{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,10,14]],"date-time":"2025-10-14T07:17:35Z","timestamp":1760426255658,"version":"3.41.2"},"reference-count":61,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2024,7,1]],"date-time":"2024-07-01T00:00:00Z","timestamp":1719792000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Big Data"],"abstract":"<jats:p>With the increasing popularity of Graph Neural Networks (GNNs) for predictive tasks on graph structured data, research on their explainability is becoming more critical and achieving significant progress. Although many methods are proposed to explain the predictions of GNNs, their focus is mainly on \u201chow to generate explanations.\u201d However, other important research questions like \u201cwhether the GNN explanations are inaccurate,\u201d \u201cwhat if the explanations are inaccurate,\u201d and \u201chow to adjust the model to generate more accurate explanations\u201d have gained little attention. Our previous GNN Explanation Supervision (GNES) framework demonstrated effectiveness on improving the reasonability of the local explanation while still keep or even improve the backbone GNNs model performance. In many applications instead of per sample explanations, we need to find global explanations which are reasonable and faithful to the domain data. Simply learning to explain GNNs locally is not an optimal solution to a global understanding of the model. To improve the explainability power of the GNES framework, we propose the Global GNN Explanation Supervision (GGNES) technique which uses a basic trained GNN and a global extension of the loss function used in the GNES framework. This GNN creates local explanations which are fed to a Global Logic-based GNN Explainer, an existing technique that can learn the global Explanation in terms of a logic formula. These two frameworks are then trained iteratively to generate reasonable global explanations. Extensive experiments demonstrate the effectiveness of the proposed model on improving the global explanations while keeping the performance similar or even increase the model prediction power.<\/jats:p>","DOI":"10.3389\/fdata.2024.1410424","type":"journal-article","created":{"date-parts":[[2024,7,1]],"date-time":"2024-07-01T05:02:14Z","timestamp":1719810134000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Global explanation supervision for Graph Neural Networks"],"prefix":"10.3389","volume":"7","author":[{"given":"Negar","family":"Etemadyrad","sequence":"first","affiliation":[]},{"given":"Yuyang","family":"Gao","sequence":"additional","affiliation":[]},{"given":"Sai","family":"Manoj Pudukotai Dinakarrao","sequence":"additional","affiliation":[]},{"given":"Liang","family":"Zhao","sequence":"additional","affiliation":[]}],"member":"1965","published-online":{"date-parts":[[2024,7,1]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"52138","DOI":"10.1109\/ACCESS.2018.2870052","article-title":"Peeking inside the black-box: a survey on explainable artificial intelligence (XAI)","volume":"6","author":"Adadi","year":"2018","journal-title":"IEEE Access"},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/N18-1029","article-title":"Learning beyond datasets: knowledge graph augmented neural networks for natural language processing","author":"Annervaz","year":"2018","journal-title":"arXiv preprint arXiv:1802.05930"},{"key":"B3","doi-asserted-by":"publisher","first-page":"82","DOI":"10.1016\/j.inffus.2019.12.012","article-title":"Explainable artificial intelligence (XAI): concepts, taxonomies, opportunities and challenges toward responsible AI","volume":"58","author":"Arrieta","year":"2020","journal-title":"Informa. Fus"},{"key":"B4","unstructured":"AtlasT. S. B.\n          Brain Atlas2023"},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2210.07147","article-title":"Global explainability of GNNs via logic combination of learned concepts","author":"Azzolin","year":"2022","journal-title":"arXiv preprint arXiv:2210.07147"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1905.13686","article-title":"Explainability techniques for graph convolutional networks","author":"Baldassarre","year":"2019","journal-title":"arXiv preprint arXiv:1905.13686"},{"key":"B7","doi-asserted-by":"publisher","first-page":"938","DOI":"10.1016\/j.neuroimage.2012.01.090","article-title":"Resting state fMRI: a personal history","volume":"62","author":"Biswal","year":"2012","journal-title":"Neuroimage"},{"key":"B8","doi-asserted-by":"publisher","first-page":"832","DOI":"10.1038\/nn.3423","article-title":"Opportunities