{"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/discrete-variational-autoencoders","title":"Discrete Variational Autoencoders","arxiv_id":"1609.02200","date":"2016-09-07","proceeding":null,"authors":["Jason Tyler Rolfe"],"abstract":"Probabilistic models with discrete latent variables naturally capture\ndatasets composed of discrete classes. However, they are difficult to train\nefficiently, since backpropagation through discrete variables is generally not\npossible. We present a novel method to train a class of probabilistic models\nwith discrete latent variables using the variational autoencoder framework,\nincluding backpropagation through the discrete latent variables. The associated\nclass of probabilistic models comprises an undirected discrete component and a\ndirected hierarchical continuous component. The discrete component captures the\ndistribution over the disconnected smooth manifolds induced by the continuous\ncomponent. As a result, this class of models efficiently learns both the class\nof objects in an image, and their specific realization in pixels, from\nunsupervised data, and outperforms state-of-the-art methods on the\npermutation-invariant MNIST, Omniglot, and Caltech-101 Silhouettes datasets.","url_abs":"http://arxiv.org/abs/1609.02200v2","url_pdf":"http://arxiv.org/pdf/1609.02200v2.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":"discrete-variational-autoencoders","repo_url":"https://github.com/ezeeEric/DiVAE","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"discrete-variational-autoencoders","repo_url":"https://github.com/qalosim/caloqvae","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1609.02200","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}