{"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/best-sources-forward-domain-generalization","title":"Best sources forward: domain generalization through source-specific nets","arxiv_id":"1806.05810","date":"2018-06-15","proceeding":null,"authors":["Massimiliano Mancini","Samuel Rota Bulò","Barbara Caputo","Elisa Ricci"],"abstract":"A long standing problem in visual object categorization is the ability of\nalgorithms to generalize across different testing conditions. The problem has\nbeen formalized as a covariate shift among the probability distributions\ngenerating the training data (source) and the test data (target) and several\ndomain adaptation methods have been proposed to address this issue. While these\napproaches have considered the single source-single target scenario, it is\nplausible to have multiple sources and require adaptation to any possible\ntarget domain. This last scenario, named Domain Generalization (DG), is the\nfocus of our work. Differently from previous DG methods which learn domain\ninvariant representations from source data, we design a deep network with\nmultiple domain-specific classifiers, each associated to a source domain. At\ntest time we estimate the probabilities that a target sample belongs to each\nsource domain and exploit them to optimally fuse the classifiers predictions.\nTo further improve the generalization ability of our model, we also introduced\na domain agnostic component supporting the final classifier. Experiments on two\npublic benchmarks demonstrate the power of our approach.","url_abs":"http://arxiv.org/abs/1806.05810v1","url_pdf":"http://arxiv.org/pdf/1806.05810v1.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":"object-categorization","task_name":"Object Categorization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/domain-generalization-on-pacs-2","task":"Domain Generalization","dataset":"PACS","model":"BestSources (Alexnet)","rank_in_archive_order":122,"of":133,"metrics":{"Average Accuracy":"70.30"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.05810","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}