{"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/adversarial-feedback-loop","title":"Adversarial Feedback Loop","arxiv_id":"1811.08126","date":"2018-11-20","proceeding":"ICCV 2019 10","authors":["Firas Shama","Roey Mechrez","Alon Shoshan","Lihi Zelnik-Manor"],"abstract":"Thanks to their remarkable generative capabilities, GANs have gained great\npopularity, and are used abundantly in state-of-the-art methods and\napplications. In a GAN based model, a discriminator is trained to learn the\nreal data distribution. To date, it has been used only for training purposes,\nwhere it's utilized to train the generator to provide real-looking outputs. In\nthis paper we propose a novel method that makes an explicit use of the\ndiscriminator in test-time, in a feedback manner in order to improve the\ngenerator results. To the best of our knowledge it is the first time a\ndiscriminator is involved in test-time. We claim that the discriminator holds\nsignificant information on the real data distribution, that could be useful for\ntest-time as well, a potential that has not been explored before.\n  The approach we propose does not alter the conventional training stage. At\ntest-time, however, it transfers the output from the generator into the\ndiscriminator, and uses feedback modules (convolutional blocks) to translate\nthe features of the discriminator layers into corrections to the features of\nthe generator layers, which are used eventually to get a better generator\nresult. Our method can contribute to both conditional and unconditional GANs.\nAs demonstrated by our experiments, it can improve the results of\nstate-of-the-art networks for super-resolution, and image generation.","url_abs":"http://arxiv.org/abs/1811.08126v1","url_pdf":"http://arxiv.org/pdf/1811.08126v1.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":"adversarial-feedback-loop","repo_url":"https://github.com/shamafiras/AFL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"super-resolution","task_name":"Super-Resolution"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1811.08126","atlas_url":"https://app.syntology.ai/?focus=1811.08126","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}