Research map / Graph learning / Graph algorithms and clustering
Graph Matching with Bi-level Noisy Correspondence
Yijie Lin, Mouxing Yang, Jun Yu, Peng Hu, Changqing Zhang, Xi Peng
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 momentum distillation to the quadratic contrastive loss (ablation) | COMMON · Pascal VOC | mean matching accuracy (%): 81.77 -> 82.67 | Improved |
| add momentum distillation to the quadratic contrastive loss (ablation) | COMMON · SPair-71k | mean matching accuracy (%): 83.94 -> 84.54 | Improved |
| add within-graph and cross-graph consistency regularizers (ablation) | COMMON · Pascal VOC | mean matching accuracy (%): 81.95 -> 82.67 | Improved |
| add within-graph and cross-graph consistency regularizers (ablation) | COMMON · SPair-71k | mean matching accuracy (%): 83.83 -> 84.54 | Improved |
| use proposed quadratic loss instead of prior quadratic regularization with InfoNCE (ablation) | COMMON · Pascal VOC | mean matching accuracy (%): 81.67 -> 82.67 | Improved |
| add robust quadratic terms on top of plain InfoNCE (ablation) | COMMON · SPair-71k | mean matching accuracy (%): 83.38 -> 84.54 | Improved |
| vary distillation temperature alpha from 0.1 to 0.9 (ablation) | COMMON · Pascal VOC | mean matching accuracy (%): 82.34 -> 82.41 (range 82.34-82.69 for alpha 0.1-0.9) | No clear effect |
| turn on distillation (alpha 0 -> 0.4) (ablation) | COMMON · Pascal VOC | mean matching accuracy (%): 81.90 -> 82.67 | Improved |
| apply COMMON to hard-viewpoint image pairs | COMMON · SPair-71k (hard viewpoint pairs) | matching accuracy (%): 73.6 -> 76.4 | Improved |
| apply COMMON to easy-viewpoint image pairs | COMMON · SPair-71k (easy viewpoint pairs) | matching accuracy (%): 86.5 -> 86.6 | No clear effect |
| add momentum distillation under synthetic keypoint noise | COMMON · Willow Object synthetic noise 0-0.5 | matching accuracy | Improved |
Closest papers by meaning
- Learning With Partial and Noisy Correspondence in Graph Matching (TPAMI 2026 · similarity 0.91)
- Deep Learning of Partial Graph Matching via Differentiable Top-K (CVPR 2023 · similarity 0.87)
- Combinatorial Learning of Robust Deep Graph Matching: An Embedding Based Approach (TPAMI 2020 · similarity 0.87)
- Multi-Graph Matching via Affinity Optimization with Graduated Consistency Regularization (TPAMI 2016 · similarity 0.87)
- Return to Basics: Very Simple Graph Contrastive Learning via Noise Cancellation Principle (NeurIPS 2026 · similarity 0.86)
- Learning Partial Graph Matching via Optimal Partial Transport (ICLR 2025 · similarity 0.86)
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.
