{"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/pixelvae-a-latent-variable-model-for-natural","title":"PixelVAE: A Latent Variable Model for Natural Images","arxiv_id":"1611.05013","date":"2016-11-15","proceeding":null,"authors":["Ishaan Gulrajani","Kundan Kumar","Faruk Ahmed","Adrien Ali Taiga","Francesco Visin","David Vazquez","Aaron Courville"],"abstract":"Natural image modeling is a landmark challenge of unsupervised learning.\nVariational Autoencoders (VAEs) learn a useful latent representation and model\nglobal structure well but have difficulty capturing small details. PixelCNN\nmodels details very well, but lacks a latent code and is difficult to scale for\ncapturing large structures. We present PixelVAE, a VAE model with an\nautoregressive decoder based on PixelCNN. Our model requires very few expensive\nautoregressive layers compared to PixelCNN and learns latent codes that are\nmore compressed than a standard VAE while still capturing most non-trivial\nstructure. Finally, we extend our model to a hierarchy of latent variables at\ndifferent scales. Our model achieves state-of-the-art performance on binarized\nMNIST, competitive performance on 64x64 ImageNet, and high-quality samples on\nthe LSUN bedrooms dataset.","url_abs":"http://arxiv.org/abs/1611.05013v1","url_pdf":"http://arxiv.org/pdf/1611.05013v1.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":"pixelvae-a-latent-variable-model-for-natural","repo_url":"https://github.com/pluu2/SeparationFactor","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"model","task_name":"model"}],"methods":[{"method_slug":"pixelcnn","method_name":"PixelCNN"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.05013","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}