Robust Graph Neural Networks via Ensemble Learning
Graph neural networks (GNNs) have demonstrated a remarkable ability in the task of semi-supervised node classification. However, most existing GNNs suffer from the nonrobustness issues, which poses a great challenge for applying GNNs into sensitive scenarios. Some researchers concentrate on construc...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-04-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/10/8/1300 |
| _version_ | 1827619659061395456 |
|---|---|
| author | Qi Lin Shuo Yu Ke Sun Wenhong Zhao Osama Alfarraj Amr Tolba Feng Xia |
| author_facet | Qi Lin Shuo Yu Ke Sun Wenhong Zhao Osama Alfarraj Amr Tolba Feng Xia |
| author_sort | Qi Lin |
| collection | DOAJ |
| description | Graph neural networks (GNNs) have demonstrated a remarkable ability in the task of semi-supervised node classification. However, most existing GNNs suffer from the nonrobustness issues, which poses a great challenge for applying GNNs into sensitive scenarios. Some researchers concentrate on constructing an ensemble model to mitigate the nonrobustness issues. Nevertheless, these methods ignore the interaction among base models, leading to similar graph representations. Moreover, due to the deterministic propagation applied in most existing GNNs, each node highly relies on its neighbors, leaving the nodes to be sensitive to perturbations. Therefore, in this paper, we propose a novel framework of graph ensemble learning based on knowledge passing (called GEL) to address the above issues. In order to achieve interaction, we consider the predictions of prior models as knowledge to obtain more reliable predictions. Moreover, we design a multilayer DropNode propagation strategy to reduce each node’s dependence on particular neighbors. This strategy also empowers each node to aggregate information from diverse neighbors, alleviating oversmoothing issues. We conduct experiments on three benchmark datasets, including Cora, Citeseer, and Pubmed. GEL outperforms GCN by more than 5% in terms of accuracy across all three datasets and also performs better than other state-of-the-art baselines. Extensive experimental results also show that the GEL alleviates the nonrobustness and oversmoothing issues. |
| first_indexed | 2024-03-09T10:32:11Z |
| format | Article |
| id | doaj.art-8cbad9e43b7c45529729a9b99be1d6e5 |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-09T10:32:11Z |
| publishDate | 2022-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-8cbad9e43b7c45529729a9b99be1d6e52023-12-01T21:12:12ZengMDPI AGMathematics2227-73902022-04-01108130010.3390/math10081300Robust Graph Neural Networks via Ensemble LearningQi Lin0Shuo Yu1Ke Sun2Wenhong Zhao3Osama Alfarraj4Amr Tolba5Feng Xia6School of Software, Dalian University of Technology, Dalian 116620, ChinaSchool of Software, Dalian University of Technology, Dalian 116620, ChinaSchool of Software, Dalian University of Technology, Dalian 116620, ChinaUltraprecision Machining Center, Zhejiang University of Technology, Hangzhou 310014, ChinaComputer Science Department, Community College, King Saud University, Riyadh 11437, Saudi ArabiaComputer Science Department, Community College, King Saud University, Riyadh 11437, Saudi ArabiaSchool of Engineering, IT and Physical Sciences, Federation University Australia, Ballarat, VIC 3353, AustraliaGraph neural networks (GNNs) have demonstrated a remarkable ability in the task of semi-supervised node classification. However, most existing GNNs suffer from the nonrobustness issues, which poses a great challenge for applying GNNs into sensitive scenarios. Some researchers concentrate on constructing an ensemble model to mitigate the nonrobustness issues. Nevertheless, these methods ignore the interaction among base models, leading to similar graph representations. Moreover, due to the deterministic propagation applied in most existing GNNs, each node highly relies on its neighbors, leaving the nodes to be sensitive to perturbations. Therefore, in this paper, we propose a novel framework of graph ensemble learning based on knowledge passing (called GEL) to address the above issues. In order to achieve interaction, we consider the predictions of prior models as knowledge to obtain more reliable predictions. Moreover, we design a multilayer DropNode propagation strategy to reduce each node’s dependence on particular neighbors. This strategy also empowers each node to aggregate information from diverse neighbors, alleviating oversmoothing issues. We conduct experiments on three benchmark datasets, including Cora, Citeseer, and Pubmed. GEL outperforms GCN by more than 5% in terms of accuracy across all three datasets and also performs better than other state-of-the-art baselines. Extensive experimental results also show that the GEL alleviates the nonrobustness and oversmoothing issues.https://www.mdpi.com/2227-7390/10/8/1300graph neural networksgraph learningensemble learningmultilayer DropNode propagationknowledge passing |
| spellingShingle | Qi Lin Shuo Yu Ke Sun Wenhong Zhao Osama Alfarraj Amr Tolba Feng Xia Robust Graph Neural Networks via Ensemble Learning Mathematics graph neural networks graph learning ensemble learning multilayer DropNode propagation knowledge passing |
| title | Robust Graph Neural Networks via Ensemble Learning |
| title_full | Robust Graph Neural Networks via Ensemble Learning |
| title_fullStr | Robust Graph Neural Networks via Ensemble Learning |
| title_full_unstemmed | Robust Graph Neural Networks via Ensemble Learning |
| title_short | Robust Graph Neural Networks via Ensemble Learning |
| title_sort | robust graph neural networks via ensemble learning |
| topic | graph neural networks graph learning ensemble learning multilayer DropNode propagation knowledge passing |
| url | https://www.mdpi.com/2227-7390/10/8/1300 |
| work_keys_str_mv | AT qilin robustgraphneuralnetworksviaensemblelearning AT shuoyu robustgraphneuralnetworksviaensemblelearning AT kesun robustgraphneuralnetworksviaensemblelearning AT wenhongzhao robustgraphneuralnetworksviaensemblelearning AT osamaalfarraj robustgraphneuralnetworksviaensemblelearning AT amrtolba robustgraphneuralnetworksviaensemblelearning AT fengxia robustgraphneuralnetworksviaensemblelearning |