{"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/boosting-domain-adaptation-by-discovering","title":"Boosting Domain Adaptation by Discovering Latent Domains","arxiv_id":"1805.01386","date":"2018-05-03","proceeding":"CVPR 2018 6","authors":["Massimiliano Mancini","Lorenzo Porzi","Samuel Rota Bulò","Barbara Caputo","Elisa Ricci"],"abstract":"Current Domain Adaptation (DA) methods based on deep architectures assume\nthat the source samples arise from a single distribution. However, in practice,\nmost datasets can be regarded as mixtures of multiple domains. In these cases\nexploiting single-source DA methods for learning target classifiers may lead to\nsub-optimal, if not poor, results. In addition, in many applications it is\ndifficult to manually provide the domain labels for all source data points,\ni.e. latent domains should be automatically discovered. This paper introduces a\nnovel Convolutional Neural Network (CNN) architecture which (i) automatically\ndiscovers latent domains in visual datasets and (ii) exploits this information\nto learn robust target classifiers. Our approach is based on the introduction\nof two main components, which can be embedded into any existing CNN\narchitecture: (i) a side branch that automatically computes the assignment of a\nsource sample to a latent domain and (ii) novel layers that exploit domain\nmembership information to appropriately align the distribution of the CNN\ninternal feature representations to a reference distribution. We test our\napproach on publicly-available datasets, showing that it outperforms\nstate-of-the-art multi-source DA methods by a large margin.","url_abs":"http://arxiv.org/abs/1805.01386v1","url_pdf":"http://arxiv.org/pdf/1805.01386v1.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":"boosting-domain-adaptation-by-discovering","repo_url":"https://github.com/mancinimassimiliano/latent_domains_DA","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"boosting-domain-adaptation-by-discovering","repo_url":"https://github.com/mancinimassimiliano/pytorch_wbn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.01386","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}