{"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/mode-normalization","title":"Mode Normalization","arxiv_id":"1810.05466","date":"2018-10-12","proceeding":"ICLR 2019 5","authors":["Lucas Deecke","Iain Murray","Hakan Bilen"],"abstract":"Normalization methods are a central building block in the deep learning\ntoolbox. They accelerate and stabilize training, while decreasing the\ndependence on manually tuned learning rate schedules. When learning from\nmulti-modal distributions, the effectiveness of batch normalization (BN),\narguably the most prominent normalization method, is reduced. As a remedy, we\npropose a more flexible approach: by extending the normalization to more than a\nsingle mean and variance, we detect modes of data on-the-fly, jointly\nnormalizing samples that share common features. We demonstrate that our method\noutperforms BN and other widely used normalization techniques in several\nexperiments, including single and multi-task datasets.","url_abs":"http://arxiv.org/abs/1810.05466v1","url_pdf":"http://arxiv.org/pdf/1810.05466v1.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":"mode-normalization","repo_url":"https://github.com/ldeecke/mn-torch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"mode-normalization","repo_url":"https://github.com/philipperemy/mode-normalization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"mode-normalization","method_name":"Mode Normalization"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1810.05466","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}