Research map / Optimization / Implicit bias and generalization
Complex Dynamics in Simple Neural Networks: Understanding Gradient Flow in Phase Retrieval
Stefano Sarao Mannelli, Giulio Biroli, Chiara Cammarota, Florent Krzakala, Pierfrancesco Urbani, Lenka Zdeborová
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 |
|---|---|---|---|
| Increase the sample-to-dimension ratio above the empirical recovery regime | Randomly initialized gradient descent for phase retrieval · Phase retrieval | zero test error: before α∼7.0 | Improved |
| Increase the sample-to-dimension ratio from α=6 to α=18 or α=24 | Gradient flow for phase retrieval · Phase retrieval | training loss convergence to zero: α=6; α=18; 24 | Improved |
| Operate near the critical sample-to-dimension ratio | Gradient flow for phase retrieval · Phase retrieval, α=12 | perfect generalization across runs: α=12 | Mixed |
| Use the threshold-state Hessian's signal-aligned unstable direction after the BBP transition | Gradient flow for phase retrieval · Phase retrieval, above the critical ratio | recovery of the global minimum | Improved |
Closest papers by meaning
- Fast Escape, Slow Convergence: Learning Dynamics of Phase Retrieval under Power-Law Data (ICLR 2026 · similarity 0.90)
- SAD Neural Networks: Divergent Gradient Flows and Asymptotic Optimality via o-minimal Structures (NeurIPS 2025 · similarity 0.87)
- No Wrong Turns: The Simple Geometry Of Neural Networks Optimization Paths (ICML 2024 · similarity 0.87)
- A Theoretical Framework for Grokking: Interpolation followed by Riemannian Norm Minimisation (NeurIPS 2025 · similarity 0.87)
- Random Matrix Theory of Early-Stopped Gradient Flow: A Transient BBP Scenario (NeurIPS 2026 · similarity 0.86)
- The Global Convergence Time of Stochastic Gradient Descent in Non-Convex Landscapes: Sharp Estimates via Large Deviations (ICML 2025 · 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.
