{"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/graphgan-graph-representation-learning-with","title":"GraphGAN: Graph Representation Learning with Generative Adversarial Nets","arxiv_id":"1711.08267","date":"2017-11-22","proceeding":null,"authors":["Hongwei Wang","Jia Wang","Jialin Wang","Miao Zhao","Wei-Nan Zhang","Fuzheng Zhang","Xing Xie","Minyi Guo"],"abstract":"The goal of graph representation learning is to embed each vertex in a graph\ninto a low-dimensional vector space. Existing graph representation learning\nmethods can be classified into two categories: generative models that learn the\nunderlying connectivity distribution in the graph, and discriminative models\nthat predict the probability of edge existence between a pair of vertices. In\nthis paper, we propose GraphGAN, an innovative graph representation learning\nframework unifying above two classes of methods, in which the generative model\nand discriminative model play a game-theoretical minimax game. Specifically,\nfor a given vertex, the generative model tries to fit its underlying true\nconnectivity distribution over all other vertices and produces \"fake\" samples\nto fool the discriminative model, while the discriminative model tries to\ndetect whether the sampled vertex is from ground truth or generated by the\ngenerative model. With the competition between these two models, both of them\ncan alternately and iteratively boost their performance. Moreover, when\nconsidering the implementation of generative model, we propose a novel graph\nsoftmax to overcome the limitations of traditional softmax function, which can\nbe proven satisfying desirable properties of normalization, graph structure\nawareness, and computational efficiency. Through extensive experiments on\nreal-world datasets, we demonstrate that GraphGAN achieves substantial gains in\na variety of applications, including link prediction, node classification, and\nrecommendation, over state-of-the-art baselines.","url_abs":"http://arxiv.org/abs/1711.08267v1","url_pdf":"http://arxiv.org/pdf/1711.08267v1.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":"graphgan-graph-representation-learning-with","repo_url":"https://github.com/hwwang55/GraphGAN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"graphgan-graph-representation-learning-with","repo_url":"https://github.com/hyunjin72/GraphGAN-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"graphgan-graph-representation-learning-with","repo_url":"https://github.com/liutongyang/GraphGAN-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"graphgan-graph-representation-learning-with","repo_url":"https://github.com/liuxinkai94/Graph-embedding","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"graphgan-graph-representation-learning-with","repo_url":"https://github.com/shuo-git/GraphGAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"computational-efficiency","task_name":"Computational Efficiency"},{"task_slug":"graph-representation-learning","task_name":"Graph Representation Learning"},{"task_slug":"link-prediction","task_name":"Link Prediction"},{"task_slug":"node-classification","task_name":"Node Classification"},{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/node-classification-on-blogcatalog-1","task":"Node Classification","dataset":"BlogCatalog","model":"GraphGAN","rank_in_archive_order":4,"of":6,"metrics":{"Accuracy":"23.20","Macro-F1":"0.221"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-wikipedia","task":"Node Classification","dataset":"Wikipedia","model":"GraphGAN","rank_in_archive_order":1,"of":6,"metrics":{"Accuracy":"21.3%","Macro-F1":"0.194"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.08267","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}