Research map / Learning theory and algorithms / Prediction and decision losses
Identifying a Minimal Class of Models for High--dimensional Data
Daniel Nevo, Ya'acov Ritov
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 |
|---|---|---|---|
| increase signal to noise ratio from 1 to 12 | simulated annealing model search, model (I) · Simulation n=100, p=200 | proportion chosen as best model: 0.00 -> 0.86 | Improved |
| increase number of candidate predictors from 200 to 1000 | simulated annealing model search, model (I) · Simulation n=100, SNR=12 | proportion chosen as best model: 0.86 -> 0.49 | Hurt |
| increase number of candidate predictors from 200 to 1000 | simulated annealing model search, model (II) · Simulation n=100, SNR=1 | proportion chosen as best model: 0.42 -> 0.23 | Hurt |
| increase number of candidate predictors from 200 to 1000 | simulated annealing model search, model (II) · Simulation n=100, SNR=12 | proportion chosen as best model: 0.98 -> 0.96 | No clear effect |
| increase SNR from 1 to 2 | simulated annealing model search, model (II) · Simulation n=100, p=200 | proportion chosen as best model: 0.42 -> 0.94 | Improved |
| competition between equal-size models (III) and (IV) at high SNR | simulated annealing model search · Simulation n=100 | proportion chosen as best model: about 50% | Hurt |
| start the search from three different initial models instead of one (ablation) | simulated annealing model search, models (III) and (IV) · Simulation n=100 | proportion chosen in top five models | Improved |
| use the stability selection three-gene model instead of the search's size-3 models | stability selection screening · Riboflavin (n=71, p=4088) | MSE: 0.2047 -> 0.2703 | Hurt |
Closest papers by meaning
- A Comparison of Hamming Errors of Representative Variable Selection Methods (ICLR 2022 · similarity 0.86)
- Rank-based Lasso - efficient methods for high-dimensional robust model selection (JMLR 2020 · similarity 0.86)
- Statistical-Computational Tradeoff in Single Index Models (NeurIPS 2019 · similarity 0.85)
- The Complete Lasso Tradeoff Diagram (NeurIPS 2020 · similarity 0.85)
- No penalty no tears: Least squares in high-dimensional linear models (ICML 2016 · similarity 0.85)
- Towards a statistical theory of data selection under weak supervision (ICLR 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.
