On the Unreasonable Effectiveness of Feature Propagation in Learning on Graphs With Missing Node Features

Emanuele Rossi, Henry Kenlay, Maria I. Gorinova, Benjamin Paul Chamberlain, Xiaowen Dong, Michael M. Bronstein
Proceedings of the First Learning on Graphs Conference, PMLR 198:11:1-11:16, 2022.

Abstract

While Graph Neural Networks (GNNs) have recently become the de facto standard for modeling relational data, they impose a strong assumption on the availability of the node or edge features of the graph. In many real-world applications, however, features are only partially available; for example, in social networks, age and gender are available only for a small subset of users. We present a general approach for handling missing features in graph machine learning applications that is based on minimization of the Dirichlet energy and leads to a diffusion-type differential equation on the graph. The discretization of this equation produces a simple, fast and scalable algorithm which we call Feature Propagation. We experimentally show that the proposed approach outperforms previous methods on seven common node-classification benchmarks and can withstand surprisingly high rates of missing features: on average we observe only around 4% relative accuracy drop when 99% of the features are missing. Moreover, it takes only 10 seconds to run on a graph with ~2.5M nodes and ~23M edges on a single GPU. The code is available at https://github.com/twitter-research/feature-propagation.

Cite this Paper

BibTeX
@InProceedings{pmlr-v198-rossi22a, title = {On the Unreasonable Effectiveness of Feature Propagation in Learning on Graphs With Missing Node Features}, author = {Rossi, Emanuele and Kenlay, Henry and Gorinova, Maria I. and Chamberlain, Benjamin Paul and Dong, Xiaowen and Bronstein, Michael M.}, booktitle = {Proceedings of the First Learning on Graphs Conference}, pages = {11:1--11:16}, year = {2022}, editor = {Rieck, Bastian and Pascanu, Razvan}, volume = {198}, series = {Proceedings of Machine Learning Research}, month = {09--12 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v198/rossi22a/rossi22a.pdf}, url = {https://proceedings.mlr.press/v198/rossi22a.html}, abstract = {While Graph Neural Networks (GNNs) have recently become the de facto standard for modeling relational data, they impose a strong assumption on the availability of the node or edge features of the graph. In many real-world applications, however, features are only partially available; for example, in social networks, age and gender are available only for a small subset of users. We present a general approach for handling missing features in graph machine learning applications that is based on minimization of the Dirichlet energy and leads to a diffusion-type differential equation on the graph. The discretization of this equation produces a simple, fast and scalable algorithm which we call Feature Propagation. We experimentally show that the proposed approach outperforms previous methods on seven common node-classification benchmarks and can withstand surprisingly high rates of missing features: on average we observe only around 4% relative accuracy drop when 99% of the features are missing. Moreover, it takes only 10 seconds to run on a graph with ~2.5M nodes and ~23M edges on a single GPU. The code is available at https://github.com/twitter-research/feature-propagation.} }
Endnote
%0 Conference Paper %T On the Unreasonable Effectiveness of Feature Propagation in Learning on Graphs With Missing Node Features %A Emanuele Rossi %A Henry Kenlay %A Maria I. Gorinova %A Benjamin Paul Chamberlain %A Xiaowen Dong %A Michael M. Bronstein %B Proceedings of the First Learning on Graphs Conference %C Proceedings of Machine Learning Research %D 2022 %E Bastian Rieck %E Razvan Pascanu %F pmlr-v198-rossi22a %I PMLR %P 11:1--11:16 %U https://proceedings.mlr.press/v198/rossi22a.html %V 198 %X While Graph Neural Networks (GNNs) have recently become the de facto standard for modeling relational data, they impose a strong assumption on the availability of the node or edge features of the graph. In many real-world applications, however, features are only partially available; for example, in social networks, age and gender are available only for a small subset of users. We present a general approach for handling missing features in graph machine learning applications that is based on minimization of the Dirichlet energy and leads to a diffusion-type differential equation on the graph. The discretization of this equation produces a simple, fast and scalable algorithm which we call Feature Propagation. We experimentally show that the proposed approach outperforms previous methods on seven common node-classification benchmarks and can withstand surprisingly high rates of missing features: on average we observe only around 4% relative accuracy drop when 99% of the features are missing. Moreover, it takes only 10 seconds to run on a graph with ~2.5M nodes and ~23M edges on a single GPU. The code is available at https://github.com/twitter-research/feature-propagation.
APA
Rossi, E., Kenlay, H., Gorinova, M.I., Chamberlain, B.P., Dong, X. & Bronstein, M.M.. (2022). On the Unreasonable Effectiveness of Feature Propagation in Learning on Graphs With Missing Node Features. Proceedings of the First Learning on Graphs Conference, in Proceedings of Machine Learning Research 198:11:1-11:16 Available from https://proceedings.mlr.press/v198/rossi22a.html.

Related Material