Uncertainty Estimation for GNNs based on SPDEs

This repository provides the source code for uncertainty estimation in Graph Neural Networks (GNNs) on both static and dynamic graphs based on specific SPDEs.

Datasets

Static Graph Datasets

Cora: Downloadable from https://github.com/kimiyoung/planetoid/tree/master/data

CiteSeer: Downloadable from https://github.com/kimiyoung/planetoid/tree/master/data

Pubmed: Downloadable from https://github.com/kimiyoung/planetoid/tree/master/data

OGBN-Arxiv: Downloadable from https://ogb.stanford.edu/docs/nodeprop/

Amazon-Computers: Downloadable from https://github.com/shchur/gnn-benchmark/

Dynamic Graph Datasets

BC-OTC: Downloadable from http://snap.stanford.edu/data/soc-sign-bitcoin-otc.html

Reddit: Downloadable from http://snap.stanford.edu/data/soc-RedditHyperlinks.html

UCI: Downloadable from http://konect.uni-koblenz.de/networks/opsahl-ucsocial

AS: Downloadable from http://snap.stanford.edu/data/as-733.html

Elliptic: Downloadable from https://www.kaggle.com/ellipticco/elliptic-data-set

Brain: Downloadable from https://www.dropbox.com/sh/33p0gk4etgdjfvz/AACe2INXtp3N0u9xRdszq4vua?dl=0

Code Usage

For static graphs: python -u main.py --use_bn --dataset --ood_type --lr --weight_decay --input_dropout --device

For dynamic graphs: python main_dgnn.py --config_file --OOD

References

[1] Xixun Lin, Wenxiao Zhang, Fengzhao Shi, Chuan Zhou, Lixin Zou, Xiangyu Zhao, Dawei Yin, Shirui Pan, Yanan Cao. Graph Neural Stochastic Diffusion for Estimating Uncertainty in Node Classification. International Conference on Machine Learning (ICML 2024)

[2] Xixun Lin, Zhiheng Zhou, Yanan Cao, Chuan Zhou, Nan Sun, Ge Zhang, Xiangyu Zhao, Peng Zhang, Peilin Zhao, Shirui Pan, Philip S. Yu. Graph Stochastic Jump-Diffusion Equation for Uncertainty Estimation on Dynamic Graphs. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI, under review)