{"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/deep-weighted-averaging-classifiers","title":"Deep Weighted Averaging Classifiers","arxiv_id":"1811.02579","date":"2018-11-06","proceeding":null,"authors":["Dallas Card","Michael Zhang","Noah A. Smith"],"abstract":"Recent advances in deep learning have achieved impressive gains in\nclassification accuracy on a variety of types of data, including images and\ntext. Despite these gains, however, concerns have been raised about the\ncalibration, robustness, and interpretability of these models. In this paper we\npropose a simple way to modify any conventional deep architecture to\nautomatically provide more transparent explanations for classification\ndecisions, as well as an intuitive notion of the credibility of each\nprediction. Specifically, we draw on ideas from nonparametric kernel\nregression, and propose to predict labels based on a weighted sum of training\ninstances, where the weights are determined by distance in a learned\ninstance-embedding space. Working within the framework of conformal methods, we\npropose a new measure of nonconformity suggested by our model, and\nexperimentally validate the accompanying theoretical expectations,\ndemonstrating improved transparency, controlled error rates, and robustness to\nout-of-domain data, without compromising on accuracy or calibration.","url_abs":"http://arxiv.org/abs/1811.02579v2","url_pdf":"http://arxiv.org/pdf/1811.02579v2.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":"deep-weighted-averaging-classifiers","repo_url":"https://github.com/dallascard/DWAC","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"deep-weighted-averaging-classifiers","repo_url":"https://github.com/joe32140/CSCI-7000-replication","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification","task_name":"General Classification"}],"methods":[{"method_slug":"interpretability","method_name":"Interpretability"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.02579","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}