Class-Imbalanced Graph Learning without Class Rebalancing

Zhining Liu, Ruizhong Qiu, Zhichen Zeng, Hyunsik Yoo, David Zhou, Zhe Xu, Yada Zhu, Kommy Weldemariam, Jingrui He, Hanghang Tong
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:31747-31772, 2024.

Abstract

Class imbalance is prevalent in real-world node classification tasks and poses great challenges for graph learning models. Most existing studies are rooted in a class-rebalancing (CR) perspective and address class imbalance with class-wise reweighting or resampling. In this work, we approach the root cause of class-imbalance bias from an topological paradigm. Specifically, we theoretically reveal two fundamental phenomena in the graph topology that greatly exacerbate the predictive bias stemming from class imbalance. On this basis, we devise a lightweight topological augmentation framework BAT to mitigate the class-imbalance bias without class rebalancing. Being orthogonal to CR, BAT can function as an efficient plug-and-play module that can be seamlessly combined with and significantly boost existing CR techniques. Systematic experiments on real-world imbalanced graph learning tasks show that BAT can deliver up to 46.27% performance gain and up to 72.74% bias reduction over existing techniques. Code, examples, and documentations are available at https://github.com/ZhiningLiu1998/BAT.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-liu24ay, title = {Class-Imbalanced Graph Learning without Class Rebalancing}, author = {Liu, Zhining and Qiu, Ruizhong and Zeng, Zhichen and Yoo, Hyunsik and Zhou, David and Xu, Zhe and Zhu, Yada and Weldemariam, Kommy and He, Jingrui and Tong, Hanghang}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {31747--31772}, 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/liu24ay/liu24ay.pdf}, url = {https://proceedings.mlr.press/v235/liu24ay.html}, abstract = {Class imbalance is prevalent in real-world node classification tasks and poses great challenges for graph learning models. Most existing studies are rooted in a class-rebalancing (CR) perspective and address class imbalance with class-wise reweighting or resampling. In this work, we approach the root cause of class-imbalance bias from an topological paradigm. Specifically, we theoretically reveal two fundamental phenomena in the graph topology that greatly exacerbate the predictive bias stemming from class imbalance. On this basis, we devise a lightweight topological augmentation framework BAT to mitigate the class-imbalance bias without class rebalancing. Being orthogonal to CR, BAT can function as an efficient plug-and-play module that can be seamlessly combined with and significantly boost existing CR techniques. Systematic experiments on real-world imbalanced graph learning tasks show that BAT can deliver up to 46.27% performance gain and up to 72.74% bias reduction over existing techniques. Code, examples, and documentations are available at https://github.com/ZhiningLiu1998/BAT.} }
Endnote
%0 Conference Paper %T Class-Imbalanced Graph Learning without Class Rebalancing %A Zhining Liu %A Ruizhong Qiu %A Zhichen Zeng %A Hyunsik Yoo %A David Zhou %A Zhe Xu %A Yada Zhu %A Kommy Weldemariam %A Jingrui He %A Hanghang Tong %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-liu24ay %I PMLR %P 31747--31772 %U https://proceedings.mlr.press/v235/liu24ay.html %V 235 %X Class imbalance is prevalent in real-world node classification tasks and poses great challenges for graph learning models. Most existing studies are rooted in a class-rebalancing (CR) perspective and address class imbalance with class-wise reweighting or resampling. In this work, we approach the root cause of class-imbalance bias from an topological paradigm. Specifically, we theoretically reveal two fundamental phenomena in the graph topology that greatly exacerbate the predictive bias stemming from class imbalance. On this basis, we devise a lightweight topological augmentation framework BAT to mitigate the class-imbalance bias without class rebalancing. Being orthogonal to CR, BAT can function as an efficient plug-and-play module that can be seamlessly combined with and significantly boost existing CR techniques. Systematic experiments on real-world imbalanced graph learning tasks show that BAT can deliver up to 46.27% performance gain and up to 72.74% bias reduction over existing techniques. Code, examples, and documentations are available at https://github.com/ZhiningLiu1998/BAT.
APA
Liu, Z., Qiu, R., Zeng, Z., Yoo, H., Zhou, D., Xu, Z., Zhu, Y., Weldemariam, K., He, J. & Tong, H.. (2024). Class-Imbalanced Graph Learning without Class Rebalancing. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:31747-31772 Available from https://proceedings.mlr.press/v235/liu24ay.html.

Related Material