Research map / Optimization / Implicit bias and generalization
Random Smoothing Regularization in Kernel Gradient Descent Learning
Liang Ding, Tianyang Hu, Jiahang Jiang, Donghao Li, Wenjia Wang, Yuan Yao
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 |
|---|---|---|---|
| add Gaussian random smoothing (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=1, early stopping, n=50 | test l2 loss: 2.2168e-03 -> 1.3468e-03 | Improved |
| add Laplace random smoothing (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=1, weight decay, n=200 | test l2 loss: 1.1135e-03 -> 8.2806e-04 | Improved |
| add Gaussian random smoothing (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=2, early stopping, n=50 | test l2 loss: 1.2628e-02 -> 6.7205e-03 | Improved |
| add Laplace random smoothing (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=2, early stopping, n=200 | test l2 loss: 3.1396e-03 -> 1.7674e-03 | Improved |
| add Gaussian random smoothing (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=3, weight decay, n=50 | test l2 loss: 2.0987e-02 -> 1.6498e-02 | Improved |
| add Laplace random smoothing (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=3, early stopping, n=200 | test l2 loss: 4.9460e-03 -> 3.8615e-03 | Improved |
| use Gaussian instead of Laplace noise (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=1, early stopping, n=50 | test l2 loss: 2.0638e-03 -> 1.3468e-03 | Improved |
| use Laplace instead of Gaussian noise (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=1, weight decay, n=50 | test l2 loss: 1.7466e-03 -> 1.6765e-03 | Improved |
| vary smoothing scale (U-shaped) (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=1 | test l2 loss: U-shaped curve | Mixed |
| increase training size from 50 to 200 (ablation) | 2-layer ReLU MLP trained with SGD · Synthetic D=1, optimal smoothing scale | test l2 loss: optimal scale decreases | Improved |
Closest papers by meaning
- Sinkhorn Normalization of Diffusion Kernels (ICML 2026 · similarity 0.87)
- Fast kernel methods: Sobolev, physics-informed, and additive models (ICML 2026 · similarity 0.87)
- Sobolev Regularized MMD Gradient Flow (NeurIPS 2026 · similarity 0.87)
- Robust Representation Consistency Model via Contrastive Denoising (ICLR 2025 · similarity 0.86)
- Generalization in Kernel Regression Under Realistic Assumptions (ICML 2024 · similarity 0.86)
- The Lipschitz-Variance-Margin Tradeoff for Enhanced Randomized Smoothing (ICLR 2024 · similarity 0.85)
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.
