{"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/fix-your-classifier-the-marginal-value-of","title":"Fix your classifier: the marginal value of training the last weight layer","arxiv_id":"1801.04540","date":"2018-01-14","proceeding":"ICLR 2018 1","authors":["Elad Hoffer","Itay Hubara","Daniel Soudry"],"abstract":"Neural networks are commonly used as models for classification for a wide\nvariety of tasks. Typically, a learned affine transformation is placed at the\nend of such models, yielding a per-class value used for classification. This\nclassifier can have a vast number of parameters, which grows linearly with the\nnumber of possible classes, thus requiring increasingly more resources. In this\nwork we argue that this classifier can be fixed, up to a global scale constant,\nwith little or no loss of accuracy for most tasks, allowing memory and\ncomputational benefits. Moreover, we show that by initializing the classifier\nwith a Hadamard matrix we can speed up inference as well. We discuss the\nimplications for current understanding of neural network models.","url_abs":"http://arxiv.org/abs/1801.04540v2","url_pdf":"http://arxiv.org/pdf/1801.04540v2.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":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/eladhoffer/fix_your_classifier","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/eladhoffer/convNet.pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/vaapopescu/gradient-pruning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/MindCode-4/code-11/tree/main/fixed-classifier","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/MindCode-4/code-7/tree/main/fixed-classifier","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/MindSpore-scientific-2/code-12/tree/main/fixed-classifier","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"fix-your-classifier-the-marginal-value-of","repo_url":"https://github.com/reeered/FixedClassifier","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1801.04540","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}