A Kernel Perspective of Skip Connections in Convolutional Networks
Authors: Daniel Barzilai, Amnon Geifman, Meirav Galun, Ronen Basri
ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Figure 1: Left: The eigenvalues of Res CGPK (filled circles and solid lines) computed numerically for various eigenfunctions... Figure 2: The effective receptive field (ERF) induced by Res CNTK compared to that of a network and to the kernel and network with the skip connections removed. We followed (Luo et al., 2016) in computing the ERF, where the networks are first trained on CIFAR-10. |
| Researcher Affiliation | Academia | Daniel Barzilai, Amnon Geifman, Meirav Galun & Ronen Basri Weizmann Institute of Science {daniel.barzilai,amnon.geifman,meirav.galun,ronen.basri}@weizmann.ac.il |
| Pseudocode | No | The paper does not contain any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | No | The paper does not provide an explicit statement about open-source code availability or a link to a code repository for its methodology. |
| Open Datasets | Yes | Figure 2: The effective receptive field (ERF) induced by Res CNTK compared to that of a network and to the kernel and network with the skip connections removed. We followed (Luo et al., 2016) in computing the ERF, where the networks are first trained on CIFAR-10. |
| Dataset Splits | No | The paper mentions training on 'CIFAR-10' in Figure 2 caption, but does not specify any dataset splits (e.g., percentages, sample counts, or specific split files) for training, validation, or testing. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers needed to replicate the experiments. |
| Experiment Setup | Yes | Figure 1 specifies 'L = 3, q = 2, d = 4'. Figure 2 mentions 'L = 8'. Figure 3 specifies 'n = 100'. Appendix G confirms: 'Thus the kernel computed is with d = 4, q = 2, α = 1'. |