{"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/robust-spatial-filtering-with-graph","title":"Robust Spatial Filtering with Graph Convolutional Neural Networks","arxiv_id":"1703.00792","date":"2017-03-02","proceeding":null,"authors":["Felipe Petroski Such","Shagan Sah","Miguel Dominguez","Suhas Pillai","Chao Zhang","Andrew Michael","Nathan Cahill","Raymond Ptucha"],"abstract":"Convolutional Neural Networks (CNNs) have recently led to incredible\nbreakthroughs on a variety of pattern recognition problems. Banks of finite\nimpulse response filters are learned on a hierarchy of layers, each\ncontributing more abstract information than the previous layer. The simplicity\nand elegance of the convolutional filtering process makes them perfect for\nstructured problems such as image, video, or voice, where vertices are\nhomogeneous in the sense of number, location, and strength of neighbors. The\nvast majority of classification problems, for example in the pharmaceutical,\nhomeland security, and financial domains are unstructured. As these problems\nare formulated into unstructured graphs, the heterogeneity of these problems,\nsuch as number of vertices, number of connections per vertex, and edge\nstrength, cannot be tackled with standard convolutional techniques. We propose\na novel neural learning framework that is capable of handling both homogeneous\nand heterogeneous data, while retaining the benefits of traditional CNN\nsuccesses.\n  Recently, researchers have proposed variations of CNNs that can handle graph\ndata. In an effort to create learnable filter banks of graphs, these methods\neither induce constraints on the data or require preprocessing. As opposed to\nspectral methods, our framework, which we term Graph-CNNs, defines filters as\npolynomials of functions of the graph adjacency matrix. Graph-CNNs can handle\nboth heterogeneous and homogeneous graph data, including graphs having entirely\ndifferent vertex or edge sets. We perform experiments to validate the\napplicability of Graph-CNNs to a variety of structured and unstructured\nclassification problems and demonstrate state-of-the-art results on document\nand molecule classification problems.","url_abs":"http://arxiv.org/abs/1703.00792v3","url_pdf":"http://arxiv.org/pdf/1703.00792v3.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":"robust-spatial-filtering-with-graph","repo_url":"https://github.com/fps7806/Graph-CNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.00792","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}