Model-Preserving Sensitivity Analysis for Families of Gaussian Distributions

Leonelli, Manuele; G¨orgen, Christiane

Publication:
Model-Preserving Sensitivity Analysis for Families of Gaussian Distributions

dc.contributor.author	Leonelli, Manuele
dc.contributor.author	G¨orgen, Christiane
dc.contributor.editor	Peter Spirtes
dc.contributor.ror	https://ror.org/02jjdwm75
dc.date.accessioned	2025-03-18T17:42:44Z
dc.date.available	2025-03-18T17:42:44Z
dc.date.issued	2020
dc.description.abstract	The accuracy of probability distributions inferred using machine-learning algorithms heavily depends on data availability and quality. In practical applications it is therefore fundamental to investigate the robustness of a statistical model to misspecification of some of its underlying probabilities. In the context of graphical models, investigations of robustness fall under the notion of sensitivity analyses. These analyses consist in varying some of the model’s probabilities or parameters and then assessing how far apart the original and the varied distributions are. However, for Gaussian graphical models, such variations usually make the original graph an incoherent representation of the model’s conditional independence structure. Here we develop an approach to sensitivity analysis which guarantees the original graph remains valid after any probability variation and we quantify the effect of such variations using different measures. To achieve this we take advantage of algebraic techniques to both concisely represent conditional independence and to provide a straightforward way of checking the validity of such relationships. Our methods are demonstrated to be robust and comparable to standard ones, which can break the conditional independence structure of the model, using an artificial example and a medical real-world application.
dc.description.peerreviewed	yes
dc.description.status	Published
dc.format	application/pdf
dc.identifier.citation	Görgen, C., & Leonelli, M. (2020). Model-preserving sensitivity analysis for families of Gaussian distributions. Journal of Machine Learning Research, 21(84), 1-32. http://jmlr.org/papers/v21/18-668.html.
dc.identifier.doi	http://jmlr.org/papers/v21/18-668.html
dc.identifier.issn	1533-7928
dc.identifier.uri	https://hdl.handle.net/20.500.14417/3665
dc.journal.title	Journal of Machine Learning Research
dc.language.iso	en
dc.page.final	32
dc.page.initial	1
dc.page.total	32
dc.publisher	JMLR
dc.relation.department	Computer Science and AI
dc.relation.entity	IE University
dc.relation.school	IE School of Science & Technology
dc.rights	Attribution 4.0 International
dc.rights.accessRights	info:eu-repo/semantics/openAccess
dc.rights.uri	https://creativecommons.org/licenses/by/4.0/deed
dc.subject.keyword	Conditional independence
dc.subject.keyword	Gaussian models
dc.subject.keyword	Graphical models
dc.subject.keyword	KullbackLeibler divergence
dc.subject.keyword	Sensitivity analysis
dc.title	Model-Preserving Sensitivity Analysis for Families of Gaussian Distributions
dc.type	info:eu-repo/semantics/article
dc.version.type	info:eu-repo/semantics/publishedVersion
dc.volume.number	21
dspace.entity.type	Publication
relation.isAuthorOfPublication	bc86b9eb-18b3-4fab-bf14-ad6f5509312f
relation.isAuthorOfPublication.latestForDiscovery	bc86b9eb-18b3-4fab-bf14-ad6f5509312f