TIDE: Time Derivative Diffusion for Deep Learning on Graphs

Maysam Behmanesh, Maximilian Krahn, Maks Ovsjanikov
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:2015-2030, 2023.

Abstract

A prominent paradigm for graph neural networks is based on the message-passing framework. In this framework, information communication is realized only between neighboring nodes. The challenge of approaches that use this paradigm is to ensure efficient and accurate long-distance communication between nodes, as deep convolutional networks are prone to over smoothing. In this paper, we present a novel method based on time derivative graph diffusion (TIDE) to overcome these structural limitations of the message-passing framework. Our approach allows for optimizing the spatial extent of diffusion across various tasks and network channels, thus enabling medium and long-distance communication efficiently. Furthermore, we show that our architecture design also enables local message-passing and thus inherits from the capabilities of local message-passing approaches. We show that on both widely used graph benchmarks and synthetic mesh and graph datasets, the proposed framework outperforms state-of-the-art methods by a significant margin.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-behmanesh23a, title = {{TIDE}: Time Derivative Diffusion for Deep Learning on Graphs}, author = {Behmanesh, Maysam and Krahn, Maximilian and Ovsjanikov, Maks}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {2015--2030}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/behmanesh23a/behmanesh23a.pdf}, url = {https://proceedings.mlr.press/v202/behmanesh23a.html}, abstract = {A prominent paradigm for graph neural networks is based on the message-passing framework. In this framework, information communication is realized only between neighboring nodes. The challenge of approaches that use this paradigm is to ensure efficient and accurate long-distance communication between nodes, as deep convolutional networks are prone to over smoothing. In this paper, we present a novel method based on time derivative graph diffusion (TIDE) to overcome these structural limitations of the message-passing framework. Our approach allows for optimizing the spatial extent of diffusion across various tasks and network channels, thus enabling medium and long-distance communication efficiently. Furthermore, we show that our architecture design also enables local message-passing and thus inherits from the capabilities of local message-passing approaches. We show that on both widely used graph benchmarks and synthetic mesh and graph datasets, the proposed framework outperforms state-of-the-art methods by a significant margin.} }
Endnote
%0 Conference Paper %T TIDE: Time Derivative Diffusion for Deep Learning on Graphs %A Maysam Behmanesh %A Maximilian Krahn %A Maks Ovsjanikov %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-behmanesh23a %I PMLR %P 2015--2030 %U https://proceedings.mlr.press/v202/behmanesh23a.html %V 202 %X A prominent paradigm for graph neural networks is based on the message-passing framework. In this framework, information communication is realized only between neighboring nodes. The challenge of approaches that use this paradigm is to ensure efficient and accurate long-distance communication between nodes, as deep convolutional networks are prone to over smoothing. In this paper, we present a novel method based on time derivative graph diffusion (TIDE) to overcome these structural limitations of the message-passing framework. Our approach allows for optimizing the spatial extent of diffusion across various tasks and network channels, thus enabling medium and long-distance communication efficiently. Furthermore, we show that our architecture design also enables local message-passing and thus inherits from the capabilities of local message-passing approaches. We show that on both widely used graph benchmarks and synthetic mesh and graph datasets, the proposed framework outperforms state-of-the-art methods by a significant margin.
APA
Behmanesh, M., Krahn, M. & Ovsjanikov, M.. (2023). TIDE: Time Derivative Diffusion for Deep Learning on Graphs. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:2015-2030 Available from https://proceedings.mlr.press/v202/behmanesh23a.html.

Related Material