{"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/scatter-component-analysis-a-unified","title":"Scatter Component Analysis: A Unified Framework for Domain Adaptation and Domain Generalization","arxiv_id":"1510.04373","date":"2015-10-15","proceeding":null,"authors":["Muhammad Ghifary","David Balduzzi","W. Bastiaan Kleijn","Mengjie Zhang"],"abstract":"This paper addresses classification tasks on a particular target domain in\nwhich labeled training data are only available from source domains different\nfrom (but related to) the target. Two closely related frameworks, domain\nadaptation and domain generalization, are concerned with such tasks, where the\nonly difference between those frameworks is the availability of the unlabeled\ntarget data: domain adaptation can leverage unlabeled target information, while\ndomain generalization cannot. We propose Scatter Component Analyis (SCA), a\nfast representation learning algorithm that can be applied to both domain\nadaptation and domain generalization. SCA is based on a simple geometrical\nmeasure, i.e., scatter, which operates on reproducing kernel Hilbert space. SCA\nfinds a representation that trades between maximizing the separability of\nclasses, minimizing the mismatch between domains, and maximizing the\nseparability of data; each of which is quantified through scatter. The\noptimization problem of SCA can be reduced to a generalized eigenvalue problem,\nwhich results in a fast and exact solution. Comprehensive experiments on\nbenchmark cross-domain object recognition datasets verify that SCA performs\nmuch faster than several state-of-the-art algorithms and also provides\nstate-of-the-art classification accuracy in both domain adaptation and domain\ngeneralization. We also show that scatter can be used to establish a\ntheoretical generalization bound in the case of domain adaptation.","url_abs":"http://arxiv.org/abs/1510.04373v2","url_pdf":"http://arxiv.org/pdf/1510.04373v2.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":"classification","task_name":"General Classification"},{"task_slug":"object-recognition","task_name":"Object Recognition"},{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/domain-adaptation-on-office-caltech","task":"Domain Adaptation","dataset":"Office-Caltech","model":"SCA[[Ghifary et al.2016]]","rank_in_archive_order":7,"of":8,"metrics":{"Average Accuracy":"85.9"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1510.04373","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}