Research map / Optimization / Nonconvex and smooth optimization
Towards Fully Parameter-Free Stochastic Optimization: Grid Search with Self-Bounding Analysis
Yuheng Zhao, Yu-Hu Yan, Amit Attia, Tomer Koren, Lijun Zhang, Peng Zhao
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 |
|---|---|---|---|
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-NC · CIFAR-10 ResNet-18 non-convex, (L=0.001,F=0.001) | relative difference rho: 0.1836 -> 0.1699 | Improved |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-NC · CIFAR-10 ResNet-18 non-convex, (L=0.1,F=0.01) | relative difference rho: 0.2023 -> 0.1141 | Improved |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-NC · CIFAR-10 ResNet-18 non-convex, (L=0.01,F=0.1) | relative difference rho: 0.2094 -> 0.1772 | Improved |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-NC · CIFAR-10 ResNet-18 non-convex, (L=0.1,F=0.1) | relative difference rho: 0.1141 -> 0.1141 | No clear effect |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-NC · CIFAR-10 ResNet-18 non-convex, (L=0.01,F=0.01) | relative difference rho: 0.2094 -> 0.2094 | No clear effect |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-C Option-I · CIFAR-10 ResNet-18 convex, (d=0.001,L=0.001) | relative difference rho: 0.1640 -> 0.1419 | Improved |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-C Option-I · CIFAR-10 ResNet-18 convex, (d=0.01,L=0.01) | relative difference rho: 0.1303 -> 0.1190 | Improved |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-C Option-I · CIFAR-10 ResNet-18 convex, (d=0.1,L=0.01) | relative difference rho: 0.0920 -> 0.0866 | Improved |
| reduce initial sampling budget M from T/4 to T/4096 (ablation) | Grasp-C Option-II · CIFAR-10 ResNet-18 convex, (d=0.001,L=0.1) | relative difference rho: 0.1616 -> 0.1411 | Improved |
| use Option-II (known lower bound 0) instead of Option-I | Grasp-C · CIFAR-10 ResNet-18 convex, (d=0.001,L=0.1) | relative difference rho: 0.1408 -> 0.1411 | No clear effect |
| increase input L_eps from 0.001 to 0.1 (F_eps=0.01) | Grasp-NC · CIFAR-10 ResNet-18 non-convex, M=T/4096 | relative difference rho: 0.1468 -> 0.1141 | Improved |
| increase input d_eps from 0.001 to 0.1 (L_eps=0.01) | Grasp-C Option-I · CIFAR-10 ResNet-18 convex, M=T/4096 | relative difference rho: 0.1419 -> 0.0866 | Improved |
Closest papers by meaning
- How Free is Parameter-Free Stochastic Optimization? (ICML 2024 · similarity 0.93)
- Tuning-Free Stochastic Optimization (ICML 2024 · similarity 0.87)
- Stochastic Optimization with Random Search (NeurIPS 2026 · similarity 0.87)
- Projection-Free Variance Reduction Methods for Stochastic Constrained Multi-Level Compositional Optimization (ICML 2024 · similarity 0.86)
- Improved Lower Bounds for First-order Stochastic Non-convex Optimization under Markov Sampling (ICML 2025 · similarity 0.86)
- Sample-and-Bound for Non-convex Optimization (AAAI 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.
