Research map / Optimization / Nonconvex and smooth optimization
Generalization Bounds for Gradient Methods via Discrete and Continuous Prior
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 FGD with a discrete data-dependent prior | CNN · MNIST | test error upper bound: 0.11 -> 0.026 | Helped |
| Use FSGD with a discrete data-dependent prior | SimpleNet · CIFAR10 | test error upper bound: 0.23 -> 0.198 | Helped |
| Use finite-precision FGD instead of GD | CNN · MNIST | training and testing curves | No clear effect |
| Use finite-precision FSGD instead of SGD | SimpleNet · CIFAR10 | training and testing curves | No clear effect |
| Increase the prior subset size m | FGD and FSGD · MNIST and CIFAR10 | cumulative squared gradient difference | Helped |
Closest papers by meaning
- On Generalization Error Bounds of Noisy Gradient Methods for Non-Convex Learning (ICLR 2020 · similarity 0.95)
- Time-Independent Information-Theoretic Generalization Bounds for SGLD (NeurIPS 2023 · similarity 0.92)
- Time-independent Generalization Bounds for SGLD in Non-convex Settings (NeurIPS 2021 · similarity 0.92)
- Generalization of noisy SGD in unbounded non-convex settings (ICML 2025 · similarity 0.90)
- Temperature is All You Need for Generalization in Langevin Dynamics and other Markov Processes (NeurIPS 2025 · similarity 0.89)
- Generalization Bound of Gradient Flow through Training Trajectory and Data-dependent Kernel (NeurIPS 2025 · 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.