Graph-Structured Gaussian Processes for Transferable Graph Learning

Authors: Jun Wu, Lisa Ainsworth, Andrew Leakey, Haixun Wang, Jingrui He

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on several transferable graph learning benchmarks demonstrate the efficacy of Graph GP over state-of-the-art Gaussian process baselines.
Researcher Affiliation Collaboration 1University of Illinois at Urbana-Champaign 2USDA ARS Global Change and Photosynthesis Research Unit 3Instacart
Pseudocode Yes Algorithm 1 Graph GP
Open Source Code Yes Code is available at https://github.com/jwu4sml/Graph GP.
Open Datasets Yes Twitch [44]: It has 6 different domains (...). Agriculture [34, 60]: It has 3 different domains (...). Airports [43]: It has 3 different domains (...). Wikipedia [44]: It has 3 different domains (...). Web KB [41]: It has 3 different domains (...).
Dataset Splits Yes For Airport, Wikipedia, and Web KB data sets, we randomly select 10% of target nodes for the training set, 10% for the validation set, and 80% for the testing set. For Agriculture and Twitch data sets, we randomly select 1% of target nodes for the training set, 1% for the validation set, and 98% for the testing set.
Hardware Specification Yes All the experiments are performed on a Windows machine with four 3.80GHz Intel Cores, 64GB RAM, and two NVIDIA Quadro RTX 5000 GPUs.
Software Dependencies No The paper mentions using "GPy Torch [16]" and "Adam [24]" but does not provide specific version numbers for these software dependencies, which is required for reproducibility.
Experiment Setup Yes The hyperparameters are optimized using Adam [24] with a learning rate of 0.01 and a total number of training epochs of 500.