Research map / Label-efficient and robust learning / Recognition and long-tail
P2SGrad: Refined Gradients for Optimizing Deep Face Models
Check 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 |
|---|---|---|---|
| Replace class probabilities with cosine similarities in gradient lengths | P2SGrad · MegaFace | recognition accuracy: P2SGrad exceeds other compared cosine-based losses ... with every size of distractor | Helped |
| Replace class probabilities with cosine similarities in gradient lengths | P2SGrad · IJB-C 1:1 verification | TAR: 99.07% -> 99.03%; 97.75% -> 97.79%; 95.55% -> 95.58%; 92.13% -> 92.25%; 87.28% -> 87.84%; 82.15% -> 82.44%; 72.28% -> 73 | Mixed |
| Replace class probabilities with cosine similarities in gradient lengths | P2SGrad · LFW | recognition accuracy: 81.50 -> 91.25; 83.63 -> 91.25; 85.32 -> 91.25 | Helped |
| Use P2SGrad instead of NormFace, CosFace, or ArcFace | face model training · initial learning rate 10^{-1} | trainability | Helped |
Closest papers by meaning
- CosFace: Large Margin Cosine Loss for Deep Face Recognition (CVPR 2018 · similarity 0.89)
- AdaCos: Adaptively Scaling Cosine Logits for Effectively Learning Deep Face Representations (CVPR 2019 · similarity 0.87)
- Killing Two Birds With One Stone: Efficient and Robust Training of Face Recognition CNNs by Partial FC (CVPR 2022 · similarity 0.86)
- SphereFace2: Binary Classification is All You Need for Deep Face Recognition (ICLR 2022 · similarity 0.85)
- Mis-Classified Vector Guided Softmax Loss for Face Recognition (AAAI 2020 · similarity 0.85)
- HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere (ICLR 2025 · 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.