Research map / Optimization / Implicit bias and generalization
Generalization Error Bounds for Optimization Algorithms via Stability
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 |
|---|---|---|---|
| Continue optimization training | GD, SGD, and SVRG for R-ERM · Linear regression, rcv1 logistic regression, and fully connected neural networks | test error | Helped |
| Use SVRG instead of GD | R-ERM optimization · Linear regression, rcv1 logistic regression, and fully connected neural networks | training time to reduce test error | Helped |
| Use SVRG instead of SGD | R-ERM optimization · Linear regression, rcv1 logistic regression, and fully connected neural networks | training time to reduce test error | Helped |
| Use SGD instead of GD or SVRG | R-ERM optimization · Large training sets | test error after sufficient training | Hurt |
| Use SVRG instead of GD or SGD | R-ERM optimization · rcv1 logistic regression | training time to reduce test error | Helped |
| Use SVRG instead of GD or SGD | R-ERM optimization · Fully connected neural networks | training time to reduce test error | Helped |
| Apply the convex stability-based generalization analysis | R-ERM optimization algorithms · Convex R-ERM | expected generalization error bound: O(1/n+Eρ(T)) | Helped |
| Apply the high-probability stability-based analysis | R-ERM optimization algorithms · Convex R-ERM | high-probability generalization error bound: O(log(1/δ)/√n+ρ(T)) | Helped |
| Use stability analysis with convergence error | R-ERM optimization algorithms · Nonconvex R-ERM | expected generalization error bound | Helped |
Closest papers by meaning
- Sharper Generalization Bounds for Learning with Gradient-dominated Objective Functions (ICLR 2021 · similarity 0.92)
- On Generalization Error Bounds of Noisy Gradient Methods for Non-Convex Learning (ICLR 2020 · similarity 0.92)
- Stability and Generalization Analysis of Decentralized SGD: Sharper Bounds Beyond Lipschitzness and Smoothness (ICML 2025 · similarity 0.90)
- Generalization Analysis of Biased Stochastic Gradient Methods for Minimax Problems (NeurIPS 2026 · similarity 0.89)
- Algorithmic Stability Unleashed: Generalization Bounds with Unbounded Losses (ICML 2024 · similarity 0.89)
- Generalizability of Neural Networks Minimizing Empirical Risk Based on Expressive Power (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.