Graph Structure Extrapolation for Out-of-Distribution Generalization

Xiner Li, Shurui Gui, Youzhi Luo, Shuiwang Ji
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:27846-27874, 2024.

Abstract

Out-of-distribution (OOD) generalization deals with the prevalent learning scenario where test distribution shifts from training distribution. With rising application demands and inherent complexity, graph OOD problems call for specialized solutions. While data-centric methods exhibit performance enhancements on many generic machine learning tasks, there is a notable absence of data augmentation methods tailored for graph OOD generalization. In this work, we propose to achieve graph OOD generalization with the novel design of non-Euclidean-space linear extrapolation. The proposed augmentation strategy extrapolates structure spaces to generate OOD graph data. Our design tailors OOD samples for specific shifts without corrupting underlying causal mechanisms. Theoretical analysis and empirical results evidence the effectiveness of our method in solving target shifts, showing substantial and constant improvements across various graph OOD tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-li24y, title = {Graph Structure Extrapolation for Out-of-Distribution Generalization}, author = {Li, Xiner and Gui, Shurui and Luo, Youzhi and Ji, Shuiwang}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {27846--27874}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/li24y/li24y.pdf}, url = {https://proceedings.mlr.press/v235/li24y.html}, abstract = {Out-of-distribution (OOD) generalization deals with the prevalent learning scenario where test distribution shifts from training distribution. With rising application demands and inherent complexity, graph OOD problems call for specialized solutions. While data-centric methods exhibit performance enhancements on many generic machine learning tasks, there is a notable absence of data augmentation methods tailored for graph OOD generalization. In this work, we propose to achieve graph OOD generalization with the novel design of non-Euclidean-space linear extrapolation. The proposed augmentation strategy extrapolates structure spaces to generate OOD graph data. Our design tailors OOD samples for specific shifts without corrupting underlying causal mechanisms. Theoretical analysis and empirical results evidence the effectiveness of our method in solving target shifts, showing substantial and constant improvements across various graph OOD tasks.} }
Endnote
%0 Conference Paper %T Graph Structure Extrapolation for Out-of-Distribution Generalization %A Xiner Li %A Shurui Gui %A Youzhi Luo %A Shuiwang Ji %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-li24y %I PMLR %P 27846--27874 %U https://proceedings.mlr.press/v235/li24y.html %V 235 %X Out-of-distribution (OOD) generalization deals with the prevalent learning scenario where test distribution shifts from training distribution. With rising application demands and inherent complexity, graph OOD problems call for specialized solutions. While data-centric methods exhibit performance enhancements on many generic machine learning tasks, there is a notable absence of data augmentation methods tailored for graph OOD generalization. In this work, we propose to achieve graph OOD generalization with the novel design of non-Euclidean-space linear extrapolation. The proposed augmentation strategy extrapolates structure spaces to generate OOD graph data. Our design tailors OOD samples for specific shifts without corrupting underlying causal mechanisms. Theoretical analysis and empirical results evidence the effectiveness of our method in solving target shifts, showing substantial and constant improvements across various graph OOD tasks.
APA
Li, X., Gui, S., Luo, Y. & Ji, S.. (2024). Graph Structure Extrapolation for Out-of-Distribution Generalization. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:27846-27874 Available from https://proceedings.mlr.press/v235/li24y.html.

Related Material