Research map / Label-efficient and robust learning / Recognition and long-tail
Adaptive Deep Disturbance-Disentangled Learning for Facial Expression Recognition
Delian Ruan, Rongyun Mo, Yan Yan, Si Chen, Jing‐Hao Xue, Hanzi Wang
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 attention blocks (Se subnetwork) (ablation) | Baseline · MMI | accuracy (%): 79.16 -> 81.22 | Improved |
| add multi-level attention mechanism (ablation) | Baseline at · MMI | accuracy (%): 81.22 -> 81.47 | Improved |
| add disturbance subnetwork Sd (gender) (ablation) | Baseline mat · MMI | accuracy (%): 81.47 -> 82.88 | Improved |
| add pose to the disturbing factors of Sd (ablation) | DDL gar&id&il · MMI | accuracy (%): 84.25 -> 83.56 | Hurt |
| add adaptive disturbance feature learning (ADFL) (ablation) | DDL gar&id&il&p · MMI | accuracy (%): 1.84% and 0.91% improvements (MMI, RAF-DB) | Improved |
| add mutual information neural estimator (MINE) (ablation) | DDL gar&id&il&p · MMI | accuracy (%): 83.56 -> 85.56 | Improved |
| combine ADFL and MINE (ablation) | DDL gar&id&il&p · MMI | accuracy (%): 2.57% and 1.17% (MMI, RAF-DB) | Improved |
| add DFEM prior knowledge for the disturbance subnetwork (ablation) | ADDL · MMI | accuracy (%): 83.13 -> 86.13 | Improved |
| train DFEM on Multi-PIE and RAF-DB instead of Multi-PIE only (ablation) | ADDL · MMI | accuracy (%): 85.00 -> 86.13 | Improved |
| pretrain backbone on MS-Celeb-1M instead of no pretraining (ablation) | ADDL · MMI | accuracy (%): 70.76 -> 86.13 | Improved |
| pretrain backbone on MS-Celeb-1M instead of ImageNet (ablation) | ADDL · RAF-DB | accuracy (%): 87.58 -> 89.34 | Improved |
| enable adversarial training (lambda1 0.0 to 1.0) (ablation) | ADDL · MMI | accuracy (%): 83.39 -> 86.13 | Improved |
Closest papers by meaning
- An Adversarial Neuro-Tensorial Approach for Learning Disentangled Representations (IJCV 2019 · similarity 0.83)
- Lifting Scheme-Based Implicit Disentanglement of Emotion-Related Facial Dynamics in the Wild (AAAI 2025 · similarity 0.82)
- Navigating Label Ambiguity for Facial Expression Recognition in the Wild (AAAI 2025 · similarity 0.82)
- Hypergraph-Guided Disentangled Spectrum Transformer Networks for Near-Infrared Facial Expression Recognition (AAAI 2024 · similarity 0.81)
- QCS:Feature Refining from Quadruplet Cross Similarity for Facial Expression Recognition (AAAI 2025 · similarity 0.80)
- Learning Adaptive Spatial Coherent Correlations for Speech-Preserving Facial Expression Manipulation (CVPR 2024 · similarity 0.80)
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.
