{"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/typed-graph-networks","title":"Typed Graph Networks","arxiv_id":"1901.07984","date":"2019-01-23","proceeding":null,"authors":["Marcelo O. R. Prates","Pedro H. C. Avelar","Henrique Lemos","Marco Gori","Luis Lamb"],"abstract":"Recently, the deep learning community has given growing attention to neural\narchitectures engineered to learn problems in relational domains. Convolutional\nNeural Networks employ parameter sharing over the image domain, tying the\nweights of neural connections on a grid topology and thus enforcing the\nlearning of a number of convolutional kernels. By instantiating trainable\nneural modules and assembling them in varied configurations (apart from grids),\none can enforce parameter sharing over graphs, yielding models which can\neffectively be fed with relational data. In this context, vertices in a graph\ncan be projected into a hyperdimensional real space and iteratively refined\nover many message-passing iterations in an end-to-end differentiable\narchitecture. Architectures of this family have been referred to with several\ndefinitions in the literature, such as Graph Neural Networks, Message-passing\nNeural Networks, Relational Networks and Graph Networks. In this paper, we\nrevisit the original Graph Neural Network model and show that it generalises\nmany of the recent models, which in turn benefit from the insight of thinking\nabout vertex \\textbf{types}. To illustrate the generality of the original\nmodel, we present a Graph Neural Network formalisation, which partitions the\nvertices of a graph into a number of types. Each type represents an entity in\nthe ontology of the problem one wants to learn. This allows - for instance -\none to assign embeddings to edges, hyperedges, and any number of global\nattributes of the graph. As a companion to this paper we provide a\nPython/Tensorflow library to facilitate the development of such architectures,\nwith which we instantiate the formalisation to reproduce a number of models\nproposed in the current literature.","url_abs":"http://arxiv.org/abs/1901.07984v3","url_pdf":"http://arxiv.org/pdf/1901.07984v3.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":"typed-graph-networks","repo_url":"https://github.com/machine-reasoning-ufrgs/typed-graph-network","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"typed-graph-networks","repo_url":"https://github.com/machine-reasoning-ufrgs/graph-neural-networks","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"graph-neural-network","task_name":"Graph Neural Network"}],"methods":[{"method_slug":"graph-neural-network","method_name":"Graph Neural Network"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}