Research map / Optimization / Implicit bias and generalization
On the Theoretical Properties of Noise Correlation in Stochastic Optimization
Aurelien Lucchi, Frank Proske, Antonio Orvieto, Francis Bach, Hans Kersting
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 |
|---|---|---|---|
| Use smaller Hurst parameter H for fPGD | fPGD · Styblinski–Tang | time to leave shallow minimum | Improved |
| Use smaller Hurst parameter H for fPGD | fPGD · Styblinski–Tang | reaching deep minimum | Improved |
| Use H=0.7 or H=0.9 rather than smaller H | fPGD · Styblinski–Tang | reaching deep minimum within 10^5 steps | Hurt |
| Use H=0.4 rather than H=0.5 or H=0 | fPGD · Embedded saddle (d=400) | speed of finding negative-curvature direction | Improved |
| Use H=0.3 or H=0.4 rather than H=0.5 or H=0 | fPGD · Embedded saddle (d=400) | final accuracy | Improved |
| Use H=0.5 rather than smaller H | fPGD · Embedded saddle (d=400) | speed of leaving saddle | Improved |
| Use H=0.5 rather than H=0.3 or H=0.4 | fPGD · Embedded saddle (d=400) | final accuracy | Hurt |
| Use H=0 rather than H=0.3 or H=0.4 | fPGD · Embedded saddle (d=400) | speed of leaving saddle | Hurt |
Closest papers by meaning
- Multiplicative Noise and Heavy Tails in Stochastic Optimization (ICML 2021 · similarity 0.90)
- Singular-limit analysis of gradient descent with noise injection (JMLR 2026 · similarity 0.90)
- The Anisotropic Noise in Stochastic Gradient Descent: Its Behavior of Escaping from Sharp Minima and Regularization Effects (ICML 2019 · similarity 0.89)
- Stochastic Zeroth-Order Optimization Under Heavy-Tailed Noise (NeurIPS 2026 · similarity 0.88)
- Nonconvex Stochastic Optimization under Heavy-Tailed Noises: Optimal Convergence without Gradient Clipping (ICLR 2025 · similarity 0.88)
- Convergence Rate in a Nonlinear Two-Time-Scale Stochastic Approximation with State (Time)-Dependence (AAAI 2025 · similarity 0.87)
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.
