Research map / Optimization / Implicit bias and generalization
A unifying view on implicit bias in training linear 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 different linear network architectures while matching the initial linear coefficients | Fully-connected, diagonal, and convolutional networks · Single-point regression, x=[1 2], y=1 | limit-point agreement | Helped |
| Decrease the initialization scale | Fully-connected, diagonal, and convolutional networks · Single-point regression, x=[1 2], y=1 | dependence of the solution on initial directions | Helped |
| Train for a finite number of iterations rather than taking the zero-loss limit | Fully-connected, diagonal, and convolutional networks · Two-point classification | dependence of coefficients on initialization | Hurt |
Closest papers by meaning
- Implicit Bias of the Step Size in Linear Diagonal Neural Networks (ICML 2022 · similarity 0.91)
- Implicit Bias of Mirror Flow for Shallow Neural Networks in Univariate Regression (ICLR 2025 · similarity 0.91)
- Flavors of Margin: Implicit Bias of Steepest Descent in Homogeneous Neural Networks (ICLR 2025 · similarity 0.89)
- Implicit Bias of Gradient Descent for Non-Homogeneous Deep Networks (ICML 2025 · similarity 0.88)
- Deep linear networks for regression are implicitly regularized towards flat minima (NeurIPS 2024 · similarity 0.88)
- Implicit Regularization for Tubal Tensor Factorizations via Gradient Descent (ICML 2025 · 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.