Inductive Graph Neural Networks for Spatiotemporal Kriging

Authors: Yuankai Wu, Dingyi Zhuang, Aurelie Labbe, Lijun Sun4478-4485

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical results on several real-world spatiotemporal datasets demonstrate the effectiveness of our model.
Researcher Affiliation Academia 1Mc Gill University, Montreal, Canada 2HEC Montreal, Montreal, Canada
Pseudocode Yes Algorithm 1 Subgraph signal and random mask generation
Open Source Code No The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets Yes METR-LA is a traffic speed dataset from 207 sensors in Los Angeles over four months (Mar 1, 2012 to Jun 30, 2012);... The two datasets [METR-LA and Pe MS-Bay] are used side by side in (Li et al. 2017).
Dataset Splits Yes For IGNNK, we take data from the first 70% of the time points as a training set X and test the kriging performance on the following 30% of the time points. ... We use the first 70% of the time points from unsampled nodes as a validation set to select the optimal hyperparameters, and evaluate recovery performance on the following 30% of the time points.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments.
Software Dependencies No The paper does not provide specific software dependencies or version numbers used for the experiments.
Experiment Setup Yes We choose h = 24 (i.e., 2 h) for the three traffic datasets, h = 16 (i.e., 80 min) for NREL, and h = 6 (i.e., 6 months) for USHCN. We perform kriging using a sliding-window approach.