Research map / Learning theory and algorithms / Clustering
Active Learning with Low-Rank Structure for Data Selection
Vincent Cohen-Addad, Sasidhar Kunapuli, Vahab Mirrokni, Mahdi Nikdan, David Woodruff, Samson Zhou
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 leverage score sampling to choose basis samples instead of uniform (ablation) | Low-rank SS · ViGGO Qwen2.5-3B 12.5% | accuracy (%): 50.70 -> 51.54 | Improved |
| use leverage score sampling instead of k-medians basis selection (ablation) | Low-rank SS · ViGGO Qwen2.5-3B 12.5% | accuracy (%): 45.80 -> 51.54 | Improved |
| select subsets with Llama3-8B instead of Qwen2.5-3B itself | Low-rank SS · ViGGO Qwen2.5-3B 12.5% | accuracy (%): 51.54 -> 55.74 | Improved |
| replace clustering-based SS with low-rank SS | Sensitivity sampling · ViGGO Qwen2.5-3B 25% | accuracy (%): 71.57 -> 75.91 | Improved |
| replace clustering-based SS with low-rank SS | Sensitivity sampling · ViGGO Llama3-8B 25% | accuracy (%): 86.6 ± 2.8 -> 88.3 ± 0.2 | Improved |
| replace clustering-based SS with low-rank SS | Sensitivity sampling · GSM8k Llama3-8B 25% | accuracy (%): 70.2 -> 68.4 | Hurt |
| replace clustering-based SS with low-rank SS | Sensitivity sampling · ViGGO Llama3-8B 12.5% | accuracy (%): 72.8 -> 69.7 | Hurt |
| replace clustering-based SS with low-rank SS | Sensitivity sampling · SQL Llama3-8B 6.25% | accuracy (%): 68.3 -> 70.4 | Improved |
| replace uniform sampling with low-rank SS | Data selection · Average over datasets Llama3-8B 6.25% | accuracy (%): 52.0 -> 54.9 | Improved |
| replace uniform sampling with low-rank SS at 12.5% average | Data selection · Average over datasets Llama3-8B 12.5% | accuracy (%): 69.2 -> 70.4 | Improved |
| replace loss with gradient norm as sensitivity objective and BERT with GTR embeddings | Low-rank SS · GSM8k Llama3-8B 12.5% | validation accuracy | No clear effect |
Closest papers by meaning
- Data-Efficient Learning via Clustering-Based Sensitivity Sampling: Foundation Models and Beyond (ICML 2024 · similarity 0.90)
- Unsupervised Active Learning via Subspace Learning (AAAI 2021 · similarity 0.86)
- Towards a statistical theory of data selection under weak supervision (ICLR 2024 · similarity 0.86)
- Structural-Entropy-Based Sample Selection for Efficient and Effective Learning (ICLR 2025 · similarity 0.84)
- The Impact of Coreset Selection on Spurious Correlations and Group Robustness (NeurIPS 2025 · similarity 0.83)
- On Socially Fair Low-Rank Approximation and Column Subset Selection (NeurIPS 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.
