{"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/fitted-learning-models-with-awareness-of","title":"Fitted Learning: Models with Awareness of their Limits","arxiv_id":"1609.02226","date":"2016-09-07","proceeding":null,"authors":["Navid Kardan","Kenneth O. Stanley"],"abstract":"Though deep learning has pushed the boundaries of classification forward, in\nrecent years hints of the limits of standard classification have begun to\nemerge. Problems such as fooling, adding new classes over time, and the need to\nretrain learning models only for small changes to the original problem all\npoint to a potential shortcoming in the classic classification regime, where a\ncomprehensive a priori knowledge of the possible classes or concepts is\ncritical. Without such knowledge, classifiers misjudge the limits of their\nknowledge and overgeneralization therefore becomes a serious obstacle to\nconsistent performance. In response to these challenges, this paper extends the\nclassic regime by reframing classification instead with the assumption that\nconcepts present in the training set are only a sample of the hypothetical\nfinal set of concepts. To bring learning models into this new paradigm, a novel\nelaboration of standard architectures called the competitive overcomplete\noutput layer (COOL) neural network is introduced. Experiments demonstrate the\neffectiveness of COOL by applying it to fooling, separable concept learning,\none-class neural networks, and standard classification benchmarks. The results\nsuggest that, unlike conventional classifiers, the amount of generalization in\nCOOL networks can be tuned to match the problem.","url_abs":"http://arxiv.org/abs/1609.02226v4","url_pdf":"http://arxiv.org/pdf/1609.02226v4.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":"fitted-learning-models-with-awareness-of","repo_url":"https://github.com/ndkn/fitted-learning","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1609.02226","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}