Revisiting Pre-Propagation GNNs: Robust Diffusion Operators and Hidden-State Re-Propagation

Zichao Yue, Zhiru Zhang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:152309-152327, 2026.

Abstract

Pre-propagation graph neural networks (PP-GNNs) decouple node feature propagation from transformation: graph diffusion is performed once as preprocessing, and training reduces to dense per-node transformations. This design enables mini-batch training without inter-node dependencies, avoids repeated sparse matrix–matrix multiplications, and better matches modern accelerators optimized for dense compute. However, their expressivity remains unclear, and empirical results show a gap between PP-GNNs and their message-passing counterparts on commonly used graph benchmarks, especially heterophilic ones. In this paper, we propose a suite of robust graph diffusion operators for preprocessing and a few-shot hidden-state re-propagation scheme during training. Our methods improve the validation and test accuracy of PP-GNNs, enabling them to match the accuracy of message-passing GNNs while maintaining training efficiency.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-yue26c, title = {Revisiting Pre-Propagation {GNN}s: Robust Diffusion Operators and Hidden-State Re-Propagation}, author = {Yue, Zichao and Zhang, Zhiru}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {152309--152327}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/yue26c/yue26c.pdf}, url = {https://proceedings.mlr.press/v306/yue26c.html}, abstract = {Pre-propagation graph neural networks (PP-GNNs) decouple node feature propagation from transformation: graph diffusion is performed once as preprocessing, and training reduces to dense per-node transformations. This design enables mini-batch training without inter-node dependencies, avoids repeated sparse matrix–matrix multiplications, and better matches modern accelerators optimized for dense compute. However, their expressivity remains unclear, and empirical results show a gap between PP-GNNs and their message-passing counterparts on commonly used graph benchmarks, especially heterophilic ones. In this paper, we propose a suite of robust graph diffusion operators for preprocessing and a few-shot hidden-state re-propagation scheme during training. Our methods improve the validation and test accuracy of PP-GNNs, enabling them to match the accuracy of message-passing GNNs while maintaining training efficiency.} }
Endnote
%0 Conference Paper %T Revisiting Pre-Propagation GNNs: Robust Diffusion Operators and Hidden-State Re-Propagation %A Zichao Yue %A Zhiru Zhang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-yue26c %I PMLR %P 152309--152327 %U https://proceedings.mlr.press/v306/yue26c.html %V 306 %X Pre-propagation graph neural networks (PP-GNNs) decouple node feature propagation from transformation: graph diffusion is performed once as preprocessing, and training reduces to dense per-node transformations. This design enables mini-batch training without inter-node dependencies, avoids repeated sparse matrix–matrix multiplications, and better matches modern accelerators optimized for dense compute. However, their expressivity remains unclear, and empirical results show a gap between PP-GNNs and their message-passing counterparts on commonly used graph benchmarks, especially heterophilic ones. In this paper, we propose a suite of robust graph diffusion operators for preprocessing and a few-shot hidden-state re-propagation scheme during training. Our methods improve the validation and test accuracy of PP-GNNs, enabling them to match the accuracy of message-passing GNNs while maintaining training efficiency.
APA
Yue, Z. & Zhang, Z.. (2026). Revisiting Pre-Propagation GNNs: Robust Diffusion Operators and Hidden-State Re-Propagation. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:152309-152327 Available from https://proceedings.mlr.press/v306/yue26c.html.

Related Material