Research map / Optimization / Nonconvex and smooth optimization
Uniform Convergence of Gradients for Non-Convex Learning and Optimization
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 norm-based vector-valued Rademacher complexity to control gradient convergence | Smooth non-convex learning problems · Generalized Linear Model and Robust Regression | sample complexity: O(\varepsilon^{-2}) | Helped |
| Use the low-dimensional \ell_2 setup | Proposed meta-algorithm · Generalized Linear Model and Robust Regression | sample complexity: O(d\varepsilon^{-1}) | Helped |
| Use the proposed general stationary-point method instead of a specialized GLM algorithm | Generalized Linear Model · Low-dimensional | sample complexity: O(d\varepsilon^{-1}) | Mixed |
| Require dimension-independent gradient convergence for a single ReLU without extra assumptions | Single ReLU · Worst-case instances | normed Rademacher complexity: \Omega(\sqrt{dn}\wedge{}n) | Hurt |
| Impose the empirical soft-margin condition with \phi(\gamma)=\gamma^{\frac{1}{2}} | Single ReLU gradient convergence · Empirical soft-margin class | gradient uniform convergence bound: O(n^{-\frac{1}{12}}) | Helped |
Closest papers by meaning
- The Sample Complexity of Gradient Descent in Stochastic Convex Optimization (NeurIPS 2024 · similarity 0.90)
- Leveraging Non-uniformity in First-order Non-convex Optimization (ICML 2021 · similarity 0.90)
- Thinking Outside the Ball: Optimal Learning with Gradient Descent for Generalized Linear Stochastic Convex Optimization (NeurIPS 2022 · similarity 0.90)
- Stability and Sharper Risk Bounds with Convergence Rate $\tilde{O}(1/n^2)$ (NeurIPS 2025 · similarity 0.88)
- Spectral Preconditioning for Gradient Methods on Graded Non-convex Functions (ICML 2024 · similarity 0.88)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.88)
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.