Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Equivariant Networks for Pixelized Spheres
Authors: Mehran Shakerinava, Siamak Ravanbakhsh
ICML 2021 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our empirical study using different pixelizations of the sphere demonstrates the effectiveness of our choice of approximation for spherical symmetry, where we report state-of-the-art on popular benchmarks for omnidirectional semantic segmentation and segmentation of extreme climate events. Code is available at https://git.io/JGi ZA. |
| Researcher Affiliation | Academia | 1School of Computer Science, Mc Gill University, Montreal, Canada 2Mila Quebec AI Institute. Correspondence to: Mehran Shakerinava <EMAIL>. |
| Pseudocode | Yes | Algorithm 1 in the Appendix gives the pseudocode for ο¬nding the orbit of a given element a A. |
| Open Source Code | Yes | Code is available at https://git.io/JGi ZA. |
| Open Datasets | Yes | The spherical MNIST dataset (Cohen et al., 2018) consists of images from the MNIST dataset projected onto a sphere with random rotation. ... The data is accessible at https://portal.nersc.gov/ project/dasrepo/deepcam/segm_h5_v3_reformat. |
| Dataset Splits | Yes | The training, validation, and test set size is 43917, 6275, and 12549, respectively. |
| Hardware Specification | Yes | All models were trained on NVIDIA Tesla V100 GPUs. |
| Software Dependencies | No | The paper mentions software like 'group convolution' and references related works, but does not specify any software names with version numbers (e.g., PyTorch 1.9, Python 3.8) that are crucial for reproduction. |
| Experiment Setup | Yes | Details of training and architectures appear in Appendix D and Appendix E. ... Appendix E provides specific hyperparameters such as learning rate, batch size, epochs, optimizer, and weight decay for each experiment. |