{"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/the-extreme-value-machine","title":"The Extreme Value Machine","arxiv_id":"1506.06112","date":"2015-06-19","proceeding":null,"authors":["Ethan M. Rudd","Lalit P. Jain","Walter J. Scheirer","Terrance E. Boult"],"abstract":"It is often desirable to be able to recognize when inputs to a recognition\nfunction learned in a supervised manner correspond to classes unseen at\ntraining time. With this ability, new class labels could be assigned to these\ninputs by a human operator, allowing them to be incorporated into the\nrecognition function --- ideally under an efficient incremental update\nmechanism. While good algorithms that assume inputs from a fixed set of classes\nexist, e.g., artificial neural networks and kernel machines, it is not\nimmediately obvious how to extend them to perform incremental learning in the\npresence of unknown query classes. Existing algorithms take little to no\ndistributional information into account when learning recognition functions and\nlack a strong theoretical foundation. We address this gap by formulating a\nnovel, theoretically sound classifier --- the Extreme Value Machine (EVM). The\nEVM has a well-grounded interpretation derived from statistical Extreme Value\nTheory (EVT), and is the first classifier to be able to perform nonlinear\nkernel-free variable bandwidth incremental learning. Compared to other\nclassifiers in the same deep network derived feature space, the EVM is accurate\nand efficient on an established benchmark partition of the ImageNet dataset.","url_abs":"http://arxiv.org/abs/1506.06112v4","url_pdf":"http://arxiv.org/pdf/1506.06112v4.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":[],"tasks":[{"task_slug":"incremental-learning","task_name":"Incremental Learning"}],"methods":[{"method_slug":"evm","method_name":"EVM"}],"datasets_introduced":[],"methods_introduced":[{"slug":"evm","name":"EVM","full_name":"Extreme Value Machine"}],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1506.06112","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}