Research map / Learning theory and algorithms / Matrix and tensor methods
Scalable Metric Learning via Weighted Approximate Rank Component Analysis
Cijo Jose, François Fleuret
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 |
|---|---|---|---|
| use harmonic rank weighting instead of uniform weighting (ablation) | WARCA · person re-ID datasets | rank-1 accuracy | Improved |
| replace the Frobenius norm regularizer with the AON regularizer (ablation) | WARCA · VIPeR (LOMO features) | rank-1 accuracy | Improved |
| add AON regularizer instead of Frobenius norm at low regularization strength (ablation) | WARCA · VIPeR (LOMO features) | rank-1 accuracy | No clear effect |
| use AON regularizer: robustness to SGD step size (ablation) | WARCA · VIPeR (LOMO features) | rank-1 accuracy | Improved |
| use chi2 kernel instead of linear kernel in WARCA | WARCA · CUHK03 | rank-1 accuracy: 62.12 -> 78.38 | Improved |
| use chi2 kernel instead of linear kernel in WARCA | WARCA · VIPeR | rank-1 accuracy: 20.86 -> 37.47 | Improved |
| use chi2 kernel instead of linear kernel in WARCA | WARCA · CUHK01 | rank-1 accuracy: 39.30 -> 58.34 | Improved |
| use WARCA instead of MLAPG with LOMO features | metric learning for re-ID · CUHK03 (LOMO features) | rank-1 accuracy: 57.96 -> 78.38 | Improved |
| use WARCA instead of MLAPG with LOMO features | metric learning for re-ID · CUHK01 (LOMO features) | rank-1 accuracy: 64.24 -> 65.64 | Improved |
| use WARCA instead of MLAPG with LOMO features | metric learning for re-ID · VIPeR (LOMO features) | rank-1 accuracy: 40.73 -> 40.22 | Hurt |
| use WARCA instead of LFDA-chi2 on small datasets | metric learning for re-ID · VIPeR | rank-1 accuracy: 36.77 -> 37.47 | No clear effect |
| use WARCA instead of LFDA-chi2 on small datasets | metric learning for re-ID · PRID450s | rank-1 accuracy: 24.31 -> 24.58 | No clear effect |
Closest papers by meaning
- Efficient Online Local Metric Adaptation via Negative Samples for Person Re-Identification (ICCV 2017 · similarity 0.85)
- Embedding Deep Metric for Person Re-identification: A Study Against Large Variations (ECCV 2016 · similarity 0.84)
- Efficient Multiple Instance Metric Learning Using Weakly Supervised Data (CVPR 2017 · similarity 0.84)
- Graph Sampling Based Deep Metric Learning for Generalizable Person Re-Identification (CVPR 2022 · similarity 0.84)
- Semi-Supervised Bayesian Attribute Learning for Person Re-Identification (AAAI 2018 · similarity 0.84)
- Improving Deep Metric Learning by Divide and Conquer (TPAMI 2021 · similarity 0.84)
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.