and limitations of intrinsic functional connectivity MRI","volume":"16","author":"Buckner","year":"2013","journal-title":"Nat. Neurosci"},{"key":"B9","first-page":"91","article-title":"\u201cAir: attention with reasoning capability,\u201d","volume-title":"European Conference on Computer Vision","author":"Chen","year":"2020"},{"key":"B10","doi-asserted-by":"publisher","first-page":"42","DOI":"10.3389\/conf.fninf.2013.09.00042","article-title":"Towards automated analysis of connectomes: the configurable pipeline for the analysis of connectomes (C-PAC)","volume":"9","author":"Craddock","year":"2013","journal-title":"Front. Neuroinform"},{"key":"B11","doi-asserted-by":"publisher","first-page":"90","DOI":"10.1016\/j.cviu.2017.10.001","article-title":"Human attention in visual question answering: do humans and deep networks look at the same regions?","volume":"163","author":"Das","year":"2017","journal-title":"Comput. Vis. Image Underst"},{"key":"B12","doi-asserted-by":"publisher","first-page":"011027","DOI":"10.1103\/PhysRevX.5.011027","article-title":"Identifying modular flows on multilayer networks reveals highly overlapping organization in interconnected systems","volume":"5","author":"De Domenico","year":"2015","journal-title":"Phys. Rev. X"},{"key":"B13","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/1471-2202-10-101","article-title":"Functional neural network analysis in frontotemporal dementia and Alzheimer's disease using EEG and graph theory","volume":"10","author":"De Haan","year":"2009","journal-title":"BMC Neurosci"},{"key":"B14","doi-asserted-by":"publisher","first-page":"968","DOI":"10.1016\/j.neuroimage.2006.01.021","article-title":"An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest","volume":"31","author":"Desikan","year":"2006","journal-title":"Neuroimage"},{"key":"B15","doi-asserted-by":"publisher","first-page":"659","DOI":"10.1038\/mp.2013.78","article-title":"The autism brain imaging data exchange: towards a large-scale evaluation of the intrinsic brain architecture in autism","volume":"19","author":"Di Martino","year":"2014","journal-title":"Mol. Psychiat"},{"key":"B16","doi-asserted-by":"publisher","first-page":"620","DOI":"10.1038\/s42256-021-00343-w","article-title":"Improving performance of deep learning models with axiomatic attribution priors and expected gradients","volume":"3","author":"Erion","year":"2021","journal-title":"Nat. Machine Intell"},{"key":"B17","doi-asserted-by":"publisher","first-page":"4862","DOI":"10.1109\/TNNLS.2022.3197337","article-title":"Functional connectivity prediction with deep learning for graph transformation","volume":"35","author":"Etemadyrad","year":"2022","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"B18","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1145\/3308558.3313488","article-title":"\u201cGraph neural networks for social recommendation,\u201d","volume-title":"The World Wide Web Conference","author":"Fan","year":"2019"},{"key":"B19","first-page":"10705","article-title":"\u201cAttention branch network: learning of attention mechanism for visual explanation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Fukui","year":"2019"},{"key":"B20","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3644073","article-title":"Going beyond XAI: a systematic survey for explanation-guided learning","volume":"56","author":"Gao","year":"2024","journal-title":"ACM Comput. Surv"},{"key":"B21","doi-asserted-by":"crossref","first-page":"131","DOI":"10.1109\/ICDM51629.2021.00023","article-title":"\u201cGNES: learning to explain graph neural networks,\u201d","volume-title":"2021 IEEE International Conference on Data Mining (ICDM)","author":"Gao","year":"2021"},{"key":"B22","first-page":"432","article-title":"\u201cRES: a robust framework for guiding visual explanation,\u201d","volume-title":"Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining","author":"Gao","year":""},{"key":"B23","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3555590","article-title":"Aligning eyes between humans and deep neural network through interactive attention alignment","volume":"6","author":"Gao","year":"","journal-title":"Proc. ACM Hum. Comput. Interact"},{"key":"B24","doi-asserted-by":"publisher","first-page":"15684","DOI":"10.1523\/JNEUROSCI.2308-09.2009","article-title":"Age-and gender-related differences in the cortical anatomical network","volume":"29","author":"Gong","year":"2009","journal-title":"J. Neurosci"},{"key":"B25","doi-asserted-by":"publisher","first-page":"15303","DOI":"10.48550\/arXiv.2311.15303","article-title":"Concept