{"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/notes-on-noise-contrastive-estimation-and","title":"Notes on Noise Contrastive Estimation and Negative Sampling","arxiv_id":"1410.8251","date":"2014-10-30","proceeding":null,"authors":["Chris Dyer"],"abstract":"Estimating the parameters of probabilistic models of language such as maxent\nmodels and probabilistic neural models is computationally difficult since it\ninvolves evaluating partition functions by summing over an entire vocabulary,\nwhich may be millions of word types in size. Two closely related\nstrategies---noise contrastive estimation (Mnih and Teh, 2012; Mnih and\nKavukcuoglu, 2013; Vaswani et al., 2013) and negative sampling (Mikolov et al.,\n2012; Goldberg and Levy, 2014)---have emerged as popular solutions to this\ncomputational problem, but some confusion remains as to which is more\nappropriate and when. This document explicates their relationships to each\nother and to other estimation techniques. The analysis shows that, although\nthey are superficially similar, NCE is a general parameter estimation technique\nthat is asymptotically unbiased, while negative sampling is best understood as\na family of binary classification models that are useful for learning word\nrepresentations but not as a general-purpose estimator.","url_abs":"http://arxiv.org/abs/1410.8251v1","url_pdf":"http://arxiv.org/pdf/1410.8251v1.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":"notes-on-noise-contrastive-estimation-and","repo_url":"https://github.com/dhyeon/ingredient-vectors","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"notes-on-noise-contrastive-estimation-and","repo_url":"https://github.com/hithisisdhara/doc2vec","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"notes-on-noise-contrastive-estimation-and","repo_url":"https://github.com/inejc/paragraph-vectors","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"binary-classification","task_name":"Binary Classification"},{"task_slug":"parameter-estimation","task_name":"parameter estimation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1410.8251","atlas_url":"https://app.syntology.ai/?focus=1410.8251","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}