{"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/biadversarial-variational-autoencoder","title":"Biadversarial Variational Autoencoder","arxiv_id":"1902.03517","date":"2019-02-09","proceeding":null,"authors":["Arnaud Fickinger"],"abstract":"In the original version of the Variational Autoencoder, Kingma et al. assume\nGaussian distributions for the approximate posterior during the inference and\nfor the output during the generative process. This assumptions are good for\ncomputational reasons, e.g. we can easily optimize the parameters of a neural\nnetwork using the reparametrization trick and the KL divergence between two\nGaussians can be computed in closed form. However it results in blurry images\ndue to its difficulty to represent multimodal distributions. We show that using\ntwo adversarial networks, we can optimize the parameters without any Gaussian\nassumptions.","url_abs":"http://arxiv.org/abs/1902.03517v2","url_pdf":"http://arxiv.org/pdf/1902.03517v2.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":"biadversarial-variational-autoencoder","repo_url":"https://github.com/ArnaudFickinger/BAVAE","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}