Research map / Architectures and efficiency / CNN design
A self consistent theory of Gaussian Processes captures feature learning effects in finite CNNs
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 |
|---|---|---|---|
| Account for higher output cumulants through a self-consistent target shift | GP regression for finite DNNs · Linear CNN teacher-student model | prediction agreement | Helped |
| Use a finite-channel CNN instead of its infinite-channel GP limit | Linear CNN teacher-student model · Training sets with n ∝ S^1.7 | prediction performance | Helped |
| Represent the non-Gaussian finite-network prior rather than using the associated GP | Quadratic FCN · Phase retrieval | training threshold: n = 2d | Helped |
| Use a finite-channel CNN rather than the GP limit | Linear CNN teacher-student model · Feature-learning regime | prediction performance | Helped |
| Increase training-set size | Self-consistent GP theory · Linear CNN teacher-student model | prediction agreement | Helped |
| Reduce channel count below the critical value | Linear CNN student · Feature-learning phase | teacher-feature signature in hidden-weight spectrum | Helped |
Closest papers by meaning
- Gaussian Process Behaviour in Wide Deep Neural Networks (ICLR 2018 · similarity 0.89)
- Critical feature learning in deep neural networks (ICML 2024 · similarity 0.87)
- Feature Learning Dynamics in Infinite-Depth Neural Networks (NeurIPS 2026 · similarity 0.86)
- Asymptotics of feature learning in two-layer networks after one gradient-step (ICML 2024 · similarity 0.86)
- Asymptotic Analysis of Two-Layer Neural Networks after One Gradient Step under Gaussian Mixtures Data with Structure (ICLR 2025 · similarity 0.86)
- Beyond IID weights: sparse and low-rank deep Neural Networks are also Gaussian Processes (ICLR 2024 · 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.