Show simple item record

dc.contributor.authorKügelgen, Julius vonen
dc.contributor.authorMey, Alexanderen
dc.contributor.authorLoog, Marcoen
dc.date.accessioned2019-12-17T12:32:24Z
dc.date.available2019-12-17T12:32:24Z
dc.identifier.urihttps://www.repository.cam.ac.uk/handle/1810/300021
dc.description.abstractCurrent methods for covariate-shift adaptation use unlabelled data to compute importance weights or domain-invariant features, while the final model is trained on labelled data only. Here, we consider a particular case of covariate shift which allows us also to learn from unlabelled data, that is, combining adaptation and semi-supervised learning. Using ideas from causality, we argue that this requires learning with both causes, X_C, and effects, X_E, of a target variable, Y, and show how this setting leads to what we call a semi-generative model, P(Y,X_E|X_C,θ). Our approach is robust to domain shifts in the distribution of causal features and leverages unlabelled data by learning a direct map from causes to effects. Experiments on synthetic data demonstrate significant improvements in classification over purely-supervised and importance-weighting baselines.
dc.publisherPMLR
dc.subjectstat.MLen
dc.subjectstat.MLen
dc.subjectcs.LGen
dc.titleSemi-Generative Modelling: Covariate-Shift Adaptation with Cause and Effect Featuresen
dc.typeConference Object
dc.identifier.doi10.17863/CAM.47092
dcterms.dateAccepted2018-12-22en
rioxxterms.versionVoR*
rioxxterms.licenseref.urihttp://creativecommons.org/licenses/by/4.0/en
rioxxterms.licenseref.startdate2018-12-22en
rioxxterms.typeConference Paper/Proceeding/Abstracten


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record