{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T22:59:57Z","timestamp":1783983597040,"version":"3.55.0"},"reference-count":40,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2022,11,17]],"date-time":"2022-11-17T00:00:00Z","timestamp":1668643200000},"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>Graph neural networks (GNNs) have been widely used in various graph analysis tasks. As the graph characteristics vary significantly in real-world systems, given a specific scenario, the architecture parameters need to be tuned carefully to identify a suitable GNN. Neural architecture search (NAS) has shown its potential in discovering the effective architectures for the learning tasks in image and language modeling. However, the existing NAS algorithms cannot be applied efficiently to GNN search problem because of two facts. First, the large-step exploration in the traditional controller fails to learn the sensitive performance variations with slight architecture modifications in GNNs. Second, the search space is composed of heterogeneous GNNs, which prevents the direct adoption of parameter sharing among them to accelerate the search progress. To tackle the challenges, we propose an automated graph neural networks (AGNN) framework, which aims to find the optimal GNN architecture efficiently. Specifically, a reinforced conservative controller is designed to explore the architecture space with small steps. To accelerate the validation, a novel constrained parameter sharing strategy is presented to regularize the weight transferring among GNNs. It avoids training from scratch and saves the computation time. Experimental results on the benchmark datasets demonstrate that the architecture identified by AGNN achieves the best performance and search efficiency, comparing with existing human-invented models and the traditional search methods.<\/jats:p>","DOI":"10.3389\/fdata.2022.1029307","type":"journal-article","created":{"date-parts":[[2022,11,17]],"date-time":"2022-11-17T12:37:35Z","timestamp":1668688655000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":82,"title":["Auto-GNN: Neural architecture search of graph neural networks"],"prefix":"10.3389","volume":"5","author":[{"given":"Kaixiong","family":"Zhou","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiao","family":"Huang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Qingquan","family":"Song","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rui","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xia","family":"Hu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2022,11,17]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1611.02167","article-title":"Designing neural network architectures using reinforcement learning","author":"Baker","year":"2016","journal-title":"arXiv"},{"key":"B2","first-page":"6657","article-title":"Rethinking graph neural architecture search from message-passing,","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Cai","year":"2021"},{"key":"B3","doi-asserted-by":"crossref","first-page":"601","DOI":"10.1109\/SocialCom.2010.94","article-title":"A cluster-based approach for automatic social network construction,","volume-title":"2010 IEEE Second International Conference on Social Computing","author":"Chang","year":"2010"},{"key":"B4","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2007.02133","article-title":"Simple and deep graph convolutional networks","author":"Chen","year":"2020","journal-title":"arXiv preprint arXiv:2007.02133"},{"key":"B5","doi-asserted-by":"publisher","first-page":"35","DOI":"10.1145\/3447556.3447567","article-title":"Techniques for automated machine learning","volume":"22","author":"Chen","year":"2021","journal-title":"ACM SIGKDD Explorat. Newslett"},{"key":"B6","doi-asserted-by":"crossref","first-page":"257","DOI":"10.1145\/3292500.3330925","article-title":"Cluster-gcn: an efficient algorithm for training deep and large graph convolutional networks,","volume-title":"Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","author":"Chiang","year":"2019"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2004.05718","article-title":"Principal neighbourhood aggregation for graph nets","author":"Corso","year":"2020","journal-title":"arXiv preprint arXiv:2004.05718"},{"key":"B8","article-title":"Convolutional neural networks on graphs with fast localized spectral filtering,","volume-title":"NeuIPS","author":"Defferrard","year":"2016"},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2010.03250","article-title":"Propagation model search for graph neural networks","author":"Ding","year":"2020","journal-title":"arXiv preprint arXiv:2010.03250"},{"key":"B10","doi-asserted-by":"crossref","first-page":"279","DOI":"10.1145\/3447548.3467447","article-title":"Diffmg: differentiable meta graph search for heterogeneous graph neural networks,","volume-title":"Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining","author":"Ding","year":"2021"},{"key":"B11","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-05318-5_3","article-title":"Neural architecture search: a survey","author":"Elsken","year":"2018","journal-title":"arXiv"},{"key":"B12","doi-asserted-by":"crossref","first-page":"1416","DOI":"10.1145\/3219819.3219947","article-title":"Large-scale learnable graph convolutional networks,","volume-title":"Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining","author":"Gao","year":"2018"},{"key":"B13","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1904.09981","article-title":"Graphnas: graph neural architecture search with reinforcement learning","author":"Gao","year":"2019","journal-title":"arXiv"},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1704.01212","article-title":"Neural message passing for quantum chemistry","author":"Gilmer","year":"2017","journal-title":"arXiv"},{"key":"B15","doi-asserted-by":"crossref","first-page":"855","DOI":"10.1145\/2939672.2939754","article-title":"node2vec: scalable feature learning for networks,","volume-title":"Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","author":"Grover","year":"2016"},{"key":"B16","first-page":"1024","article-title":"Inductive