{"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/adversarial-dropout-regularization","title":"Adversarial Dropout Regularization","arxiv_id":"1711.01575","date":"2017-11-05","proceeding":"ICLR 2018 1","authors":["Kuniaki Saito","Yoshitaka Ushiku","Tatsuya Harada","Kate Saenko"],"abstract":"We present a method for transferring neural representations from label-rich\nsource domains to unlabeled target domains. Recent adversarial methods proposed\nfor this task learn to align features across domains by fooling a special\ndomain critic network. However, a drawback of this approach is that the critic\nsimply labels the generated features as in-domain or not, without considering\nthe boundaries between classes. This can lead to ambiguous features being\ngenerated near class boundaries, reducing target classification accuracy. We\npropose a novel approach, Adversarial Dropout Regularization (ADR), to\nencourage the generator to output more discriminative features for the target\ndomain. Our key idea is to replace the critic with one that detects\nnon-discriminative features, using dropout on the classifier network. The\ngenerator then learns to avoid these areas of the feature space and thus\ncreates better features. We apply our ADR approach to the problem of\nunsupervised domain adaptation for image classification and semantic\nsegmentation tasks, and demonstrate significant improvement over the state of\nthe art. We also show that our approach can be used to train Generative\nAdversarial Networks for semi-supervised learning.","url_abs":"http://arxiv.org/abs/1711.01575v3","url_pdf":"http://arxiv.org/pdf/1711.01575v3.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":[],"tasks":[{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"unsupervised-domain-adaptation","task_name":"Unsupervised Domain Adaptation"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"dropout","method_name":"Dropout"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/synthetic-to-real-translation-on-syn2real-c","task":"Synthetic-to-Real Translation","dataset":"Syn2Real-C","model":"ADR","rank_in_archive_order":2,"of":6,"metrics":{"Accuracy":"74.8"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.01575","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}