Research map / Label-efficient and robust learning / Semi-supervised learning
TestRank: Bringing Order into Unlabeled Test Instances for Deep Learning Tasks
YU LI, Min Li, Qiuxia LAI, Yannan Liu, Qiang Xu
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 contextual attributes (GNN on similarity graph) to intrinsic attributes | TestRank · CIFAR-10 model A | ATPF: 67.47 -> 76.56 (vs DeepGini) | Improved |
| combine intrinsic and contextual attributes instead of contextual only (ablation) | TestRank · CIFAR-10 model A | ATPF: 51.39 -> 76.56 | Improved |
| use contextual attributes only vs random (ablation) | TestRank Contextual-Only · CIFAR-10 model A | ATPF: 30.15 -> 51.39 | Improved |
| contextual-only vs DeepGini on model C | TestRank Contextual-Only · SVHN model C | ATPF: 81.68 -> 93.99 | Improved |
| replace BYOL feature extractor with target model front layers (ablation) | TestRank · CIFAR-10 / STL10 | ATPF: -6.23 average (76.56 -> 68.84 on CIFAR-10 A) | Hurt |
| use approximate kNN graph instead of exact graph (ablation) | TestRank · CIFAR-10/SVHN/STL10 | ATPF: +0.95 average influence | No clear effect |
| vary number of neighbors k (20-800) (ablation) | TestRank · STL10 | ATPF | Mixed |
Closest papers by meaning
- Computing the Testing Error Without a Testing Set (CVPR 2020 · similarity 0.85)
- DarkRank: Accelerating Deep Metric Learning via Cross Sample Similarities Transfer (AAAI 2018 · similarity 0.84)
- Model Similarity Mitigates Test Set Overuse (NeurIPS 2019 · similarity 0.84)
- Testing Semantic Importance via Betting (NeurIPS 2024 · similarity 0.82)
- Label-Efficient Dataset Pruning via Semi-Supervised Pseudo-Labeling (NeurIPS 2026 · similarity 0.82)
- SPADE: Sparsity-Guided Debugging for Deep Neural Networks (ICML 2024 · similarity 0.82)
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.
