Research map / Learning theory and algorithms / Kernels and regression theory
Learning sparse features can lead to overfitting in neural networks
Leonardo Petrini, Francesco Cagnetta, Eric Vanden-Eijnden, Matthieu Wyart
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 lazy NTK training instead of feature learning | One-hidden-layer ReLU network · Smooth Gaussian random functions on the d-dimensional unit sphere | generalization error exponent β | Improved |
| Use RFK lazy training instead of feature learning | One-hidden-layer ReLU network · Smooth Gaussian random functions on the d-dimensional unit sphere | generalization error exponent β | Improved |
| Learn sparse features rather than use lazy training | Fully-connected networks on image datasets · Benchmark image datasets | relative sensitivity to small diffeomorphisms | Hurt |
| Learn sparse features rather than use lazy training | Fully-connected networks on image datasets · Benchmark image datasets | classification performance | Hurt |
| Use feature learning instead of lazy training | One-hidden-layer ReLU network · Gaussian random functions on the d-dimensional unit sphere with insufficiently smooth targets | generalization error exponent β | No clear effect |
Closest papers by meaning
- How Feature Learning Can Improve Neural Scaling Laws (ICLR 2025 · similarity 0.87)
- Grokking as the transition from lazy to rich training dynamics (ICLR 2024 · similarity 0.87)
- From Kernels to Features: A Multi-Scale Adaptive Theory of Feature Learning (ICML 2025 · similarity 0.87)
- How Deep Networks Learn Sparse and Hierarchical Data: the Sparse Random Hierarchy Model (ICML 2024 · similarity 0.87)
- Dynamic Sparse Training versus Dense Training: The Unexpected Winner in Image Corruption Robustness (ICLR 2025 · similarity 0.86)
- Complexity Matters: Feature Learning in the Presence of Spurious Correlations (ICML 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.
