Research map / Optimization / Implicit bias and generalization
Gradient Flow Through Diagram Expansions: Learning Regimes and Explicit Solutions
Dmitry Yarotsky, Eugene Golikov, Yaroslav Gusev
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 asymmetric instead of symmetric CP model in overparameterized free evolution (ablation) | CP decomposition, free evolution (point B) · Free evolution, nu=3 ASYM vs even-nu SYM | loss decay form: power law (SYM) -> exponential (ASYM) | Mixed |
| use asymmetric instead of symmetric model for NTK regime existence (ablation) | CP decomposition gradient flow · identity target, order nu | existence of NTK limit | Improved |
| large initialization (sigma) so target is negligible (ablation) | CP decomposition gradient flow · identity target | target learning | Hurt |
| gradient ascent with high initialization noise (p^3 sigma^4 above threshold rho*) (ablation) | SYM nu=4 identity target · gradient ascent, p=64 | loss divergence: divergence exponent -4/3 | Hurt |
Closest papers by meaning
- Neural Mechanics: Symmetry and Broken Conservation Laws in Deep Learning Dynamics (ICLR 2021 · similarity 0.87)
- Noether’s Learning Dynamics: Role of Symmetry Breaking in Neural Networks (NeurIPS 2021 · similarity 0.86)
- How to build a consistency model: Learning flow maps via self-distillation (NeurIPS 2025 · similarity 0.85)
- SAD Neural Networks: Divergent Gradient Flows and Asymptotic Optimality via o-minimal Structures (NeurIPS 2025 · similarity 0.85)
- Context-Informed Neural ODEs Unexpectedly Identify Broken Symmetries: Insights from the Poincaré–Hopf Theorem (ICML 2025 · similarity 0.85)
- Random Matrix Theory of Early-Stopped Gradient Flow: A Transient BBP Scenario (NeurIPS 2026 · similarity 0.85)
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.
