{"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/flow-gan-combining-maximum-likelihood-and","title":"Flow-GAN: Combining Maximum Likelihood and Adversarial Learning in Generative Models","arxiv_id":"1705.08868","date":"2017-05-24","proceeding":null,"authors":["Aditya Grover","Manik Dhar","Stefano Ermon"],"abstract":"Adversarial learning of probabilistic models has recently emerged as a\npromising alternative to maximum likelihood. Implicit models such as generative\nadversarial networks (GAN) often generate better samples compared to explicit\nmodels trained by maximum likelihood. Yet, GANs sidestep the characterization\nof an explicit density which makes quantitative evaluations challenging. To\nbridge this gap, we propose Flow-GANs, a generative adversarial network for\nwhich we can perform exact likelihood evaluation, thus supporting both\nadversarial and maximum likelihood training. When trained adversarially,\nFlow-GANs generate high-quality samples but attain extremely poor\nlog-likelihood scores, inferior even to a mixture model memorizing the training\ndata; the opposite is true when trained by maximum likelihood. Results on MNIST\nand CIFAR-10 demonstrate that hybrid training can attain high held-out\nlikelihoods while retaining visual fidelity in the generated samples.","url_abs":"http://arxiv.org/abs/1705.08868v2","url_pdf":"http://arxiv.org/pdf/1705.08868v2.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":"flow-gan-combining-maximum-likelihood-and","repo_url":"https://github.com/ermongroup/flow-gan","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"flow-gan-combining-maximum-likelihood-and","repo_url":"https://github.com/RuiLiFeng/flow-gan","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"flow-gan-combining-maximum-likelihood-and","repo_url":"https://github.com/eyalbetzalel/flow-gan","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":null,"task_name":"Generative Adversarial Network"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1705.08868","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}