP (%): improve the mAP scores by 2.95%, 2.66%, 3.56%, 3.42%, and 3.85% | Improved |
| add CCIM | EmotiCon · CAER-S | accuracy (%): 88.65 -> 91.17 | Improved |
| add CCIM | five CAER models · GroupWalk | mAP (%): increased by 2.41%, 2.25%, 2.99%, 2.72%, and 3.73% | Improved |
| add CCIM | EMOT-Net and GNN-CNN · EMOTIC (VAD regression) | mean AAE (lower is better): 0.0571 -> 0.0559 and 0.0555 -> 0.0538 | Improved |
| add CCIM | CD-Net · EMOTIC (C-F1) | C-F1 (%): 13.29 -> 11.66 | Hurt |
| use random dictionary instead of context-prototype dictionary Z (ablation) | CCIM on EmotiCon · EMOTIC (discrete, mAP) | mAP (%): 35.12 -> 39.13 | Improved |
| use Places365 instead of ImageNet pretrained context extractor (ablation) | CCIM on EmotiCon · EMOTIC (discrete, mAP) | mAP (%): 37.48 -> 39.13 | Improved |
| add attention weight lambda_i (ablation) | CCIM on EmotiCon · EMOTIC (discrete, mAP) | mAP (%): 38.53 -> 39.13 | Improved |
| add prior probability P(z_i) (ablation) | CCIM on EmotiCon · CAER-S | accuracy (%): 90.06 -> 91.17 | Improved |
| add subject masking strategy (ablation) | CCIM on EmotiCon · GroupWalk | mAP (%): 67.79 -> 69.31 | Improved |
| swap K-Means++ for Dichotomous K-Means or K-Medoids (ablation) | CCIM on EmotiCon · EMOTIC (discrete, mAP) | mAP (%): 39.13 vs 39.11 and 39.16 | No clear effect |
Closest papers by meaning
- Robust Emotion Recognition in Context Debiasing (CVPR 2024 · similarity 0.93)
- Bridge Then Begin Anew: Generating Target-Relevant Intermediate Model for Source-Free Visual Emotion Adaptation (AAAI 2025 · similarity 0.86)
- CocoER: Aligning Multi-Level Feature by Competition and Coordination for Emotion Recognition (CVPR 2025 · similarity 0.86)
- Facial-R1: Aligning Reasoning and Recognition for Facial Emotion Analysis (AAAI 2026 · similarity 0.85)
- Customizing Visual Emotion Evaluation for MLLMs: An Open-vocabulary, Multifaceted, and Scalable Approach (ICLR 2026 · similarity 0.85)
- VidEmo: Affective-Tree Reasoning for Emotion-Centric Video Foundation Models (NeurIPS 2025 · similarity 0.85)
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.
