{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/splinecnn-fast-geometric-deep-learning-with","title":"SplineCNN: Fast Geometric Deep Learning with Continuous B-Spline Kernels","arxiv_id":"1711.08920","date":"2017-11-24","proceeding":"CVPR 2018 6","authors":["Matthias Fey","Jan Eric Lenssen","Frank Weichert","Heinrich Müller"],"abstract":"We present Spline-based Convolutional Neural Networks (SplineCNNs), a variant\nof deep neural networks for irregular structured and geometric input, e.g.,\ngraphs or meshes. Our main contribution is a novel convolution operator based\non B-splines, that makes the computation time independent from the kernel size\ndue to the local support property of the B-spline basis functions. As a result,\nwe obtain a generalization of the traditional CNN convolution operator by using\ncontinuous kernel functions parametrized by a fixed number of trainable\nweights. In contrast to related approaches that filter in the spectral domain,\nthe proposed method aggregates features purely in the spatial domain. In\naddition, SplineCNN allows entire end-to-end training of deep architectures,\nusing only the geometric structure as input, instead of handcrafted feature\ndescriptors. For validation, we apply our method on tasks from the fields of\nimage graph classification, shape correspondence and graph node classification,\nand show that it outperforms or pars state-of-the-art approaches while being\nsignificantly faster and having favorable properties like domain-independence.","url_abs":"http://arxiv.org/abs/1711.08920v2","url_pdf":"http://arxiv.org/pdf/1711.08920v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"splinecnn-fast-geometric-deep-learning-with","repo_url":"https://github.com/rusty1s/pytorch_geometric","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"splinecnn-fast-geometric-deep-learning-with","repo_url":"https://github.com/MikelBros/Deep_Spammer_Detection_GCN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"splinecnn-fast-geometric-deep-learning-with","repo_url":"https://github.com/mikel-brostrom/Deep_Spammer_Detection_GCN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"splinecnn-fast-geometric-deep-learning-with","repo_url":"https://github.com/rusty1s/pytorch_spline_conv","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"splinecnn-fast-geometric-deep-learning-with","repo_url":"https://github.com/abhilash1910/SpectralEmbeddings","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"graph-classification","task_name":"Graph Classification"},{"task_slug":"node-classification","task_name":"Node Classification"},{"task_slug":"superpixel-image-classification","task_name":"Superpixel Image Classification"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/node-classification-on-citeseer","task":"Node Classification","dataset":"Citeseer","model":"SplineCNN","rank_in_archive_order":7,"of":71,"metrics":{"Accuracy":"79.20%"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora","task":"Node Classification","dataset":"Cora","model":"SplineCNN","rank_in_archive_order":2,"of":73,"metrics":{"Accuracy":"89.48% ± 0.31%"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed","task":"Node Classification","dataset":"Pubmed","model":"SplineCNN","rank_in_archive_order":13,"of":70,"metrics":{"Accuracy":"88.88%"},"uses_additional_data":false},{"leaderboard":"/sota/superpixel-image-classification-on-75","task":"Superpixel Image Classification","dataset":"75 Superpixel MNIST","model":"SplineCNN","rank_in_archive_order":5,"of":6,"metrics":{"Classification Error":"4.78"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.08920","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}