Research map / Optimization / Implicit bias and generalization
Implicit Gradient Regularization
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 |
|---|---|---|---|
| Increase the learning rate (ablation) | Gradient descent on 5-layer MLPs · MNIST | test accuracy | Helped |
| Increase network size (ablation) | Gradient descent on 5-layer MLPs · MNIST | test accuracy | Helped |
| Increase the learning rate (ablation) | Gradient descent on 5-layer MLPs · MNIST | loss surface slope | Helped |
| Increase the learning rate (ablation) | Gradient descent on 5-layer MLPs · MNIST | robustness to parameter perturbations | Helped |
| Increase the learning rate (ablation) | Gradient descent on ResNet-18 · CIFAR-10 | test accuracy | Helped |
| Add an explicit parameter-gradient penalty to the loss | ResNet-18 · CIFAR-10 | test accuracy: more than 12% | Helped |
| Increase the learning rate (ablation) | Gradient descent on the two-parameter linear model · Single training point | R_IG/E at the end of training | Helped |
| Use excessively large learning rates (ablation) | Gradient descent on the two-parameter linear model · Single training point | training stability | Hurt |
Closest papers by meaning
- Combining Explicit and Implicit Regularization for Efficient Learning in Deep Networks (NeurIPS 2022 · similarity 0.93)
- Estimating Implicit Regularization in Deep Learning (NeurIPS 2026 · similarity 0.90)
- On the Implicit Bias of Adam (ICML 2024 · similarity 0.90)
- Implicit Regularization for Tubal Tensor Factorizations via Gradient Descent (ICML 2025 · similarity 0.89)
- Improving Generalization and Convergence by Enhancing Implicit Regularization (NeurIPS 2024 · similarity 0.89)
- Implicit regularization of deep residual networks towards neural ODEs (ICLR 2024 · similarity 0.89)
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.