Research map / Optimization / Implicit bias and generalization
A Kernel Perspective of Skip Connections in Convolutional 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 |
|---|---|---|---|
| Add skip connections | Convolutional networks · Uniform multi-sphere inputs | eigenvalue decay rate | No clear effect |
| Add skip connections | Convolutional networks · Effective receptive field | locality bias | Helped |
| Remove skip connections (ablation) | ResNet kernels · Effective receptive field | locality bias | Hurt |
| Add skip connections | CGPK kernel matrices · Finite depths | condition number | Helped |
| Set the balancing parameter to α=L^-γ, with γ in (0.5,1] (ablation) | ResCNTK · As depth tends to infinity | kernel richness | Hurt |
| Use α=1 instead of a decaying balancing parameter (ablation) | Convolutional residual architectures · Large depth | kernel richness | Helped |
Closest papers by meaning
- Towards Understanding the Importance of Shortcut Connections in Residual Networks (NeurIPS 2019 · similarity 0.89)
- On the Spectral Bias of Convolutional Neural Tangent and Gaussian Process Kernels (NeurIPS 2022 · similarity 0.89)
- Residual Alignment: Uncovering the Mechanisms of Residual Networks (NeurIPS 2023 · similarity 0.88)
- Deep Convolutional Networks as shallow Gaussian Processes (ICLR 2019 · similarity 0.88)
- Skip-Convolutions for Efficient Video Processing (CVPR 2021 · similarity 0.87)
- Revisiting Residual Connections: Orthogonal Updates for Stable and Efficient Deep Networks (NeurIPS 2025 · 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.