Deformable Graph Convolutional Networks

Authors: Jinyoung Park, Sungdong Yoo, Jihwan Park, Hyunwoo J. Kim7949-7956

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our extensive experiments demonstrate that Deformable GCNs flexibly handles the heterophily and achieve the best performance in node classification tasks on six heterophilic graph datasets.
Researcher Affiliation Academia Jinyoung Park, Sungdong Yoo, Jihwan Park, Hyunwoo J. Kim* Department of Computer Science and Engineering, Korea University {lpmn678, ysd424, jseven7071, hyunwoojkim}@korea.ac.kr
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code Yes Our code is publicly available at https://github.com/mlvlab/Deformable GCN.
Open Datasets Yes For validating our model, we use six heterophilic graph datasets and three homophilic graph datasets... For all datasets, we apply the splits (48%/ 32%/ 20%)1 of nodes per class for (train/ validation/ test) provided by (Pei et al. 2020) for a fair comparison as (Zhu et al. 2020). 1https://github.com/graphdml-uiuc-jlu/geom-gcn.
Dataset Splits Yes For all datasets, we apply the splits (48%/ 32%/ 20%)1 of nodes per class for (train/ validation/ test) provided by (Pei et al. 2020) for a fair comparison as (Zhu et al. 2020). 1https://github.com/graphdml-uiuc-jlu/geom-gcn.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using the Adam optimizer but does not specify version numbers for any software dependencies or libraries (e.g., Python, PyTorch, CUDA).
Experiment Setup Yes We use the Adam optimizer (Kingma and Ba 2014) with ℓ2-regularization and 500 epochs for training our model and the baselines. The performance is reported with the best model on the validation datasets. All experiments are repeated 10 times as (Pei et al. 2020; Zhu et al. 2020) and accuracy is used as an evaluation metric.