Research map / Label-efficient and robust learning / Semi-supervised learning
Deep Semi-Supervised Metric Learning With Mixed Label Propagation
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 |
|---|---|---|---|
| Use edge-removed propagated-neighbor labels to estimate hard-negative edges, then apply mixed label propagation | Semi-supervised metric learning · aircraft (family), cars, CIFAR100, CUB-200-2011, Stanford Dogs, Oxford-IIIT Pets | NMI: aircraft (family): 32.798 -> 33.837; cars: 34.755 -> 36.397; CIFAR100: 39.150 -> 39.377; cub: 55.824 -> 57.780; dogs: 71 | Helped |
| Use edge-removed propagated-neighbor labels to estimate hard-negative edges | Label propagation · Aircraft, cars, CIFAR100, CUB, dogs, pets, traffic sign, vgg flowers | accuracy: aircraft (family): 0.195 ± 0.006 -> 0.207 ± 0.005; aircraft (manufacturer): 0.189 ± 0.019 -> 0.218 ± 0.020; aircraft (va | Helped |
| Identify nearest-neighbor but not mutual-nearest-neighbor pairs as negative edges (MoM) (ablation) | Label propagation · Datasets in Table 3 | accuracy | Hurt |
| Use LP predictions directly, rather than first-order-neighbor predictions, to calculate dissimilarity weights (Control) (ablation) | Label propagation · Datasets in Table 3 | accuracy | Mixed |
| Replace the proposed mixed LP dissimilarity loss with the LPDR dissimilarity loss (ablation) | Mixed label propagation · Oracle-provided close negative edges | accuracy | Hurt |
| Vary β in the mixed LP dissimilarity loss (ablation) | Mixed label propagation · CUB-200-2011, cars, Stanford Dogs | accuracy | No clear effect |
| Vary λ in the dissimilarity-weight calculation (ablation) | Hard-negative mining · CUB-200-2011, cars, Stanford Dogs | accuracy | Mixed |
Closest papers by meaning
- Implicit Relative Labeling-Importance Aware Multi-Label Metric Learning (AAAI 2025 · similarity 0.87)
- Semi-Supervised Metric Learning: A Deep Resurrection (AAAI 2021 · similarity 0.87)
- Self-Taught Metric Learning Without Labels (CVPR 2022 · similarity 0.86)
- Embedding Transfer With Label Relaxation for Improved Metric Learning (CVPR 2021 · similarity 0.86)
- Deep Metric Learning Beyond Binary Supervision (CVPR 2019 · similarity 0.86)
- T-MDML: Triplet-based Multiple Distance Metric Learning for Multi-Instance Multi-Label Classification with Label Correlation (AAAI 2025 · similarity 0.84)
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.