Research map / Learning theory and algorithms / Matrix and tensor methods
On Computationally Tractable Selection of Experiments in Measurement-Constrained Regression Models
Yining Wang, Adams Wei Yu, Aarti Singh
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 greedy selection with pi* instead of Fedorov exchange | L6 Fedorov exchange · Synthetic, isotropic t df=1, k=10p | running time (seconds): 1077 -> 0.3 | Improved |
| use greedy selection with pi* instead of Fedorov exchange | L6 Fedorov exchange · Synthetic, isotropic t df=1, k=2p | running time (seconds): 140.1 -> 1.8 | Improved |
| use greedy selection with pi* instead of Fedorov exchange | L6 Fedorov exchange · Synthetic, skewed Gaussian alpha=3, k=10p | number of iterations: 250 -> 8 | Improved |
| use proposed L4* and L5* instead of uniform, leverage, predictive length sampling | L1, L2, L3 · Synthetic designs (n=1000, p=50) | selection objective and MSE ratio | Improved |
| use proposed L4*/L5* instead of Fedorov exchange | L6 Fedorov exchange · Synthetic designs (n=1000, p=50) | selection performance | No clear effect |
| use greedy L5* instead of uniform sampling | L1 uniform sampling · CPU performance dataset, k=50 | norm of beta error: .142 -> .048 | Improved |
| use sampling based L4* instead of leverage score sampling | L2 leverage score sampling · CPU performance dataset, k=30 | norm of beta error: .201 -> .060 | Improved |
| use greedy L5* instead of sampling based L4* | L4* sampling · CPU performance dataset, k=30 | norm of beta error: .060 -> .104 | Hurt |
| use proposed methods instead of uniform sampling and Fedorov exchange | L1, L6 · Material synthesis dataset (133 experiments, k=21 to 30) | normalized prediction and estimation MSE | Improved |
| use greedy selection instead of uniform and leverage score sampling | L1, L2 · Material synthesis dataset, k=30 | diversity of selected design points | Improved |
| use greedy methods L5*, L6 instead of sampling methods at k near p | L1 to L4* · Minnesota wind dataset | mean-square prediction error ratio | Improved |
| use Fedorov exchange instead of proposed methods on wind data | L4*, L5* · Minnesota wind dataset | running time: more than 10 times running time | Hurt |
Closest papers by meaning
- Experimental Design on a Budget for Sparse Linear Models and Applications (ICML 2016 · similarity 0.88)
- Near-Optimal Design of Experiments via Regret Minimization (ICML 2017 · similarity 0.86)
- Online A-Optimal Design and Active Linear Regression (ICML 2021 · similarity 0.85)
- Near-Optimal Experiment Design in Linear non-Gaussian Cyclic Models (NeurIPS 2025 · similarity 0.84)
- DARTS: Targeting Prognostic Covariates in Budget-Constrained Sequential Experiments (NeurIPS 2026 · similarity 0.84)
- Efficient Adaptive Experimentation with Noncompliance (NeurIPS 2025 · 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.
