Research map / Optimization / Nonconvex and smooth optimization
High-Probability Bounds for Stochastic Optimization and Variational Inequalities: the Case of Unbounded Variance
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 |
|---|---|---|---|
| add gradient clipping to SGD for alpha-th moment noise (alpha in (1,2]), smooth convex/strongly convex settings | SGD · smooth minimization, unbounded variance | high-probability complexity | Helped |
| SGD without clipping under bounded variance | SGD · strongly convex quadratic, alpha=2 lower-bound construction | iterations to reach P{||x^K-x*||^2>=eps}<=beta | Hurt |
| restart clipped-SSTM in strongly convex case | clipped-SSTM · strongly convex minimization | high-probability complexity | Helped |
| use clipped-SEG instead of unclipped methods for monotone VIP with alpha<2 | extragradient for VIP · Lipschitz monotone VIP, unbounded variance | high-probability complexity | No clear effect |
Closest papers by meaning
- High Probability Convergence of Stochastic Gradient Methods (ICML 2023 · similarity 0.93)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.92)
- High-Probability Convergence for Composite and Distributed Stochastic Minimization and Variational Inequalities with Heavy-Tailed Noise (ICML 2024 · similarity 0.92)
- High-probability Bounds for Non-Convex Stochastic Optimization with Heavy Tails (NeurIPS 2021 · similarity 0.91)
- High-Probability Bound for Non-Smooth Non-Convex Stochastic Optimization with Heavy Tails (ICML 2024 · similarity 0.90)
- Complexity Lower Bounds of Adaptive Gradient Algorithms for Non-convex Stochastic Optimization under Relaxed Smoothness (ICLR 2025 · similarity 0.89)
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.