Research map / Learning theory and algorithms / Kernels and regression theory
Benefits of Overparameterized Convolutional Residual Networks: Function Approximation under Smoothness Constraint
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 widening factor from 1 to 4 | WideResNet at depth 16 · CIFAR-10 test, PGD-20 | robust accuracy: 34.31 \pm 0.45\% -> 51.40 \pm 0.16\% | Helped |
| Increase widening factor from 1 to 4 | WideResNet at depth 22 · CIFAR-10 test, PGD-20 | robust accuracy: 41.47 \pm 0.11\% -> 53.36 \pm 0.21\% | Helped |
| Increase widening factor from 1 to 4 | WideResNet at depth 28 · CIFAR-10 test, PGD-20 | robust accuracy: 43.33 \pm 0.57\% -> 54.45 \pm 0.14\% | Helped |
| Increase widening factor from 1 to 4 | WideResNet at depth 34 · CIFAR-10 test, PGD-20 | robust accuracy: 46.14 \pm 0.16\% -> 55.28 \pm 0.35\% | Helped |
| Increase depth from 16 to 34 | WideResNet with WF 1 · CIFAR-10 test, PGD-20 | robust accuracy: 34.31 \pm 0.45\% -> 46.14 \pm 0.16\% | Helped |
| Increase depth from 16 to 34 | WideResNet with WF 2 · CIFAR-10 test, PGD-20 | robust accuracy: 46.14 \pm 0.21\% -> 51.63 \pm 0.28\% | Helped |
| Increase depth from 16 to 34 | WideResNet with WF 4 · CIFAR-10 test, PGD-20 | robust accuracy: 51.40 \pm 0.16\% -> 55.28 \pm 0.35\% | Helped |
Closest papers by meaning
- Nonparametric Classification on Low Dimensional Manifolds using Overparameterized Convolutional Residual Networks (NeurIPS 2024 · similarity 0.91)
- Besov Function Approximation and Binary Classification on Low-Dimensional Manifolds Using Convolutional Residual Networks (ICML 2021 · similarity 0.90)
- Approximation with CNNs in Sobolev Space: with Applications to Classification (NeurIPS 2022 · similarity 0.88)
- More is Better: when Infinite Overparameterization is Optimal and Overfitting is Obligatory (ICLR 2024 · similarity 0.88)
- Position: A Theory of Deep Learning Must Include Compositional Sparsity (ICML 2025 · similarity 0.86)
- Theoretical Analysis of Robust Overfitting for Wide DNNs: An NTK Approach (ICLR 2024 · similarity 0.86)
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.