Research map / Label-efficient and robust learning / Recognition and long-tail
Multi-Similarity Loss With General Pair Weighting for Deep Metric Learning
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 |
|---|---|---|---|
| use multi-similarity mining and weighting rather than ABE | image retrieval · CUB200 | Recall@1: 60.6 -> 65.7 | Helped |
| use multi-similarity mining and weighting rather than ABE | image retrieval · Cars-196 | Recall@1: 85.2 -> 84.1 | Hurt |
| use multi-similarity mining and weighting rather than ABE | image retrieval · In-Shop | Recall@1: 87.3 -> 89.7 | Helped |
| use multi-similarity mining and weighting rather than ABE | image retrieval · SOP | Recall@1: 76.3 -> 78.2 | Helped |
| add MS mining to Binomial weighting (ablation) | Binomial on Cars-196 · Cars-196 | Recall@1: 71.9 -> 74.6 | Helped |
| add MS mining to LiftedStruct* weighting (ablation) | LiftedStruct* on Cars-196 · Cars-196 | Recall@1: 69.7% -> 72.2% | Helped |
| add Similarity-N weighting to Binomial (ablation) | Binomial on Cars-196 · Cars-196 | Recall@1: 71.9% -> 73.2% | Helped |
| add Similarity-S weighting to MS mining (ablation) | MS mining on Cars-196 · Cars-196 | Recall@1: 67.0% -> 74.6% | Helped |
| add Similarity-P-based mining to Binomial (ablation) | Binomial on Cars-196 · Cars-196 | Recall@1: 71.9% -> 74.6% | Helped |
| combine Binomial and LiftedStruct* weighting directly (ablation) | Cars-196 metric learning · Cars-196 | Recall@1: 71.9 -> 70.4 | Hurt |
| increase MS embedding size from 64 to 512 (ablation) | MS · Cars-196 | Recall@1: 77.3 -> 84.1 | Helped |
Closest papers by meaning
- Ranked List Loss for Deep Metric Learning (CVPR 2019 · similarity 0.89)
- Improved Deep Metric Learning with Multi-class N-pair Loss Objective (NeurIPS 2016 · similarity 0.89)
- Multi-level Distance Regularization for Deep Metric Learning (AAAI 2021 · similarity 0.87)
- Moving in the Right Direction: A Regularization for Deep Metric Learning (CVPR 2020 · similarity 0.87)
- Proxy Anchor Loss for Deep Metric Learning (CVPR 2020 · similarity 0.87)
- Mean Field Theory in Deep Metric Learning (ICLR 2024 · 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.