Research map / Label-efficient and robust learning / Recognition and long-tail
MetricFormer: A Unified Perspective of Correlation Exploring in Similarity Learning
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 |
|---|---|---|---|
| add feature decoupling block to margin loss (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 63.1 -> 66.8 | Helped |
| add batch-wise correlation block on top of FDB (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 66.8 -> 68.7 | Helped |
| add feature-wise correlation block on top of FDB (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 66.8 -> 68.4 | Helped |
| add both BCB and FCB on top of FDB (full MetricFormer) (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 68.7 -> 69.2 | Helped |
| add feature-wise correlation block to FDB+BCB (ablation) | MetricFormer-M · CUB-200-2011 | NMI: 70.8 -> 71.0 | Helped |
| set sub-feature count k=3 instead of k=2 (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 65.8 -> 69.2 | Helped |
| set sub-feature count k=3 instead of k=4 (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 68.5 -> 69.2 | Helped |
| increase batch-wise blocks from T=1 to T=3 (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 67.7 -> 69.2 | Helped |
| increase batch-wise blocks from T=2 to T=3 (ablation) | MetricFormer-M · CUB-200-2011 | Recall@1: 68.4 -> 69.2 | Helped |
| add MetricFormer to margin loss | MetricFormer-M · CUB-200-2011 | Recall@1: 63.1 -> 69.2 | Helped |
| add MetricFormer to margin loss | MetricFormer-M · Cars196 | Recall@1: 79.9 -> 89.5 | Helped |
| add MetricFormer to ProxyAnchor loss | ProxyAnchor · CUB-200-2011 | Recall@1: 69.7 -> 74.4 | Helped |
Closest papers by meaning
- SLADE: A Self-Training Framework for Distance Metric Learning (CVPR 2021 · similarity 0.85)
- Deep Compositional Metric Learning (CVPR 2021 · similarity 0.85)
- Hierarchical Correlation Clustering and Tree Preserving Embedding (CVPR 2024 · similarity 0.84)
- Stabilizing Sample Similarity in Representation via Mitigating Random Consistency (ICML 2025 · similarity 0.84)
- T-MDML: Triplet-based Multiple Distance Metric Learning for Multi-Instance Multi-Label Classification with Label Correlation (AAAI 2025 · similarity 0.84)
- Volume-Aware Distance for Robust Similarity Learning (ICML 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.