representation learning on large graphs,","volume-title":"Advances in Neural Information Processing Systems","author":"Hamilton","year":"2017"},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2005.00687","article-title":"Open graph benchmark: datasets for machine learning on graphs","author":"Hu","year":"2020","journal-title":"arXiv preprint arXiv:2005.00687"},{"key":"B18","article-title":"Semi-supervised classification with graph convolutional networks,","volume-title":"ICLR","author":"Kipf","year":"2017"},{"key":"B19","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1810.05997","article-title":"Predict then propagate: graph neural networks meet personalized pagerank","author":"Klicpera","year":"2018","journal-title":"arXiv preprint arXiv:1810.05997"},{"key":"B20","doi-asserted-by":"crossref","first-page":"189","DOI":"10.1007\/978-3-030-63833-7_16","article-title":"Autograph: automated graph neural network,","volume-title":"International Conference on Neural Information Processing","author":"Li","year":"2020"},{"key":"B21","first-page":"82","article-title":"Auto-deeplab: hierarchical neural architecture search for semantic image segmentation,","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Liu","year":"2019"},{"key":"B22","doi-asserted-by":"publisher","first-page":"498","DOI":"10.1109\/TNN.2008.2010350","article-title":"Neural network for graphs: a contextual constructive approach","volume":"20","author":"Micheli","year":"2009","journal-title":"IEEE Trans. Neural Netw"},{"key":"B23","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1802.03268","article-title":"Efficient neural architecture search via parameter sharing","author":"Pham","year":"2018","journal-title":"arXiv"},{"key":"B24","doi-asserted-by":"publisher","first-page":"93","DOI":"10.1609\/aimag.v29i3.2157","article-title":"Collective classification in network data","volume":"29","author":"Sen","year":"2008","journal-title":"AI Mag"},{"key":"B25","doi-asserted-by":"publisher","first-page":"108752","DOI":"10.1016\/j.knosys.2022.108752","article-title":"Evolutionary architecture search for graph neural networks","volume":"247","author":"Shi","year":"2020","journal-title":"Knowl. Based Syst"},{"key":"B26","first-page":"1057","article-title":"Policy gradient methods for reinforcement learning with function approximation,","volume-title":"Advances in Neural Information Processing Systems","author":"Sutton","year":"2000"},{"key":"B27","first-page":"5998","article-title":"Attention is all you need,","volume-title":"Advances in Neural Information Processing Systems","author":"Vaswani","year":"2017"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1710.10903","article-title":"Graph attention networks","author":"Velickovic","year":"2017","journal-title":"arXiv"},{"key":"B29","first-page":"2091","article-title":"Pooling architecture search for graph classification,","volume-title":"Proceedings of the 30th ACM International Conference on Information and Knowledge Management","author":"Wei","year":"2021"},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1902.07153","article-title":"Simplifying graph convolutional networks","author":"Wu","year":"2019","journal-title":"arXiv preprint arXiv:1902.07153"},{"key":"B31","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1806.03536","article-title":"Representation learning on graphs with jumping knowledge networks","author":"Xu","year":"2018","journal-title":"arXiv preprint arXiv:1806.03536"},{"key":"B32","article-title":"Design space for graph neural networks","volume-title":"Advances in Neural Information Processing Systems, Vol. 33","author":"You","year":"2020"},{"key":"B33","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1907.04931","article-title":"Graphsaint: graph sampling based inductive learning method","author":"Zeng","year":"2019","journal-title":"arXiv preprint arXiv:1907.04931"},{"key":"B34","article-title":"Simplifying architecture search for graph neural network,","volume-title":"International Conference on Information and Knowledge Management","author":"Zhao","year":""},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1909.12223","article-title":"Pairnorm: tackling oversmoothing in gnns","author":"Zhao","year":"2019","journal-title":"arXiv preprint arXiv:1909.12223"},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2003.09676","article-title":"Probabilistic dual network architecture search on graphs","author":"Zhao","year":"","journal-title":"arXiv preprint arXiv:2003.09676"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2006.06972","article-title":"Towards deeper graph neural networks with differentiable group normalization","author":"Zhou","year":"2020","journal-title":"arXiv preprint arXiv:2006.06972"},{"key":"B38","doi-asserted-by":"publisher","first-page":"i190","DOI":"10.1093\/bioinformatics\/btx252","article-title":"Predicting multicellular function through multi-layer tissue networks","volume":"33","author":"Zitnik","year":"2017","journal-title":"Bioinformatics"},{"key":"B39","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1611.01578","article-title":"Neural architecture search with reinforcement learning","author":"Zoph","year":"2016","journal-title":"arXiv"},{"key":"B40","first-page":"8697","article-title":"Learning transferable architectures for scalable image recognition,","author":"Zoph","year":"2018","journal-title":"CVPR"}],"container-title":["Frontiers in Big Data"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2022.1029307\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,11,17]],"date-time":"2022-11-17T12:38:03Z","timestamp":1668688683000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2022.1029307\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,11,17]]},"references-count":40,"alternative-id":["10.3389\/fdata.2022.1029307"],"URL":"https:\/\/doi.org\/10.3389\/fdata.2022.1029307","relation":{},"ISSN":["2624-909X"],"issn-type":[{"value":"2624-909X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,11,17]]},"article-number":"1029307"}}