Research map / Label-efficient and robust learning / Recognition and long-tail
Contrastive Adversarial Learning for Person Independent Facial Emotion Recognition
Daeha Kim, Byung Cheol Song
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 contrastive adversarial learning with ResNet18 | FER on AffectNet · AffectNet | RMSE (valence): 0.2668 -> 0.2186 | Improved |
| add contrastive adversarial learning with ResNet18 | FER on AFEW-VA · AFEW-VA | RMSE (valence): 0.24 -> 0.17 | Improved |
| add contrastive adversarial learning with ResNet18 | FER on Aff-Wild · Aff-Wild | CCC (arousal): 0.31 -> 0.56 | Improved |
| replace the CAF critic loss with generic adversarial learning (ablation) | ResNet18 FER on AffectNet · AffectNet | RMSE (valence): 0.2186 -> 0.3687 | Hurt |
| replace the CAF critic loss with FisherGAN (ablation) | ResNet18 FER on AffectNet · AffectNet | RMSE (valence): 0.2186 -> 0.2221 | Hurt |
| replace the CAF critic loss with SphereGAN (ablation) | ResNet18 FER on AffectNet · AffectNet | RMSE (valence): 0.2186 -> 0.2467 | Hurt |
| replace the CAF critic loss with GramNet (ablation) | ResNet18 FER on AffectNet · AffectNet | RMSE (valence): 0.2186 -> 0.2251 | Hurt |
| use a constant instead of the adaptive margin interval (ablation) | ResNet18 CAF on AffectNet · AffectNet | RMSE (valence): 0.2186 -> 0.2289 | Hurt |
| use a constant instead of the adaptive margin interval (ablation) | ResNet18 CAF on AffectNet · AffectNet | RMSE (arousal): 0.1873 -> 0.2082 | Hurt |
| use a linear model instead of the adaptive margin interval (ablation) | ResNet18 CAF on AffectNet · AffectNet | RMSE (valence): 0.2186 -> 0.2201 | Hurt |
| use a linear model instead of the adaptive margin interval (ablation) | ResNet18 CAF on AffectNet · AffectNet | RMSE (arousal): 0.1873 -> 0.1982 | Hurt |
| replace Otsu grouping with k-means grouping (ablation) | CAF on AffectNet · AffectNet | RMSE: about 0.05 | No clear effect |
Closest papers by meaning
- A Personalized Affective Memory Model for Improving Emotion Recognition (ICML 2019 · similarity 0.86)
- Emotion-Aware Multi-View Contrastive Learning for Facial Emotion Recognition (ECCV 2022 · similarity 0.86)
- Attack Can Benefit: An Adversarial Approach to Recognizing Facial Expressions under Noisy Annotations (AAAI 2023 · similarity 0.85)
- PerReactor: Offline Personalised Multiple Appropriate Facial Reaction Generation (AAAI 2025 · similarity 0.83)
- EmoNet-Face: An Expert-Annotated Benchmark for Synthetic Emotion Recognition (NeurIPS 2025 · similarity 0.82)
- Multi-to-Single: Reducing Multimodal Dependency in Emotion Recognition Through Contrastive Learning (AAAI 2025 · similarity 0.82)
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.
