Research map / Optimization / Implicit bias and generalization
The Dynamics of Learning: A Random Matrix Approach
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 |
|---|---|---|---|
| stop gradient descent early rather than continue toward the least-squares solution | single-layer linear classifier · Gaussian-mixture binary classification | generalization error | Helped |
| continue training toward the least-squares solution instead of stopping early | single-layer linear classifier · Gaussian-mixture binary classification | training error | Helped |
| use smaller random-initialization variance | single-layer linear classifier · Gaussian-mixture binary classification | generalization performance | Helped |
| train with a sample-to-dimension ratio closer to 1 | least-squares solution · Gaussian-mixture binary classification | generalization performance | Hurt |
| increase training time when the sample-to-dimension ratio is zero | single-layer linear classifier · Gaussian-mixture binary classification, c = 0 | generalization performance | Helped |
Closest papers by meaning
- Deep Linear Network Training Dynamics from Random Initialization: Data, Width, Depth, and Hyperparameter Transfer (ICML 2025 · similarity 0.89)
- Mixed Dynamics In Linear Networks: Unifying the Lazy and Active Regimes (NeurIPS 2024 · similarity 0.88)
- Three Mechanisms of Feature Learning in a Linear Network (ICLR 2025 · similarity 0.88)
- Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer Networks (NeurIPS 2025 · similarity 0.88)
- Dynamics of Supervised and Reinforcement Learning in the Non-Linear Perceptron (NeurIPS 2024 · similarity 0.87)
- From Lazy to Rich: Exact Learning Dynamics in Deep Linear Networks (ICLR 2025 · similarity 0.87)
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.