Dynamic Graph Neural Networks Under Spatio-Temporal Distribution Shift

Authors: Zeyang Zhang, Xin Wang, Ziwei Zhang, Haoyang Li, Zhou Qin, Wenwu Zhu

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on three real-world datasets and one synthetic dataset demonstrate the superiority of our method over state-of-the-art baselines under distribution shifts.
Researcher Affiliation Collaboration 1Tsinghua University, 2Alibaba Group
Pseudocode Yes Algorithm 1 Training pipeline for DIDA
Open Source Code Yes 3Our codes are publicly available at https://github.com/wondergo2017/DIDA
Open Datasets Yes COLLAB [51]4 is an academic collaboration dataset... https://www.aminer.cn/collaboration. Yelp [43]5 is a business review dataset... https://www.yelp.com/dataset
Dataset Splits Yes To measure models performance under spatio-temporal distribution shift, we choose one field as w/ DS and the left others are further split into training, validation and test data ( w/o DS ) chronologically.
Hardware Specification No The paper discusses computational complexity but does not provide specific details on the hardware used for experiments, such as GPU or CPU models.
Software Dependencies No The paper does not provide specific version numbers for software dependencies or libraries used in the implementation.
Experiment Setup No The paper states 'More Details of the settings and other results can be found in Appendix' but does not include specific hyperparameters such as learning rate, batch size, or optimizer settings within the main text.