Mathematical Methods of Modern Statistics 2 / Méthodes mathématiques en statistiques modernes 2

Collection Mathematical Methods of Modern Statistics 2 / Méthodes mathématiques en statistiques modernes 2

Organisateur(s) Bogdan, Malgorzata ; Graczyk, Piotr ; Panloup, Fabien ; Proïa, Frédéric ; Roquain, Etienne
Date(s) 15/06/2020 - 19/06/2020
URL associée https://www.cirm-math.com/cirm-virtual-event-2146.html
00:00:00 / 00:00:00
15 25

High-dimensional, multiscale online changepoint detection

De Richard Samworth

We introduce a new method for high-dimensional, online changepoint detection in settings where a p-variate Gaussian data stream may undergo a change in mean. The procedure works by performing likelihood ratio tests against simple alternatives of different scales in each coordinate, and then aggregating test statistics across scales and coordinates. The algorithm is online in the sense that its worst-case computational complexity per new observation, namely O(p2log(ep)), is independent of the number of previous observations; in practice, it may even be significantly faster than this. We prove that the patience, or average run length under the null, of our procedure is at least at the desired nominal level, and provide guarantees on its response delay under the alternative that depend on the sparsity of the vector of mean change. Simulations confirm the practical effectiveness of our proposal.

Informations sur la vidéo

Données de citation

  • DOI 10.24350/CIRM.V.19643203
  • Citer cette vidéo Samworth, Richard (04/06/2020). High-dimensional, multiscale online changepoint detection. CIRM. Audiovisual resource. DOI: 10.24350/CIRM.V.19643203
  • URL https://dx.doi.org/10.24350/CIRM.V.19643203

Bibliographie

  • CHEN, Yudong, WANG, Tengyao, et SAMWORTH, Richard J. High-dimensional, multiscale online changepoint detection. arXiv preprint arXiv:2003.03668, 2020. - https://arxiv.org/abs/2003.03668

Dernières questions liées sur MathOverflow

Pour poser une question, votre compte Carmin.tv doit être connecté à mathoverflow

Poser une question sur MathOverflow




Inscrivez-vous

  • Mettez des vidéos en favori
  • Ajoutez des vidéos à regarder plus tard &
    conservez votre historique de consultation
  • Commentez avec la communauté
    scientifique
  • Recevez des notifications de mise à jour
    de vos sujets favoris
Donner son avis