{"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/multi-view-data-generation-without-view","title":"Multi-View Data Generation Without View Supervision","arxiv_id":"1711.00305","date":"2017-11-01","proceeding":"ICLR 2018 1","authors":["Mickaël Chen","Ludovic Denoyer","Thierry Artières"],"abstract":"The development of high-dimensional generative models has recently gained a\ngreat surge of interest with the introduction of variational auto-encoders and\ngenerative adversarial neural networks. Different variants have been proposed\nwhere the underlying latent space is structured, for example, based on\nattributes describing the data to generate. We focus on a particular problem\nwhere one aims at generating samples corresponding to a number of objects under\nvarious views. We assume that the distribution of the data is driven by two\nindependent latent factors: the content, which represents the intrinsic\nfeatures of an object, and the view, which stands for the settings of a\nparticular observation of that object. Therefore, we propose a generative model\nand a conditional variant built on such a disentangled latent space. This\napproach allows us to generate realistic samples corresponding to various\nobjects in a high variety of views. Unlike many multi-view approaches, our\nmodel doesn't need any supervision on the views but only on the content.\nCompared to other conditional generation approaches that are mostly based on\nbinary or categorical attributes, we make no such assumption about the factors\nof variations. Our model can be used on problems with a huge, potentially\ninfinite, number of categories. We experiment it on four image datasets on\nwhich we demonstrate the effectiveness of the model and its ability to\ngeneralize.","url_abs":"http://arxiv.org/abs/1711.00305v2","url_pdf":"http://arxiv.org/pdf/1711.00305v2.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":"multi-view-data-generation-without-view","repo_url":"https://github.com/mickaelChen/GMV","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.00305","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}