Research map / Optimization / Implicit bias and generalization
Quantitative Propagation of Chaos for SGD in Wide Neural Networks
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 step size scaling beta=1 instead of beta=0.75 (implicit regularization noise regime) (ablation) | 2-layer ReLU network trained with SGD · CIFAR-10, N=5000 | test accuracy (%): 55.5 -> 56.5 | Helped |
| use beta=1 instead of beta=0.75 (ablation) | 2-layer ReLU network trained with SGD · CIFAR-10, N=10000 | test accuracy (%): 56.0 -> 56.5 | Helped |
| use beta=1 instead of beta=0.75 (ablation) | 2-layer ReLU network trained with SGD · CIFAR-10, N=50000 | test accuracy (%): 56.7 -> 57.7 | Helped |
Closest papers by meaning
- Uniform-in-Time Weak Propagation of Chaos in Shallow Neural Networks (NeurIPS 2026 · similarity 0.90)
- Improved Particle Approximation Error for Mean Field Neural Networks (NeurIPS 2024 · similarity 0.89)
- High-dimensional Limit of SGD for Diagonal Linear Networks (NeurIPS 2026 · similarity 0.89)
- Emergence and scaling laws in SGD learning of shallow neural networks (NeurIPS 2025 · similarity 0.89)
- Propagation of Chaos for Mean-Field Langevin Dynamics and its Application to Model Ensemble (ICML 2025 · similarity 0.88)
- Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD (ICML 2024 · 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.