Research map / Label-efficient and robust learning / Recognition and long-tail
Mis-Classified Vector Guided Softmax Loss for Face Recognition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| apply MV-AM-Softmax-a instead of AM-Softmax | face recognition model · MegaFace | identification Rank-1 (%): 97.69 -> 98.00 | Helped |
| apply MV-AM-Softmax-a instead of AM-Softmax | face recognition model · Trillion-Pairs | verification TPR@FAR=1e-9 (%): 71.57 -> 75.93 | Helped |
| use adaptive rather than fixed mis-classified-vector re-weighting (ablation) | MV-AM-Softmax · MegaFace verification | TPR@FAR=1e-6 (%): 97.85 -> 98.31 | Helped |
| use adaptive rather than fixed mis-classified-vector re-weighting (ablation) | MV-AM-Softmax · Trillion-Pairs verification | TPR@FAR=1e-9 (%): 74.45 -> 75.93 | Helped |
| vary t across the tested range (ablation) | MV-Softmax · BLUFR, CALFW, and AgeDB | verification performance (%): 94.09 to 96.27; 95.29 to 95.63; 97.86 to 98.13 | Mixed |
| combine margin-based loss with mining by Focal weighting | face recognition model · LFW BLUFR TPR@FAR=1e-5 | TPR (%): 93.75 -> 92.81 | Hurt |
Closest papers by meaning
- Learning Towards The Largest Margins (ICLR 2022 · similarity 0.92)
- CosFace: Large Margin Cosine Loss for Deep Face Recognition (CVPR 2018 · similarity 0.91)
- Killing Two Birds With One Stone: Efficient and Robust Training of Face Recognition CNNs by Partial FC (CVPR 2022 · similarity 0.90)
- Loss Function Search for Face Recognition (ICML 2020 · similarity 0.90)
- ArcFace: Additive Angular Margin Loss for Deep Face Recognition (CVPR 2019 · similarity 0.89)
- Large-Margin Softmax Loss for Convolutional Neural Networks (ICML 2016 · similarity 0.89)
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.