Research map / Label-efficient and robust learning / Recognition and long-tail
MIC: Mining Interclass Characteristics for Improved Metric Learning
Karsten Roth, Biagio Brattoli, Bjorn Ommer
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 clustering-based auxiliary encoder without standardization (ablation) | margin loss baseline · CUB200-2011 | Recall@1: 62.9 -> 59.1 | Hurt |
| add clustering-based auxiliary encoder without standardization (ablation) | margin loss baseline · CARS196 | Recall@1: 80.0 -> 79.2 | Hurt |
| add class-wise standardization before clustering (ablation) | MIC · CUB200-2011 | Recall@1: 59.1 -> 64.9 | Improved |
| add class-wise standardization before clustering (ablation) | MIC · CARS196 | Recall@1: 79.2 -> 81.3 | Improved |
| add class-wise standardization before clustering (ablation) | MIC · Stanford Online Products | Recall@1: 71.9 -> 75.8 | Improved |
| add mutual information loss between encoders (ablation) | MIC · CUB200-2011 | Recall@1: 64.9 -> 66.1 | Improved |
| add mutual information loss between encoders (ablation) | MIC · CARS196 | Recall@1: 81.3 -> 82.6 | Improved |
| add mutual information loss between encoders (ablation) | MIC · Stanford Online Products | Recall@1: 75.8 -> 77.2 | Improved |
| apply MIC on margin loss | Margin* · CUB200-2011 | Recall@1: 62.9 -> 66.1 | Improved |
| apply MIC on margin loss | Margin* · In-Shop Clothes | Recall@1: 84.5 -> 88.2 | Improved |
| apply MIC on ProxyNCA | ProxyNCA* · CARS196 | Recall@1: 73.0 -> 75.9 | Improved |
| apply MIC on semihard triplet loss | Semihard* · CARS196 | Recall@1: 65.5 -> 70.5 | Improved |
Closest papers by meaning
- Rethinking Preventing Class-Collapsing in Metric Learning With Margin-Based Losses (ICCV 2021 · similarity 0.89)
- Divide and Conquer the Embedding Space for Metric Learning (CVPR 2019 · similarity 0.89)
- Deep Metric Learning with Self-Supervised Ranking (AAAI 2021 · similarity 0.89)
- Sharing Matters for Generalization in Deep Metric Learning (TPAMI 2020 · similarity 0.88)
- Improving Deep Metric Learning by Divide and Conquer (TPAMI 2021 · similarity 0.88)
- Energy Confused Adversarial Metric Learning for Zero-Shot Image Retrieval and Clustering (AAAI 2019 · similarity 0.88)
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.
