Research map / Graph learning / Graph algorithms and clustering
Structure-Aware Face Clustering on a Large-Scale Graph With 107 Nodes
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 |
|---|---|---|---|
| add node-intimacy refinement after naive edge pruning (ablation) | STAR-FC on MS1M · MS1M part1_test | pairwise F-score: 82.50 -> 90.45 | Helped |
| use maximum rather than mean or minimum aggregation for node intimacy (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 89.35 / 88.06 -> 90.45 | Helped |
| add GCN graph parsing before node-intimacy refinement (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 80.00 -> 90.45 | Helped |
| replace node intimacy with Jaccard similarity (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 91.97 -> 88.39 | Hurt |
| use structure-preserved subgraph sampling instead of full-graph training (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 90.45 -> 91.21 | Helped |
| add cluster randomness to SPSS (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 91.21 -> 91.80 | Helped |
| add sample randomness to SPSS (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 91.21 -> 91.92 | Helped |
| add both cluster randomness and sample randomness to SPSS (ablation) | STAR-FC · MS1M part1_test | pairwise F-score: 91.21 -> 91.97 | Helped |
| increase WebFace42M training data from 0.4M to 20M images (ablation) | STAR-FC · WebFace42M, 12M unlabeled test images | pairwise F-score: 90.2 -> 95.1 | Helped |
| use STAR-FC rather than the strongest listed baseline | MS1M part1_test · MS1M part1_test (0.5M images) | pairwise F-score: 90.60 -> 91.97 | Helped |
| use STAR-FC rather than ARO | WebFace42M clustering · WebFace42M, 12M unlabeled test images; 4M training images | pairwise F-score: 76.05 -> 92.43 | Helped |
Closest papers by meaning
- Linkage Based Face Clustering via Graph Convolution Network (CVPR 2019 · similarity 0.92)
- Learning to Cluster Faces via Confidence and Connectivity Estimation (CVPR 2020 · similarity 0.92)
- TopoFR: A Closer Look at Topology Alignment on Face Recognition (NeurIPS 2024 · similarity 0.86)
- Local Graph Clustering with Noisy Labels (ICLR 2024 · similarity 0.86)
- THESAURUS: Contrastive Graph Clustering by Swapping Fused Gromov-Wasserstein Couplings (AAAI 2025 · similarity 0.85)
- Attribute-Missing Multi-view Graph Clustering (CVPR 2025 · 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.