Research map / Optimization / Implicit bias and generalization
The Implicit Regularization of Momentum Gradient Descent in Overparametrized Models
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 MGF with calibration λ = 2/t² instead of ridge | Least-squares regression · Least squares | estimation risk relative to ridge: 1.5376 | Mixed |
| Tune MGF at t = √(2α) instead of optimally tuned ridge | MGF · Least squares, spherical prior | Bayes risk relative to ridge: 1 ≤ inf_t>0 Risk(ˆβmgf(t))/inf_λ>0 Risk(ˆβridge(λ)) < 1.035 | Mixed |
| Use MGF rather than GF | Least-squares regression · Least squares, spherical prior | maximum ratio of Bayes risk to ridge: 1.3663 -> 1.1097 | Helped |
| Use MGF rather than GF | Least-squares regression · Least squares, spherical prior | maximum ratio of optimal Bayes risk to ridge: 1.0914 -> 1.0208 | Helped |
| Increase depth from 1 to 2 or greater | MGF on diagonal linear networks · Depth-N linear model | evolution-rate gap between weight components | Mixed |
| Apply early stopping | MGD on depth-N linear models with small initialization · Depth-N linear model | sparsity of learned solution | Helped |
Closest papers by meaning
- Does Momentum Change the Implicit Regularization on Separable Data? (NeurIPS 2022 · similarity 0.90)
- Algorithmic Regularization in Learning Deep Homogeneous Models: Layers are Automatically Balanced (NeurIPS 2018 · similarity 0.90)
- Implicit regularization in Heavy-ball momentum accelerated stochastic gradient descent (ICLR 2023 · similarity 0.90)
- Effects of Momentum in Implicit Bias of Gradient Flow for Diagonal Linear Networks (AAAI 2025 · similarity 0.89)
- Combining Explicit and Implicit Regularization for Efficient Learning in Deep Networks (NeurIPS 2022 · similarity 0.89)
- The Marginal Value of Momentum for Small Learning Rate SGD (ICLR 2024 · 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.