Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networks
Authors: Renan A. Rojas-Gomez, Teck-Yian Lim, Alex Schwing, Minh Do, Raymond A. Yeh
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate LPS on image classification and semantic segmentation. Experiments show that LPS is on-par with or outperforms existing methods in both performance and shift consistency. For the first time, we achieve true shift-equivariance on semantic segmentation (PASCAL VOC), i.e., 100% shift consistency, outperforming baselines by an absolute 3.3%. To validate the proposed LPS, we conduct extensive experiments: (a) image classification on CIFAR10 [21] and Image Net [12]; (b) semantic segmentation on PASCAL VOC [13]. |
| Researcher Affiliation | Academia | Department of Electrical Engineering, University of Illinois at Urbana-Champaign Department of Computer Science, Purdue University |
| Pseudocode | No | The paper describes the proposed model and its components in text and diagrams (e.g., Figure 3), but does not contain a structured pseudocode or algorithm block. |
| Open Source Code | Yes | Our project page and code are available at https://raymondyeh07.github.io/learnable_polyphase_sampling/ Please refer to the Python code included with the supplementary material. |
| Open Datasets | Yes | To validate the proposed LPS, we conduct extensive experiments: (a) image classification on CIFAR10 [21] and Image Net [12]; (b) semantic segmentation on PASCAL VOC [13]. |
| Dataset Splits | No | The paper states it uses CIFAR10, Image Net, and PASCAL VOC and that it follows prior works for training setup, and refers to Appendix Sec. A4 for more experimental details, but the provided text does not specify exact training/validation/test splits (e.g., percentages or counts) or refer to a specific citation for those splits. |
| Hardware Specification | No | The paper states that computational details are included in Appendix Sec. A4.4, but this appendix content is not provided in the main text, so specific hardware specifications are not explicitly detailed here. |
| Software Dependencies | No | The paper refers to PyTorch Lightning and PyTorch in its citations and mentions training details are in Appendix Sec. A4.4 and the attached code, but the provided text does not explicitly list specific version numbers for software dependencies such as libraries or frameworks. |
| Experiment Setup | No | The paper mentions architectural choices like "Res Net-18" and "Deep Lab V3+", and general training aspects like "circular padding" and "Gumbel Softmax temperature decay". However, it states that "more experimental details" and "hyperparameter details are included in Sec. A4.4 and the attached code", which are not provided in the main text. |