{"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/multi-component-image-translation-for-deep","title":"Multi-component Image Translation for Deep Domain Generalization","arxiv_id":"1812.08974","date":"2018-12-21","proceeding":null,"authors":["Mohammad Mahfujur Rahman","Clinton Fookes","Mahsa Baktashmotlagh","Sridha Sridharan"],"abstract":"Domain adaption (DA) and domain generalization (DG) are two closely related\nmethods which are both concerned with the task of assigning labels to an\nunlabeled data set. The only dissimilarity between these approaches is that DA\ncan access the target data during the training phase, while the target data is\ntotally unseen during the training phase in DG. The task of DG is challenging\nas we have no earlier knowledge of the target samples. If DA methods are\napplied directly to DG by a simple exclusion of the target data from training,\npoor performance will result for a given task. In this paper, we tackle the\ndomain generalization challenge in two ways. In our first approach, we propose\na novel deep domain generalization architecture utilizing synthetic data\ngenerated by a Generative Adversarial Network (GAN). The discrepancy between\nthe generated images and synthetic images is minimized using existing domain\ndiscrepancy metrics such as maximum mean discrepancy or correlation alignment.\nIn our second approach, we introduce a protocol for applying DA methods to a DG\nscenario by excluding the target data from the training phase, splitting the\nsource data to training and validation parts, and treating the validation data\nas target data for DA. We conduct extensive experiments on four cross-domain\nbenchmark datasets. Experimental results signify our proposed model outperforms\nthe current state-of-the-art methods for DG.","url_abs":"http://arxiv.org/abs/1812.08974v1","url_pdf":"http://arxiv.org/pdf/1812.08974v1.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":"domain-generalization","task_name":"Domain Generalization"},{"task_slug":null,"task_name":"Generative Adversarial Network"},{"task_slug":"translation","task_name":"Translation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/domain-generalization-on-pacs-2","task":"Domain Generalization","dataset":"PACS","model":"JAN-COMBO (Alexnet)","rank_in_archive_order":125,"of":133,"metrics":{"Average Accuracy":"69.45"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1812.08974","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}