Research map / Architectures and efficiency / RNNs and normalization
Self-Consistent Dynamical Field Theory of Kernel Evolution 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 full self-consistent DMFT instead of gradient independence ansatz | kernel dynamics prediction · deep linear network, large gamma_0 | kernel/loss prediction accuracy | Helped |
| use full DMFT instead of leading-order perturbation theory | kernel dynamics prediction · L=5 linear network, gamma_0=1.0 | loss and kernel agreement with NN | Helped |
| increase feature learning strength gamma_0 | wide CNN · two-class CIFAR | training speed | Helped |
| increase gamma_0 in depth-7 CNN | wide CNN · two-class CIFAR | final test accuracy | No clear effect |
Closest papers by meaning
- Three Mechanisms of Feature Learning in a Linear Network (ICLR 2025 · similarity 0.90)
- Global Convergence and Rich Feature Learning in $L$-Layer Infinite-Width Neural Networks under $\mu$ Parametrization (ICML 2025 · similarity 0.89)
- Adaptive kernel predictors from feature-learning infinite limits of neural networks (ICML 2025 · similarity 0.89)
- Mean-field Analysis on Two-layer Neural Networks from a Kernel Perspective (ICML 2024 · similarity 0.89)
- Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer Networks (NeurIPS 2025 · similarity 0.88)
- Critical feature learning in deep neural networks (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.