{"id":62,"date":"2021-01-01T18:45:07","date_gmt":"2021-01-01T17:45:07","guid":{"rendered":"https:\/\/projets.litislab.fr\/api\/?p=62"},"modified":"2024-11-17T21:43:31","modified_gmt":"2024-11-17T20:43:31","slug":"learning-output-embeddings-in-structured-prediction-submitted-to-aistats21","status":"publish","type":"post","link":"https:\/\/projets.litislab.fr\/api\/2021\/01\/01\/learning-output-embeddings-in-structured-prediction-submitted-to-aistats21\/","title":{"rendered":"Learning Output Embeddings in Structured Prediction (submitted to AISTATS&#8217;21)"},"content":{"rendered":"\n<div class=\"page\" title=\"Page 7\">\n<div class=\"section\">\n<div class=\"layoutArea\">\n<div class=\"column\">\n<p>By&nbsp;Luc Brogat-Motte, Alessandro Rudi, Ce\u0301line Brouard, Juho Rousu, Florence d&#8217;Alche\u0301-Buc.<\/p>\n<p>Submitted to AISTATS, 2021<\/p>\n<p><a href=\"https:\/\/arxiv.org\/abs\/2007.14703\">https:\/\/arxiv.org\/abs\/2007.14703<\/a>.<\/p>\n<\/div>\n<p><strong>Abstract.&nbsp;<\/strong>A powerful and flexible approach to structured prediction consists in embedding the structured objects to be predicted into a feature space of possibly infinite dimension by means of output kernels, and then, solving a regression problem in this output space. A prediction in the original space is computed by solving a pre-image problem. In such an approach, the embedding, linked to the target loss, is defined prior to the learning phase. In this work, we propose to jointly learn a finite approximation of the output embedding and the regression function into the new feature space. For that purpose, we leverage a priori information on the outputs and also unexploited unsupervised output data, which are both often available in structured prediction problems. We prove that the resulting structured predictor is a consistent estimator, and derive an excess risk bound. Moreover, the novel structured prediction tool enjoys a significantly smaller computational complexity than former output kernel methods. The approach empirically tested on various structured prediction problems reveals to be versatile and able to handle large datasets.<\/p>\n<\/div>\n<\/div>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>By&nbsp;Luc Brogat-Motte, Alessandro Rudi, Ce\u0301line Brouard, Juho Rousu, Florence d&#8217;Alche\u0301-Buc. Submitted to AISTATS, 2021 https:\/\/arxiv.org\/abs\/2007.14703. Abstract.&nbsp;A powerful and flexible approach to structured prediction consists in embedding the structured objects to be predicted into a feature space of possibly infinite dimension by means of output kernels, and then, solving a regression problem in this output space. &hellip; <a href=\"https:\/\/projets.litislab.fr\/api\/2021\/01\/01\/learning-output-embeddings-in-structured-prediction-submitted-to-aistats21\/\" class=\"more-link\">Continue reading <span class=\"screen-reader-text\">Learning Output Embeddings in Structured Prediction (submitted to AISTATS&#8217;21)<\/span> <span class=\"meta-nav\">&rarr;<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[13,30,16,25,31],"class_list":["post-62","post","type-post","status-publish","format-standard","hentry","category-conferences","tag-kernel-methods","tag-output-embedding","tag-pre-image-problem","tag-representation-learning","tag-structured-prediction"],"_links":{"self":[{"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/posts\/62","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/comments?post=62"}],"version-history":[{"count":1,"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/posts\/62\/revisions"}],"predecessor-version":[{"id":63,"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/posts\/62\/revisions\/63"}],"wp:attachment":[{"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/media?parent=62"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/categories?post=62"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/projets.litislab.fr\/api\/wp-json\/wp\/v2\/tags?post=62"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}