Research map / Label-efficient and robust learning / Recognition and long-tail
Learning from Ambiguously Labeled Face Images
Ching‐Hui Chen, Vishal M. Patel, Rama Chellappa
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 instance weighting (WMCar) to MCar (ablation) | MCar · Lost (16,8) | labeling error rate (%): 8.5 % -> 8.2 % | Improved |
| add Iterative Candidate Elimination to WMCar (ablation) | WMCar · Lost (16,8) | labeling error rate (%): 8.2 % -> 5.2 % | Improved |
| add instance weighting to MCar-ICE (WMCar-ICE) (ablation) | MCar-ICE · Lost (16,8) | labeling error rate (%): 8.0 % -> 5.2 % | Improved |
| add ICE to MCar without weighting (ablation) | MCar · Lost (16,8) | labeling error rate (%): 8.5 % -> 8.0 % | Improved |
| add instance weighting to MCar-SVM (ablation) | MCar-SVM · Labeled Yahoo! News | testing error rate (%): 14.5 % -> 13.6 % | Improved |
| add ICE and weighting to MCar-SVM (WMCar-ICE-SVM) (ablation) | MCar-SVM · Labeled Yahoo! News | testing error rate (%): outperforms MCar-SVM by 1.6% | Improved |
| add ICE without weighting to MCar-SVM (ablation) | MCar-SVM · Labeled Yahoo! News | testing error rate (%): 14.5 % -> 15.0 % | Hurt |
| MCar-SVM vs LR-SVM low-rank baseline | MCar-SVM · Labeled Yahoo! News | testing accuracy: 4.7% improvement on the testing accuracy | Improved |
| MCar-SVM vs MMS discriminative method | MCar-SVM · Labeled Yahoo! News | testing error rate (%): 14.3 % -> 14.5 % | No clear effect |
| use WMCar and WMCar-ICE under high ambiguity (labeling imbalance, imbalance factor 1.33 to 3.58) (ablation) | MCar · FIW(10b) inductive, varying epsilon | testing error rate: imbalance factor varies from 1.33 to 3.58 | Improved |
| add ICE to WMCar when more than 0.7 of instances are ambiguous (ablation) | WMCar · FIW(10b) inductive, varying alpha | testing error rate | Improved |
| add weighting to MCar on LFW under mild ambiguity model (ablation) | MCar · FIW(10b) inductive, varying alpha | testing error rate | No clear effect |
Closest papers by meaning
- Navigating Label Ambiguity for Facial Expression Recognition in the Wild (AAAI 2025 · similarity 0.87)
- RepFace: Refining Closed-Set Noise with Progressive Label Correction for Face Recognition (AAAI 2025 · similarity 0.85)
- The Pursuit of Human Labeling: A New Perspective on Unsupervised Learning (NeurIPS 2023 · similarity 0.85)
- Multi-Label Iterated Learning for Image Classification With Label Ambiguity (CVPR 2022 · similarity 0.85)
- Semantic-Aware Representation Blending for Multi-Label Image Recognition with Partial Labels (AAAI 2022 · similarity 0.85)
- Flaws can be Applause: Unleashing Potential of Segmenting Ambiguous Objects in SAM (NeurIPS 2024 · similarity 0.83)
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.
