{"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/vae-with-a-vampprior","title":"VAE with a VampPrior","arxiv_id":"1705.07120","date":"2017-05-19","proceeding":null,"authors":["Jakub M. Tomczak","Max Welling"],"abstract":"Many different methods to train deep generative models have been introduced\nin the past. In this paper, we propose to extend the variational auto-encoder\n(VAE) framework with a new type of prior which we call \"Variational Mixture of\nPosteriors\" prior, or VampPrior for short. The VampPrior consists of a mixture\ndistribution (e.g., a mixture of Gaussians) with components given by\nvariational posteriors conditioned on learnable pseudo-inputs. We further\nextend this prior to a two layer hierarchical model and show that this\narchitecture with a coupled prior and posterior, learns significantly better\nmodels. The model also avoids the usual local optima issues related to useless\nlatent dimensions that plague VAEs. We provide empirical studies on six\ndatasets, namely, static and binary MNIST, OMNIGLOT, Caltech 101 Silhouettes,\nFrey Faces and Histopathology patches, and show that applying the hierarchical\nVampPrior delivers state-of-the-art results on all datasets in the unsupervised\npermutation invariant setting and the best results or comparable to SOTA\nmethods for the approach with convolutional networks.","url_abs":"http://arxiv.org/abs/1705.07120v5","url_pdf":"http://arxiv.org/pdf/1705.07120v5.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":"vae-with-a-vampprior","repo_url":"https://github.com/jmtomczak/vae_vampprior","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/KenzaB27/VAE-VampPrior","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/belaalb/CEVAE-VampPrior","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/clementchadebec/benchmark_VAE","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/morpheusthewhite/vae-vampprior","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/sebastiaanver/HVAE-MLAdv","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/MindSpore-scientific-2/code-11/tree/main/VAE-Creative-Discovery-using-QD-Search","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/MindSpore-scientific-2/code-2/tree/main/VAE-Creative-Discovery-using-QD-Search","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/MindSpore-scientific-2/code-8/tree/main/VAE-Creative-Discovery-using-QD-Search","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/MindSpore-scientific/code-10/tree/main/VAE-Creative-Discovery-using-QD-Search","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/MindSpore-scientific/code-11/tree/main/VAE-Creative-Discovery-using-QD-Search","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"vae-with-a-vampprior","repo_url":"https://github.com/MindSpore-scientific/code-8/tree/main/VAE-Creative-Discovery-using-QD-Search","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1705.07120","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}