A Gradient-Based Explanation Method for Node Classification Using Graph Convolutional Networks
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Explainable artificial intelligence is a method that explains how a complex model (e.g., a deep neural network)yields its output from a given input. Recently, graph-type data have been widely used in various fields, anddiverse graph neural networks (GNNs) have been developed for graph-type data. However, methods to explainthe behavior of GNNs have not been studied much, and only a limited understanding of GNNs is currentlyavailable. Therefore, in this paper, we propose an explanation method for node classification using graphconvolutional networks (GCNs), which is a representative type of GNN. The proposed method finds out whichfeatures of each node have the greatest influence on the classification of that node using GCN. The proposedmethod identifies influential features by backtracking the layers of the GCN from the output layer to the inputlayer using the gradients. The experimental results on both synthetic and real datasets demonstrate that theproposed explanation method accurately identifies the features of each node that have the greatest influence onits classification.
Abstract
Explainable artificial intelligence is a method that explains how a complex model (e.g., a deep neural network)yields its output from a given input. Recently, graph-type data have been widely used in various fields, anddiverse graph neural networks (GNNs) have been developed for graph-type data. However, methods to explainthe behavior of GNNs have not been studied much, and only a limited understanding of GNNs is currentlyavailable. Therefore, in this paper, we propose an explanation method for node classification using graphconvolutional networks (GCNs), which is a representative type of GNN. The proposed method finds out whichfeatures of each node have the greatest influence on the classification of that node using GCN. The proposedmethod identifies influential features by backtracking the layers of the GCN from the output layer to the inputlayer using the gradients. The experimental results on both synthetic and real datasets demonstrate that theproposed explanation method accurately identifies the features of each node that have the greatest influence onits classification.
참고문헌 (13)
- Alejandro Barredo Arrieta, "Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI", Information Fusion, 58, pp.82~115 (2020), Elsevier BV[DOI]
- 이성빈, "Explainable Machine Learning Based a Packed Red Blood Cell Transfusion Prediction and Evaluation for Major Internal Medical Condition", JIPS(Journal of Information Processing Systems), 18(3), pp.302~310 (2022), 한국정보처리학회[DOI]
- "Semi-supervised classification with graph convolutional networks" (2017)[URL]
- Zonghan Wu, "A Comprehensive Survey on Graph Neural Networks", IEEE Transactions on Neural Networks and Learning Systems, 32(1), pp.4~24 (2021), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- B. Zhou, "Learning deep features for discriminative localization", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.2921~2929 (2016)
- R. R. Selvaraju, "Grad-CAM: visual explanations from deep networks via gradient-based localization", Proceedings of the IEEE International Conference on Computer Vision, 2017, pp.618~626 (2017)
- Sebastian Bach, "On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation", PLOS ONE, 10(7), pp.e0130140~ (2015), Public Library of Science (PLoS)[DOI]
- M. T. Ribeiro, "Why should I trust you? Explaining the predictions of any classifier", Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp.1135~1144 (2016)
- Erico Tjoa, "A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI", IEEE Transactions on Neural Networks and Learning Systems, 32(11), pp.4793~4813 (2021), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- J. Hu, "GCN-LRP explanation: exploring latent attention of graph convolutional networks", Proceedings of 2020 International Joint Conference on Neural Networks (IJCNN), pp.1~8 (2020)
- C. Kim, "A gradient-based explanation method for graph convolutional neural networks", Proceedings of Annual Spring Conference of KIPS (ASK 2022), pp.670~673 (2022)
- Prithviraj Sen, "Collective Classification in Network Data", AI Magazine, 29(3), pp.93~106 (2008), Wiley[DOI]
- Giulia Vilone, "Notions of explainability and evaluation approaches for explainable artificial intelligence", Information Fusion, 76, pp.89~106 (2021), Elsevier BV[DOI]