{"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/disconnected-manifold-learning-for-generative","title":"Disconnected Manifold Learning for Generative Adversarial Networks","arxiv_id":"1806.00880","date":"2018-06-03","proceeding":"NeurIPS 2018 12","authors":["Mahyar Khayatkhoei","Ahmed Elgammal","Maneesh Singh"],"abstract":"Natural images may lie on a union of disjoint manifolds rather than one\nglobally connected manifold, and this can cause several difficulties for the\ntraining of common Generative Adversarial Networks (GANs). In this work, we\nfirst show that single generator GANs are unable to correctly model a\ndistribution supported on a disconnected manifold, and investigate how sample\nquality, mode dropping and local convergence are affected by this. Next, we\nshow how using a collection of generators can address this problem, providing\nnew insights into the success of such multi-generator GANs. Finally, we explain\nthe serious issues caused by considering a fixed prior over the collection of\ngenerators and propose a novel approach for learning the prior and inferring\nthe necessary number of generators without any supervision. Our proposed\nmodifications can be applied on top of any other GAN model to enable learning\nof distributions supported on disconnected manifolds. We conduct several\nexperiments to illustrate the aforementioned shortcoming of GANs, its\nconsequences in practice, and the effectiveness of our proposed modifications\nin alleviating these issues.","url_abs":"http://arxiv.org/abs/1806.00880v3","url_pdf":"http://arxiv.org/pdf/1806.00880v3.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":"disconnected-manifold-learning-for-generative","repo_url":"https://github.com/mahyarkoy/dmgan_release","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.00880","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}