Research map / Learning theory and algorithms / Sample complexity
A Two-Stage Approach for Learning a Sparse Model with Sharp Excess Risk Analysis
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 |
|---|---|---|---|
| use Epoch-SGD instead of SGD in the first stage | SGD · E2006-tfidf and E2006-log1p | test RMSE | Helped |
| use distribution dependent sampling instead of magnitude sampling in the sparsification stage (ablation) | MG-Sparsification · E2006-tfidf and E2006-log1p | test RMSE at same K | Helped |
| include more coordinates K in sparsified model (ablation) | DD-Sparsification · E2006-tfidf and E2006-log1p | test RMSE | Helped |
| sparsify with K = 500 (tfidf) or K = 5000 (log1p) | DD-Sparsification · E2006-tfidf and E2006-log1p | test RMSE vs full model | No clear effect |
| two-stage learning instead of single-stage l1-constrained SGD with small B | SpS single stage · E2006 | test RMSE vs sparsity | Helped |
Closest papers by meaning
- Feature Adaptation for Sparse Linear Regression (NeurIPS 2023 · similarity 0.88)
- Robust Sparse Regression with Non-Isotropic Designs (NeurIPS 2024 · similarity 0.88)
- MAST: model-agnostic sparsified training (ICLR 2025 · similarity 0.86)
- An Iterative Min-Min Optimization Method for Sparse Bayesian Learning (ICML 2024 · similarity 0.86)
- Sparse Variational Student-t Processes (AAAI 2024 · similarity 0.86)
- Establishing Linear Surrogate Regret Bounds for Convex Smooth Losses via Convolutional Fenchel–Young Losses (NeurIPS 2025 · similarity 0.86)
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.