Research map / Label-efficient and robust learning / Self-supervised and contrastive
FairSSL: Fair Multimodal Self-Supervised Learning
Jiaee Cheong, Abtin Mogharabin, Paul Pu Liang, Hatice Gunes, Sinan Kalkan
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 subject-aware single pooling and regularization (M2) to VICReg | VICReg · D-Vlog | AGG_F: 0.04 -> 0.86 | Improved |
| add FairSSL M1 to VICReg | VICReg · MODMA | AGG_F: 0.53 -> 0.90 | Improved |
| add FairSSL M2 to VICReg | VICReg · MIMIC-III | AGG_F: 0.88 -> 0.95 | Improved |
| add single pooling (M3) vs no pooling (ablation) | VICReg · D-Vlog | AGG_F: 0.04 -> 0.82 | Improved |
| use double pooling instead of single pooling (M3) (ablation) | FairSSL · D-Vlog | AGG_F: 0.82 -> 0.71 | Hurt |
| use double pooling instead of single pooling (M3) (ablation) | FairSSL · D-Vlog | accuracy: 0.60 -> 0.65 | Improved |
| add FairSSL M3 to CoMM | CoMM · D-Vlog | AGG_F: 0.47 -> 0.72 | Improved |
| add FairSSL M2 to FOCAL | FOCAL · D-Vlog | AGG_F: 0.10 -> 0.90 | Improved |
| add FairSSL M1 to FOCAL | FOCAL · D-Vlog | AGG_F: 0.10 -> 0.10 | No clear effect |
| add FairSSL M1 to QUEST | QUEST · D-Vlog | AGG_F: 0.17 -> 0.85 | Improved |
| add FairSSL M3 to DeCUR | DeCUR · D-Vlog | AGG_F: 0.06 -> 0.75 | Improved |
| add FairSSL M1 to VICReg on MIMIC-CXR | VICReg · MIMIC-CXR (gender) | accuracy: 0.74 -> 0.66 | Hurt |
Closest papers by meaning
- Rethinking Fair Representation Learning for Performance-Sensitive Tasks (ICLR 2025 · similarity 0.86)
- FairLLaVA: Fairness-Aware Parameter-Efficient Fine-Tuning for Large Vision-Language Assistants (CVPR 2026 · similarity 0.86)
- Asymmetric Reinforcing Against Multi-Modal Representation Bias (AAAI 2025 · similarity 0.86)
- FairMT: Fairness for Heterogeneous Multi-Task Learning (NeurIPS 2026 · similarity 0.86)
- Amplifying Prominent Representations in Multimodal Learning via Variational Dirichlet Process (NeurIPS 2025 · similarity 0.85)
- CLIP the Bias: How Useful is Balancing Data in Multimodal Learning? (ICLR 2024 · 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.
