Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Active Learning via Subspace Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Select samples under learned low-rank representations rather than with them (ablation) | ALSL · UTKFace | accuracy: 0.218±0.024 -> 0.242±0.018; 0.376±0.013 -> 0.398±0.014 | Helped |
| Select samples under learned low-rank representations rather than with them (ablation) | ALSL · HAM10000 | accuracy: 0.256±0.019 -> 0.271±0.020; 0.360±0.010 -> 0.407±0.020 | Helped |
| Add low-rank sample representation learning (ablation) | ALSL W · UTKFace | accuracy: 0.130±0.013 -> 0.218±0.024; 0.335±0.014 -> 0.376±0.013 | Helped |
| Add low-rank sample representation learning (ablation) | ALSL W · HAM10000 | accuracy: 0.156±0.036 -> 0.256±0.019; 0.321±0.022 -> 0.360±0.010 | Helped |
| Jointly learn the representation and select samples instead of using separate steps (ablation) | ALSL U · HMDB51 | accuracy: 0.101±0.022 -> 0.159±0.011; 0.326±0.020 -> 0.377±0.009 | Helped |
| Jointly learn the representation and select samples instead of using separate steps (ablation) | ALSL U · UCF50 | accuracy: 0.107±0.022 -> 0.249±0.010; 0.523±0.020 -> 0.684±0.014 | Helped |
| Select samples under low-rank representations rather than reconstructing original inputs | ALSL U · UCF50, 100 selected | accuracy: 0.246±0.017 -> 0.377±0.011 | Helped |
| Select samples with low-rank representations while retaining original-input reconstruction | ALSL W · UTKFace, HAM10000, HMDB51, UCF50, and Wine Quality | accuracy | Mixed |
Closest papers by meaning
- Active Learning as Nullspace Regulation: A Spectral Representation Perspective (NeurIPS 2026 · similarity 0.88)
- Discover-Then-Rank Unlabeled Support Vectors in the Dual Space for Multi-Class Active Learning (ICML 2023 · similarity 0.87)
- Sample-Efficient Linear Representation Learning from Non-IID Non-Isotropic Data (ICLR 2024 · similarity 0.84)
- On the Benefits of Active Data Collection in Operator Learning (ICML 2025 · similarity 0.84)
- Machine Unlearning in Low-Dimensional Feature Subspace (NeurIPS 2026 · similarity 0.84)
- A Unified Framework for Learning with Nonlinear Model Classes from Arbitrary Linear Samples (ICML 2024 · 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.