{"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/deep-gaussian-embedding-of-graphs","title":"Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via Ranking","arxiv_id":"1707.03815","date":"2017-07-12","proceeding":"ICLR 2018 1","authors":["Aleksandar Bojchevski","Stephan Günnemann"],"abstract":"Methods that learn representations of nodes in a graph play a critical role\nin network analysis since they enable many downstream learning tasks. We\npropose Graph2Gauss - an approach that can efficiently learn versatile node\nembeddings on large scale (attributed) graphs that show strong performance on\ntasks such as link prediction and node classification. Unlike most approaches\nthat represent nodes as point vectors in a low-dimensional continuous space, we\nembed each node as a Gaussian distribution, allowing us to capture uncertainty\nabout the representation. Furthermore, we propose an unsupervised method that\nhandles inductive learning scenarios and is applicable to different types of\ngraphs: plain/attributed, directed/undirected. By leveraging both the network\nstructure and the associated node attributes, we are able to generalize to\nunseen nodes without additional training. To learn the embeddings we adopt a\npersonalized ranking formulation w.r.t. the node distances that exploits the\nnatural ordering of the nodes imposed by the network structure. Experiments on\nreal world networks demonstrate the high performance of our approach,\noutperforming state-of-the-art network embedding methods on several different\ntasks. Additionally, we demonstrate the benefits of modeling uncertainty - by\nanalyzing it we can estimate neighborhood diversity and detect the intrinsic\nlatent dimensionality of a graph.","url_abs":"http://arxiv.org/abs/1707.03815v4","url_pdf":"http://arxiv.org/pdf/1707.03815v4.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":"deep-gaussian-embedding-of-graphs","repo_url":"https://github.com/abojchevski/graph2gauss","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"inductive-learning","task_name":"Inductive Learning"},{"task_slug":"link-prediction","task_name":"Link Prediction"},{"task_slug":"network-embedding","task_name":"Network Embedding"},{"task_slug":"node-classification","task_name":"Node Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1707.03815","atlas_url":"https://app.syntology.ai/?focus=1707.03815","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}