Detecting periodicities with Gaussian processes

Abstract : We consider the problem of detecting and quantifying the periodic component of a function given noise-corrupted observations of a limited number of input/output tuples. Our approach is based on Gaussian process regression, which provides a flexible non-parametric framework for modelling periodic data. We introduce a novel decomposition of the covariance function as the sum of periodic and aperiodic kernels. This decomposition allows for the creation of sub-models which capture the periodic nature of the signal and its complement. To quantify the periodicity of the signal, we derive a periodicity ratio which reflects the uncertainty in the fitted sub-models. Although the method can be applied to many kernels, we give a special emphasis to the Matérn family, from the expression of the reproducing kernel Hilbert space inner product to the implementation of the associated periodic kernels in a Gaussian process toolkit. The proposed method is illustrated by considering the detection of periodically expressed genes in the arabidopsis genome.
Type de document :
Article dans une revue
PeerJ Computer Science, PeerJ, 2016, 2, 〈10.7717/peerj-cs.50〉
Liste complète des métadonnées
Contributeur : Florent Breuil <>
Soumis le : mardi 2 août 2016 - 15:30:19
Dernière modification le : mardi 23 octobre 2018 - 14:36:09

Lien texte intégral



Nicolas Durrande, James Hensman, Magnus Rattray, Neil D. Lawrence. Detecting periodicities with Gaussian processes. PeerJ Computer Science, PeerJ, 2016, 2, 〈10.7717/peerj-cs.50〉. 〈emse-01351044〉



Consultations de la notice