General E(2)-Equivariant Steerable CNNs
Authors: Maurice Weiler, Gabriele Cesa
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We implement a wide range of previously proposed and entirely new equivariant network architectures and extensively compare their performances. E(2)-steerable convolutions are further shown to yield remarkable gains on CIFAR-10, CIFAR-100 and STL-10 when used as drop in replacement for non-equivariant convolutions. |
| Researcher Affiliation | Academia | Maurice Weiler University of Amsterdam, QUVA Lab m.weiler@uva.nl Gabriele Cesa University of Amsterdam cesa.gabriele@gmail.com |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. It describes processes in narrative text. |
| Open Source Code | Yes | Our implementation is provided as a Py Torch extension which is available at https://github.com/QUVA-Lab/e2cnn. |
| Open Datasets | Yes | All benchmarked models are evaluated on three different versions of the MNIST dataset, each containing 12000 training and 50000 test images. [...] classifying CIFAR-10 and CIFAR-100. [...] run experiments on STL-10 [37]. |
| Dataset Splits | No | The paper specifies training and test image counts for MNIST but does not explicitly mention a separate validation dataset split with specific numbers or percentages within the provided text. For CIFAR and STL-10, it references other papers for training procedures without detailing splits here. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU model, CPU type, memory) used for running its experiments. It describes the methods and results but omits hardware specifications. |
| Software Dependencies | No | The paper mentions that their implementation is a 'PyTorch extension' but does not provide specific version numbers for PyTorch or any other software dependencies. Without version numbers, reproducibility of the software environment is not fully described. |
| Experiment Setup | No | The paper states, 'All runs use the same training procedure as reported in [34] and Appendix K.3.' and 'As in the CIFAR experiments, we adopt the training settings and hyperparameters of [36] without changes.' This delegates the specific experimental setup details (like hyperparameters) to referenced papers or appendices not included in the provided text, meaning they are not explicitly stated within this paper. |