{"title":"Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling","status":"public","publication_status":"published","intvolume":" 51","publisher":"Elsevier","_id":"2165","user_id":"4435EBFC-F248-11E8-B48F-1D18A9856A87","year":"2014","date_created":"2018-12-11T11:56:05Z","date_updated":"2021-01-12T06:55:43Z","oa_version":"Submitted Version","volume":51,"author":[{"full_name":"Morvant, Emilie","orcid":"0000-0002-8301-7240","id":"4BAC2A72-F248-11E8-B48F-1D18A9856A87","last_name":"Morvant","first_name":"Emilie"}],"type":"journal_article","extern":"1","abstract":[{"lang":"eng","text":"In machine learning, the domain adaptation problem arrives when the test (tar-get) and the train (source) data are generated from different distributions. A key applied issue is thus the design of algorithms able to generalize on a new distribution, for which we have no label information. We focus on learning classification models defined as a weighted majority vote over a set of real-valued functions. In this context, Germain et al. (2013) have shown that a measure of disagreement between these functions is crucial to control. The core of this measure is a theoretical bound—the C-bound (Lacasse et al., 2007)—which involves the disagreement and leads to a well performing majority vote learn-ing algorithm in usual non-adaptative supervised setting: MinCq. In this work,we propose a framework to extend MinCq to a domain adaptation scenario.This procedure takes advantage of the recent perturbed variation divergence between distributions proposed by Harel and Mannor (2012). Justified by a theoretical bound on the target risk of the vote, we provide to MinCq a tar-get sample labeled thanks to a perturbed variation-based self-labeling focused on the regions where the source and target marginals appear similar. We also study the influence of our self-labeling, from which we deduce an original process for tuning the hyperparameters. Finally, our framework called PV-MinCq shows very promising results on a rotation and translation synthetic problem."}],"publist_id":"4819","ec_funded":1,"quality_controlled":"1","page":"37-43","project":[{"_id":"2532554C-B435-11E9-9278-68D0E5697425","grant_number":"308036","call_identifier":"FP7","name":"Lifelong Learning of Visual Scene Understanding"}],"publication":"Pattern Recognition Letters","main_file_link":[{"open_access":"1","url":"http://arxiv.org/abs/1410.0334"}],"oa":1,"citation":{"chicago":"Morvant, Emilie. “Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling.” Pattern Recognition Letters. Elsevier, 2014. https://doi.org/10.1016/j.patrec.2014.08.013.","mla":"Morvant, Emilie. “Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling.” Pattern Recognition Letters, vol. 51, Elsevier, 2014, pp. 37–43, doi:10.1016/j.patrec.2014.08.013.","short":"E. Morvant, Pattern Recognition Letters 51 (2014) 37–43.","ista":"Morvant E. 2014. Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling. Pattern Recognition Letters. 51, 37–43.","ieee":"E. Morvant, “Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling,” Pattern Recognition Letters, vol. 51. Elsevier, pp. 37–43, 2014.","apa":"Morvant, E. (2014). Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling. Pattern Recognition Letters. Elsevier. https://doi.org/10.1016/j.patrec.2014.08.013","ama":"Morvant E. Domain Adaptation of Weighted Majority Votes via Perturbed Variation-Based Self-Labeling. Pattern Recognition Letters. 2014;51:37-43. doi:10.1016/j.patrec.2014.08.013"},"language":[{"iso":"eng"}],"date_published":"2014-10-01T00:00:00Z","doi":"10.1016/j.patrec.2014.08.013","month":"10","day":"01"}