Research map / Optimization / Optimizers
Almost Tune-Free Variance Reduction
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 BB step sizes, weighted averaging, and an adaptive inner-loop length | SARAH · a9a | convergence | Helped |
| Use BB step sizes, weighted averaging, and an adaptive inner-loop length | SVRG · a9a | convergence | Mixed |
| Use BB step sizes, weighted averaging, and an adaptive inner-loop length | SARAH · rcv1.binary | convergence | No clear effect |
| Use BB step sizes, weighted averaging, and an adaptive inner-loop length | SVRG · rcv1.binary | convergence | Helped |
| Use BB step sizes, weighted averaging, and an adaptive inner-loop length | SARAH · real-sim | convergence | No clear effect |
| Use BB step sizes, weighted averaging, and an adaptive inner-loop length | SVRG · real-sim | convergence | No clear effect |
| Use W-Avg instead of U-Avg or L-Avg (ablation) | SARAH with a large step size on w7a · w7a | convergence: almost two orders | Helped |
| Use W-Avg instead of U-Avg or L-Avg (ablation) | SARAH with a medium step size on w7a · w7a | convergence | Mixed |
| Use L-Avg instead of W-Avg or U-Avg (ablation) | SARAH with a small step size on w7a · w7a | convergence | Helped |
| Use W-Avg instead of U-Avg or L-Avg (ablation) | BB-SARAH with large inner-loop length · a9a | convergence | Helped |
| Use W-Avg instead of U-Avg or L-Avg (ablation) | SARAH · large inner-loop length | convergence rate | Helped |
| Use a dynamically adjusted inner-loop length m^s = c/(μη^s) instead of a fixed length (ablation) | BB-SARAH · binary classification datasets | convergence | Helped |
Closest papers by meaning
- Towards closing the gap between the theory and practice of SVRG (NeurIPS 2019 · similarity 0.89)
- Barzilai-Borwein Step Size for Stochastic Gradient Descent (NeurIPS 2016 · similarity 0.87)
- Double Variance Reduction: A Smoothing Trick for Composite Optimization Problems without First-Order Gradient (ICML 2024 · similarity 0.86)
- Dissipativity Theory for Accelerating Stochastic Variance Reduction: A Unified Analysis of SVRG and Katyusha Using Semidefinite Programs (ICML 2018 · similarity 0.85)
- Nonconvex Variance Reduced Optimization with Arbitrary Sampling (ICML 2019 · similarity 0.85)
- A Coefficient Makes SVRG Effective (ICLR 2025 · 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.