Research map / Optimization / Implicit bias and generalization
Learning in the Presence of Low-dimensional Structure: A Spiked Random Matrix Perspective
Jimmy Ba, Murat A Erdogdu, Taiji Suzuki, Zhichao Wang, Denny Wu
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 spike magnitude above d^(1-1/p) threshold for KRR with pure Hermite link (k=p=2) | KRR (Gaussian RBF kernel) · Synthetic spiked covariance, sigma*=h2 | prediction risk | Improved |
| use spike magnitude above d^(1-1/k) threshold for two-layer NN with pure Hermite link (k=p=3) | Two-layer NN (one GD step + ridge) · Synthetic spiked covariance, sigma*=h3 | prediction risk | Improved |
| switch from KRR to GD-trained NN when k<p (sigma*=h1+h3) | Two-layer NN vs KRR · Synthetic spiked covariance, p=3,k=1 | prediction risk | Improved |
| switch from KRR to GD-trained NN when k<p (sigma*=h2+h3) | Two-layer NN vs KRR · Synthetic spiked covariance, p=3,k=2 | prediction risk | Improved |
| switch from KRR to NN when k=p (pure Hermite link) | Two-layer NN vs KRR · Synthetic spiked covariance, p=k=2 and 3 | prediction risk | No clear effect |
Closest papers by meaning
- Neural network learns low-dimensional polynomials with SGD near the information-theoretic limit (NeurIPS 2024 · similarity 0.89)
- Low Rank Gradients and Where to Find Them (NeurIPS 2025 · similarity 0.88)
- Learning quadratic neural networks in high dimensions: SGD dynamics and scaling laws (NeurIPS 2025 · similarity 0.86)
- Learning Reveals Invisible Structure in Low-Rank RNNs (NeurIPS 2026 · similarity 0.86)
- A Theory of Non-Linear Feature Learning with One Gradient Step in Two-Layer Neural Networks (ICML 2024 · similarity 0.86)
- Asymptotics of feature learning in two-layer networks after one gradient-step (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.
