Research map / Label-efficient and robust learning / Semi-supervised learning
A Neural Pre-Conditioning Active Learning Algorithm to Reduce Label Complexity
Seo Taek Kong, Soomin Jeon, Dongbin Na, Jaewon Lee, Hong-Seok Lee, Kyu-Hwan Jung
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 BADGE with NPC | active learning + FixMatch · CIFAR-10, 50 labels | accuracy (%): 86.43 ± 1.22 -> 94.63 ± 0.07 | Improved |
| replace BADGE with NPC | active learning + FixMatch · CIFAR-100, 300 labels | accuracy (%): 24.55 ± 1.13 -> 40.92 ± 1.65 | Improved |
| replace random (passive) sampling with NPC | active learning + FixMatch · CIFAR-100, 700 labels | accuracy (%): 52.69 ± 1.16 -> 55.76 ± 1.45 | Improved |
| replace ALBL with NPC | active learning + FixMatch · CIFAR-100, 500 labels | accuracy (%): 49.88 ± 0.92 -> 48.16 ± 0.77 | Hurt |
| replace passive sampling with NPC | active learning + FixMatch · CIFAR-10, 30 labels | accuracy (%): 78.08 ± 5.48 -> 85.09 ± 9.61, but Margin 87.36 | Mixed |
| use single-shot query model instead of zero-shot (random init) for BADGE | BADGE · CIFAR-10, 60 labels | accuracy (%): 94.42 ± 0.41 -> 70.28 ± 13.22 | Hurt |
| use single-shot query model instead of zero-shot for NPC | NPC · CIFAR-10, 60 labels | accuracy (%): 93.16 ± 2.34 -> 93.76 ± 0.58 | No clear effect |
| use NPC zero-shot (random init network) vs FixMatch on balanced data without DARP | NPC · CIFAR-10 | accuracy (%): 86.19 -> 93.14 ± 1.42 | Improved |
| replace BADGE with NPC at zero-shot 40 labels | active learning · CIFAR-10 | accuracy (%): 92.13 ± 3.24 -> 93.14 ± 1.42 | No clear effect |
| replace BADGE with NPC at single-shot 40 labels | active learning · CIFAR-10 | accuracy (%): 86.70 ± 5.42 -> 92.26 ± 3.26 | Improved |
Closest papers by meaning
- Consistency-based Semi-supervised Active Learning: Towards Minimizing Labeling Cost (ECCV 2020 · similarity 0.88)
- Enhancing Semi-Supervised Learning via Representative and Diverse Sample Selection (NeurIPS 2024 · similarity 0.87)
- Towards Cost-Effective Learning: A Synergy of Semi-Supervised and Active Learning (CVPR 2025 · similarity 0.87)
- Active Learning Guided by Efficient Surrogate Learners (AAAI 2024 · similarity 0.86)
- The Surprising Effectiveness of Infinite-Width NTKs for Characterizing and Improving Model Training (AAAI 2025 · similarity 0.86)
- Provably Neural Active Learning Succeeds via Prioritizing Perplexing Samples (ICML 2024 · similarity 0.85)
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.
