Rethinking GNNs and Missing Features: Challenges, Evaluation and a Robust Solution

Francesco Ferrini, Veronica Lachi, Antonio Longa, Bruno Lepri, Akiyoshi Matono, Andrea Passerini, Xin Liu, Manfred Jaeger
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:30897-30935, 2026.

Abstract

Handling missing node features is a key challenge for deploying Graph Neural Networks (GNNs) in real-world domains such as healthcare and sensor networks. Existing studies mostly address relatively benign scenarios, namely benchmark datasets with (a) high-dimensional but sparse node features and (b) incomplete data generated under Missing Completely At Random (MCAR) mechanisms. For (a), we theoretically prove that high sparsity substantially limits the information loss caused by missingness, making all models appear robust and preventing a meaningful comparison of their performance. To overcome this limitation, we introduce one synthetic and three real-world datasets with dense, semantically meaningful features. For (b), we move beyond MCAR and design evaluation protocols with more realistic missingness mechanisms. Moreover, we provide a theoretical background to state explicit assumptions on the missingness process and analyze their implications for different methods. Building on this analysis, we show that a simple baseline adapted to the graph domain is competitive with respect to specialized architectures across diverse datasets and missingness regimes.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-ferrini26a, title = {Rethinking {GNN}s and Missing Features: Challenges, Evaluation and a Robust Solution}, author = {Ferrini, Francesco and Lachi, Veronica and Longa, Antonio and Lepri, Bruno and Matono, Akiyoshi and Passerini, Andrea and Liu, Xin and Jaeger, Manfred}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {30897--30935}, 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/ferrini26a/ferrini26a.pdf}, url = {https://proceedings.mlr.press/v306/ferrini26a.html}, abstract = {Handling missing node features is a key challenge for deploying Graph Neural Networks (GNNs) in real-world domains such as healthcare and sensor networks. Existing studies mostly address relatively benign scenarios, namely benchmark datasets with (a) high-dimensional but sparse node features and (b) incomplete data generated under Missing Completely At Random (MCAR) mechanisms. For (a), we theoretically prove that high sparsity substantially limits the information loss caused by missingness, making all models appear robust and preventing a meaningful comparison of their performance. To overcome this limitation, we introduce one synthetic and three real-world datasets with dense, semantically meaningful features. For (b), we move beyond MCAR and design evaluation protocols with more realistic missingness mechanisms. Moreover, we provide a theoretical background to state explicit assumptions on the missingness process and analyze their implications for different methods. Building on this analysis, we show that a simple baseline adapted to the graph domain is competitive with respect to specialized architectures across diverse datasets and missingness regimes.} }
Endnote
%0 Conference Paper %T Rethinking GNNs and Missing Features: Challenges, Evaluation and a Robust Solution %A Francesco Ferrini %A Veronica Lachi %A Antonio Longa %A Bruno Lepri %A Akiyoshi Matono %A Andrea Passerini %A Xin Liu %A Manfred Jaeger %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-ferrini26a %I PMLR %P 30897--30935 %U https://proceedings.mlr.press/v306/ferrini26a.html %V 306 %X Handling missing node features is a key challenge for deploying Graph Neural Networks (GNNs) in real-world domains such as healthcare and sensor networks. Existing studies mostly address relatively benign scenarios, namely benchmark datasets with (a) high-dimensional but sparse node features and (b) incomplete data generated under Missing Completely At Random (MCAR) mechanisms. For (a), we theoretically prove that high sparsity substantially limits the information loss caused by missingness, making all models appear robust and preventing a meaningful comparison of their performance. To overcome this limitation, we introduce one synthetic and three real-world datasets with dense, semantically meaningful features. For (b), we move beyond MCAR and design evaluation protocols with more realistic missingness mechanisms. Moreover, we provide a theoretical background to state explicit assumptions on the missingness process and analyze their implications for different methods. Building on this analysis, we show that a simple baseline adapted to the graph domain is competitive with respect to specialized architectures across diverse datasets and missingness regimes.
APA
Ferrini, F., Lachi, V., Longa, A., Lepri, B., Matono, A., Passerini, A., Liu, X. & Jaeger, M.. (2026). Rethinking GNNs and Missing Features: Challenges, Evaluation and a Robust Solution. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:30897-30935 Available from https://proceedings.mlr.press/v306/ferrini26a.html.

Related Material