PDO-eConvs: Partial Differential Operator Based Equivariant Convolutions
Authors: Zhengyang Shen, Lingshen He, Zhouchen Lin, Jinwen Ma
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on rotated MNIST and natural image classification show that PDO-e Convs perform competitively yet use parameters much more efficiently. |
| Researcher Affiliation | Academia | 1School of Mathematical Sciences and LMAM, Peking University, Beijing 100871 2Key Lab. of Machine Perception (Mo E), School of EECS, Peking University, Beijing 100871. Correspondence to: Zhouchen Lin <zlin@pku.edu.cn>, Jinwen Ma <jwma@math.pku.edu.cn>. |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described in this paper. |
| Open Datasets | Yes | The most commonly used dataset for validating rotationequivariant algorithms is MNIST-rot-12k (Larochelle et al., 2007). ... CIFAR-10 (C10) and CIFAR-100 (C100) (Krizhevsky & Hinton, 2009) |
| Dataset Splits | Yes | We randomly select 2,000 training images as a validation set. ... The training and the test sets contain 50,000 and 10,000 images, respectively. We randomly select 5,000 training images as a validation set. |
| Hardware Specification | No | The paper does not provide specific hardware details used for running its experiments. |
| Software Dependencies | No | all the experiments are implemented using Tensorflow. |
| Experiment Setup | Yes | The model is trained using the Adam algorithm (Kingma & Ba, 2015) with a weight decay of 0.01. ... We train using batch size 128 for 200 epochs. The initial learning rate is set to 0.001 and is divided by 10 at 50% and 75% of the total number of training epochs. We set the dropout rate as 0.2. |