Research map / Optimization / Nonconvex and smooth optimization
Debiasing Averaged Stochastic Gradient Descent to handle missing values
Aude Sportisse, Claire Boyer, Aymeric Dieuleveut, Julie Josse
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 EM instead of AvSGD for parameter estimation | regression with missing values · Superconductivity (81 features) | relative l2 prediction error: 0.7 | Hurt |
| use averaging (AvSGD) over plain SGD with decaying step (ablation) | SGD for least squares with missing values · synthetic, n=10^5, d=10, 30% MCAR | excess empirical risk | Improved |
| run multiple passes instead of one pass (ablation) | AvSGD · synthetic, n=10^3, 100 passes | excess empirical risk | No clear effect |
| ignore heterogeneous missingness probabilities in debiasing (ablation) | AvSGD · synthetic, heterogeneous missing values | excess empirical risk | Hurt |
| use constant step size instead of decaying step size (ablation) | SGD without averaging · synthetic, n=10^5 | excess empirical risk | Mixed |
| debias the gradient instead of mean imputation (ablation) | AvSGD · Superconductivity | relative l2 prediction error | Improved |
Closest papers by meaning
- Parallelizing Stochastic Gradient Descent for Least Squares Regression: Mini-batching, Averaging, and Model Misspecification (JMLR 2018 · similarity 0.87)
- Stochastic Approximate Gradient Descent via the Langevin Algorithm (AAAI 2020 · similarity 0.86)
- Sample Average Approximation for Stochastic Optimization with Dependent Data: Performance Guarantees and Tractability (AAAI 2022 · similarity 0.86)
- Tight Nonparametric Convergence Rates for Stochastic Gradient Descent under the Noiseless Linear Model (NeurIPS 2020 · similarity 0.85)
- Generalization Analysis of Stochastic Weight Averaging with General Sampling (ICML 2024 · similarity 0.85)
- On the Convergence of Loss and Uncertainty-based Active Learning Algorithms (NeurIPS 2024 · similarity 0.84)
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.
