{"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/an-extended-framework-for-marginalized-domain","title":"An Extended Framework for Marginalized Domain Adaptation","arxiv_id":"1702.05993","date":"2017-02-20","proceeding":null,"authors":["Gabriela Csurka","Boris Chidlovski","Stephane Clinchant","Sophia Michel"],"abstract":"We propose an extended framework for marginalized domain adaptation, aimed at\naddressing unsupervised, supervised and semi-supervised scenarios. We argue\nthat the denoising principle should be extended to explicitly promote\ndomain-invariant features as well as help the classification task. Therefore we\npropose to jointly learn the data auto-encoders and the target classifiers.\nFirst, in order to make the denoised features domain-invariant, we propose a\ndomain regularization that may be either a domain prediction loss or a maximum\nmean discrepancy between the source and target data. The noise marginalization\nin this case is reduced to solving the linear matrix system $AX=B$ which has a\nclosed-form solution. Second, in order to help the classification, we include a\nclass regularization term. Adding this component reduces the learning problem\nto solving a Sylvester linear matrix equation $AX+BX=C$, for which an efficient\niterative procedure exists as well. We did an extensive study to assess how\nthese regularization terms improve the baseline performance in the three domain\nadaptation scenarios and present experimental results on two image and one text\nbenchmark datasets, conventionally used for validating domain adaptation\nmethods. We report our findings and comparison with state-of-the-art methods.","url_abs":"http://arxiv.org/abs/1702.05993v1","url_pdf":"http://arxiv.org/pdf/1702.05993v1.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":"an-extended-framework-for-marginalized-domain","repo_url":"https://github.com/sclincha/xrce_msda_da_regularization","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}