{"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-reconstruction-classification-networks","title":"Deep Reconstruction-Classification Networks for Unsupervised Domain Adaptation","arxiv_id":"1607.03516","date":"2016-07-12","proceeding":null,"authors":["Muhammad Ghifary","W. Bastiaan Kleijn","Mengjie Zhang","David Balduzzi","Wen Li"],"abstract":"In this paper, we propose a novel unsupervised domain adaptation algorithm\nbased on deep learning for visual object recognition. Specifically, we design a\nnew model called Deep Reconstruction-Classification Network (DRCN), which\njointly learns a shared encoding representation for two tasks: i) supervised\nclassification of labeled source data, and ii) unsupervised reconstruction of\nunlabeled target data.In this way, the learnt representation not only preserves\ndiscriminability, but also encodes useful information from the target domain.\nOur new DRCN model can be optimized by using backpropagation similarly as the\nstandard neural networks.\n  We evaluate the performance of DRCN on a series of cross-domain object\nrecognition tasks, where DRCN provides a considerable improvement (up to ~8% in\naccuracy) over the prior state-of-the-art algorithms. Interestingly, we also\nobserve that the reconstruction pipeline of DRCN transforms images from the\nsource domain into images whose appearance resembles the target dataset. This\nsuggests that DRCN's performance is due to constructing a single composite\nrepresentation that encodes information about both the structure of target\nimages and the classification of source images. Finally, we provide a formal\nanalysis to justify the algorithm's objective in domain adaptation context.","url_abs":"http://arxiv.org/abs/1607.03516v2","url_pdf":"http://arxiv.org/pdf/1607.03516v2.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-reconstruction-classification-networks","repo_url":"https://github.com/ghif/drcn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"deep-reconstruction-classification-networks","repo_url":"https://github.com/pmirallesr/DRCN-Torch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"object-recognition","task_name":"Object Recognition"},{"task_slug":"unsupervised-domain-adaptation","task_name":"Unsupervised Domain Adaptation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1607.03516","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}