distillation: leveraging human-centered explanations for model improvement","volume":"36","author":"Gupta","year":"2024","journal-title":"Adv. Neural Inform. Process. Syst"},{"key":"B26","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3392878","article-title":"Human factors in model interpretability: industry practices, challenges, and needs","volume":"4","author":"Hong","year":"2020","journal-title":"Proc. ACM Hum. Comput. Interact"},{"key":"B27","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2001.06216","article-title":"Graphlime: Local interpretable model explanations for graph neural networks","author":"Huang","year":"2020","journal-title":"arXiv preprint arXiv:2001.06216"},{"key":"B28","unstructured":"in St.LouisW. U.\n          2020"},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2006.01067","article-title":"Aligning faithful interpretations with their social attribution","author":"Jacovi","year":"2020","journal-title":"arXiv preprint arXiv:2006.01067"},{"key":"B30","doi-asserted-by":"publisher","first-page":"782","DOI":"10.1016\/j.neuroimage.2011.09.015","article-title":"Fsl","volume":"62","author":"Jenkinson","year":"","journal-title":"Neuroimage"},{"key":"B31","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.6980","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014","journal-title":"arXiv preprint arXiv:1412.6980"},{"key":"B32","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1609.02907","article-title":"Semi-supervised classification with graph convolutional networks","author":"Kipf","year":"2016","journal-title":"arXiv preprint arXiv:1609.02907"},{"key":"B33","doi-asserted-by":"publisher","first-page":"55","DOI":"10.1109\/TMI.2022.3203899","article-title":"Improving multi-site autism classification based on site-dependence minimisation and second-order functional connectivity","volume":"42","author":"Kunda","year":"2020","journal-title":"IEEE Trans. Med. Imag"},{"key":"B34","doi-asserted-by":"publisher","first-page":"3203","DOI":"10.48550\/arXiv.2210.07024","article-title":"Self-explaining deep models with logic rule reasoning","volume":"35","author":"Lee","year":"2022","journal-title":"Adv. Neural Inform. Process. Syst"},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1805.08819","article-title":"Learning what and where to attend","author":"Linsley","year":"2018","journal-title":"arXiv preprint arXiv:1805.08819"},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2011.04573","article-title":"Parameterized explainer for graph neural network","author":"Luo","year":"2020","journal-title":"arXiv preprint arXiv:2011.04573"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1909.10660","article-title":"Exploring graph neural networks for stock market predictions with rolling window analysis","author":"Matsunaga","year":"2019","journal-title":"arXiv preprint arXiv:1909.10660"},{"key":"B38","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1905.03540","article-title":"Embedding human knowledge into deep neural network via attention map","author":"Mitsuhara","year":"2019","journal-title":"arXiv preprint arXiv:1905.03540"},{"key":"B39","unstructured":"Fsl-Scripts2014"},{"key":"B40","first-page":"11848","article-title":"\u201cExplanation vs. attention: a two-player game to obtain attention for VQA,\u201d","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 34","author":"Patro","year":"2020"},{"key":"B41","first-page":"10772","article-title":"\u201cExplainability methods for graph convolutional neural networks,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Pope","year":"2019"},{"key":"B42","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v32i1.12272","article-title":"\u201cExploring human-like attention supervision in visual question answering,\u201d","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 32","author":"Qiao","year":"2018"},{"key":"B43","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1703.03717","article-title":"Right for the right reasons: training differentiable models by constraining their explanations","author":"Ross","year":"2017","journal-title":"arXiv preprint arXiv:1703.03717"},{"key":"B44","doi-asserted-by":"publisher","first-page":"61","DOI":"10.1109\/TNN.2008.2005605","article-title":"The graph neural network model","volume":"20","author":"Scarselli","year":"2008","journal-title":"IEEE Trans. Neural Netw"},{"key":"B45","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2010.00577","article-title":"Interpreting