Research map / Optimization / Nonconvex and smooth optimization
Adaptive SVRG Methods under Error Bound Conditions with Unknown Growth Parameter
Yi Xu, Qihang Lin, Tianbao Yang
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 |
|---|---|---|---|
| Adaptive search for unknown QEB growth parameter c (SVRGQEB-RS) versus SVRGHEB with fixed initial T1 (ablation) | SVRG under QEB · Adult, squared hinge + l1 norm | objective gap (log scale) over #grad/n | Improved |
| Adaptive search for c makes performance insensitive to initial T1 estimate (ablation) | SVRGQEB-RS · Adult, squared hinge + l1 norm | objective gap sensitivity to initial T1 | Improved |
| Replace fixed-epoch SVRG++ baseline with SVRGQEB-RS | SVRG under QEB · Adult, logistic / squared hinge / square / Huber / lp regression | objective gap over #grad/n | Improved |
| Heuristic automatic epoch length (SVRG-heuristics) with 2n cap | SVRG++ · multiple datasets | objective gap over #grad/n | Mixed |
| Restart-based search for c for theta=1/2 (Algorithm 2 cannot be applied at theta=1/2) (ablation) | SVRGHEB-RS · theta=1/2 | convergence guarantee | No clear effect |
Closest papers by meaning
- Stochastic Variance Reduction for Nonconvex Optimization (ICML 2016 · similarity 0.89)
- Towards closing the gap between the theory and practice of SVRG (NeurIPS 2019 · similarity 0.88)
- Estimate Sequences for Variance-Reduced Stochastic Composite Optimization (ICML 2019 · similarity 0.88)
- A Coefficient Makes SVRG Effective (ICLR 2025 · similarity 0.87)
- Complexity Lower Bounds of Adaptive Gradient Algorithms for Non-convex Stochastic Optimization under Relaxed Smoothness (ICLR 2025 · similarity 0.86)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · 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.
