Gaussian Process Regression with Mismatched Models

dc.creatorSollich, Peter
dc.date2001-06-22
dc.date.accessioned2026-07-07T02:41:53Z
dc.date.available2026-07-07T02:41:53Z
dc.descriptionLearning curves for Gaussian process regression are well understood when the `student' model happens to match the `teacher' (true data generation process). I derive approximations to the learning curves for the more generic case of mismatched models, and find very rich behaviour: For large input space dimensionality, where the results become exact, there are universal (student-independent) plateaux in the learning curve, with transitions in between that can exhibit arbitrarily many over-fitting maxima. In lower dimensions, plateaux also appear, and the asymptotic decay of the learning curve becomes strongly student-dependent. All predictions are confirmed by simulations.
dc.description7 pages, style file nips01e.sty included
dc.identifierhttps://arxiv.org/abs/cond-mat/0106475
dc.identifierhttp://arxiv.org/abs/cond-mat/0106475
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/17920
dc.subjectDisordered Systems and Neural Networks
dc.subjectStatistical Mechanics
dc.titleGaussian Process Regression with Mismatched Models
dc.typetext

Files

Collections