graph neural networks for NLP with differentiable edge masking","author":"Schlichtkrull","year":"2020","journal-title":"arXiv preprint arXiv:2010.00577"},{"volume-title":"Higher-Order Explanations of Graph Neural Networks via Relevant Walks","year":"2020","author":"Schnake","key":"B46"},{"key":"B47","doi-asserted-by":"publisher","first-page":"103828","DOI":"10.1016\/j.artint.2022.103828","article-title":"Rationalizing predictions by adversarial information calibration","volume":"315","author":"Sha","year":"2023","journal-title":"Artif. Intell"},{"key":"B48","first-page":"104","article-title":"\u201cEngage: explanation guided data augmentation for graph representation learning,\u201d","volume-title":"Joint European Conference on Machine Learning and Knowledge Discovery in Databases","author":"Shi","year":"2023"},{"key":"B49","doi-asserted-by":"publisher","first-page":"273","DOI":"10.1006\/nimg.2001.0978","article-title":"Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain","volume":"15","author":"Tzourio-Mazoyer","year":"2002","journal-title":"Neuroimage"},{"key":"B50","doi-asserted-by":"publisher","first-page":"62","DOI":"10.1016\/j.neuroimage.2013.05.041","article-title":"The wu-minn human connectome project: an overview","volume":"80","author":"Van Essen","year":"2013","journal-title":"Neuroimage"},{"key":"B51","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1906.03859","article-title":"Few-shot learning with per-sample rich supervision","author":"Visotsky","year":"2019","journal-title":"arXiv preprint arXiv:1906.03859"},{"key":"B52","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2010.05788","article-title":"PGM-explainer: probabilistic graphical model explanations for graph neural networks","author":"Vu","year":"2020","journal-title":"arXiv preprint arXiv:2010.05788"},{"key":"B53","doi-asserted-by":"publisher","first-page":"14034","DOI":"10.48550\/arXiv.1912.10065","article-title":"Learning deep attribution priors based on prior knowledge","volume":"33","author":"Weinberger","year":"2020","journal-title":"Adv. Neural Inform. Process. Syst"},{"volume-title":"Graph Neural Networks: Foundations, Frontiers, and Applications","year":"2021","author":"Wu","key":"B54"},{"key":"B55","unstructured":"1200 Subjects Data Release Reference Manual2017"},{"key":"B56","doi-asserted-by":"publisher","first-page":"9240","DOI":"10.48550\/arXiv.1903.03894","article-title":"GNNExplainer: generating explanations for graph neural networks","volume":"32","author":"Ying","year":"2019","journal-title":"Adv. Neural Inform. Process. Syst"},{"key":"B57","first-page":"430","article-title":"\u201cXGNN: towards model-level explanations of graph neural networks,\u201d","volume-title":"Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","author":"Yuan","year":""},{"key":"B58","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2012.15445","article-title":"Explainability in graph neural networks: a taxonomic survey","author":"Yuan","year":"","journal-title":"arXiv preprint arXiv:2012.15445"},{"key":"B59","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v32i1.11782","article-title":"\u201cAn end-to-end deep learning architecture for graph classification,\u201d","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 32","author":"Zhang","year":"2018"},{"key":"B60","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2006.00305","article-title":"RelEx: a model-agnostic relational model explainer","author":"Zhang","year":"2020","journal-title":"arXiv preprint arXiv:2006.00305"},{"key":"B61","doi-asserted-by":"crossref","first-page":"349","DOI":"10.1109\/WACV.2019.00043","article-title":"\u201cInterpretable visual question answering by visual grounding from attention supervision mining,\u201d","volume-title":"2019 IEEE Winter Conference on Applications of Computer Vision (WACV)","author":"Zhang","year":"2019"}],"container-title":["Frontiers in Big Data"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2024.1410424\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,7,1]],"date-time":"2024-07-01T05:02:30Z","timestamp":1719810150000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2024.1410424\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,7,1]]},"references-count":61,"alternative-id":["10.3389\/fdata.2024.1410424"],"URL":"https:\/\/doi.org\/10.3389\/fdata.2024.1410424","relation":{},"ISSN":["2624-909X"],"issn-type":[{"type":"electronic","value":"2624-909X"}],"subject":[],"published":{"date-parts":[[2024,7,1]]},"article-number":"1410424"}}