Research map / Label-efficient and robust learning / Recognition and long-tail
CoReS: Compatible Representations via Stationarity
Niccolò Biondi, Federico Pernici, Matteo Bruni, Alberto Del Bimbo
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 |
|---|---|---|---|
| replace BCT with CoReS fixed-simplex stationary training | compatible training · CIFAR-100/10, 1 upgrade | update gain (%): 5.9 -> 21.3 | Improved |
| CoReS vs BCT at two upgrades | compatible training · CIFAR-100/10, 2 upgrades | update gain (%): 5.2 -> 54.7 | Improved |
| CoReS vs BCT at nine upgrades | compatible training · CIFAR-100/10, 9 upgrades | AC: 0.11 -> 0.51 | Improved |
| CoReS vs BCT at nine upgrades | compatible training · CASIA-WebFace/LFW, 9 upgrades | AC: 0.09 -> 0.58 | Improved |
| CoReS vs BCT with one upgrade | compatible training · CASIA-WebFace/LFW, 1 upgrade | update gain (%): 0.29 -> 0.32 | No clear effect |
| CoReS vs BCT with two upgrades | compatible training · Market1501, 2 upgrades | AC: 0.3 -> 1 | Improved |
| CoReS vs BCT cross-test | compatible training · GLDv2 1 upgrade | mAP@100: 0.16 -> 0.20 | Improved |
| CoReS vs BCT cross-test | compatible training · MET kNN 1 upgrade | accuracy: 0.54 -> 0.65 | Improved |
| add model selection (ablation) | CoReS from scratch · CIFAR-100/10, 9 upgrades | AC: 0.51 -> 0.53 | Improved |
| add model selection to pretrained CoReS (50 classes) (ablation) | CoReS · CIFAR-100/10, 9 upgrades | AC: 0.67 -> 0.74 | Improved |
| pretrain on 50 classes instead of none (ablation) | CoReS · CIFAR-100/10, 9 upgrades | AC: 0.51 -> 0.67 | Improved |
| use different random init at each upgrade instead of same init (ablation) | CoReS · CIFAR-100/10, 9 upgrades | AC: 0.51 -> 0.45 | Hurt |
Later papers that compared with this method
3 later accepted papers report CORES as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- On Revisiting Entropy for Identifying Mislabeled Images (ICML 2026 · better on 1 of 1 shared result)
- Enhanced Sample Selection with Confidence Tracking: Identifying Correctly Labeled Yet Hard-to-Learn Samples in Noisy Data (AAAI 2025 · better on 1 of 1 shared result)
- Delving into Noisy Label Detection with Clean Data (ICML 2023 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Stationary Representations: Optimally Approximating Compatibility and Implications for Improved Model Replacements (CVPR 2024 · similarity 0.88)
- A Stationary (and Therefore Compatible) Representation is All You Need (TPAMI 2026 · similarity 0.86)
- Compositing-aware Image Search (ECCV 2018 · similarity 0.85)
- Continual Learning for Visual Search With Backward Consistent Feature Embedding (CVPR 2022 · similarity 0.85)
- Switchable Representation Learning Framework With Self-Compatibility (CVPR 2023 · similarity 0.84)
- OmniGlue: Generalizable Feature Matching with Foundation Model Guidance (CVPR 2024 · 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.
