Research map / Learning theory and algorithms / Sample complexity
DECOrrelated feature space partitioning for distributed sparse regression
Xiangyu Wang, David B Dunson, Chenlei Leng
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 |
|---|---|---|---|
| add third-stage refinement (DECO-2 to DECO-3) (ablation) | DECO · Synthetic n=500 p=10000 model (ii) | MSE: 4.636 -> 0.241 | Improved |
| add third-stage refinement (DECO-2 to DECO-3) (ablation) | DECO · Synthetic n=500 p=10000 model (iii) | MSE: 1220.5 -> 6.620 | Improved |
| add third-stage refinement (DECO-2 to DECO-3) (ablation) | DECO · Synthetic n=500 p=10000 model (i) | MSE: 3.502 -> 0.102 | Improved |
| remove decorrelation (lasso-naive vs DECO-2) (ablation) | feature-partitioned lasso · Synthetic n=500 p=10000 model (ii) | MSE: 4.636 -> 171.05 | Hurt |
| remove decorrelation (lasso-naive vs DECO-2) (ablation) | feature-partitioned lasso · Synthetic n=500 p=10000 model (i) independent | MSE: 3.667 vs 3.502 | No clear effect |
| DECO-3 instead of full-data lasso with ridge refinement | DECO-3 · Synthetic n=500 p=10000 model (iv) | MSE: 11.15 -> 0.787 | Improved |
| DECO-3 instead of lasso-full | DECO-3 · Synthetic n=500 p=10000 model (ii) | runtime: 806.3 -> 66.9 | Improved |
| vary number of subsets m from 1 to 200 (ablation) | DECO-2 and DECO-3 · p=10000, n=500, model (ii) | accuracy | No clear effect |
| increase p with n=500 on one subset | DECO-2 · n=500, varied p | performance on one subset | No clear effect |
| DECO-3 instead of lasso-full on real data | DECO-3 · Electricity load | runtime: 23,515.5 -> 67.9 | Improved |
| DECO-3 instead of lasso-full on real data | DECO-3 · Electricity load | MSE: 2.205 -> 0.691 | Improved |
| DECO-3 instead of lasso-full on real data | DECO-3 · Mammalian eye diseases | model size: 11 -> 4.3 | Improved |
Closest papers by meaning
- Feature-distributed sparse regression: a screen-and-clean approach (NeurIPS 2016 · similarity 0.90)
- Communication-efficient Sparse Regression (JMLR 2017 · similarity 0.88)
- Debiased Distributed Learning for Sparse Partial Linear Models in High Dimensions (JMLR 2022 · similarity 0.87)
- Doubly Distributed Supervised Learning and Inference with High-Dimensional Correlated Outcomes (JMLR 2020 · similarity 0.86)
- Selective Inference with Distributed Data (JMLR 2025 · similarity 0.86)
- Heterogeneous Sufficient Dimension Reduction and Subspace Clustering (ICML 2025 · 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.
