Inductive Conformal Martingales for Change-Point Detection

Denis Volkhonskiy, Evgeny Burnaev, Ilia Nouretdinov, Alexander Gammerman, Vladimir Vovk
Proceedings of the Sixth Workshop on Conformal and Probabilistic Prediction and Applications, PMLR 60:132-153, 2017.

Abstract

We consider the problem of quickest change-point detection in data streams. Classical change-point detection procedures, such as CUSUM, Shiryaev-Roberts and Posterior Probability statistics, are optimal only if the change-point model is known, which is an unrealistic assumption in typical applied problems. Instead we propose a new method for change-point detection based on Inductive Conformal Martingales, which requires only the independence and identical distribution of observations. We compare the proposed approach to standard methods, as well as to change-point detection oracles, which model a typical practical situation when we have only imprecise (albeit parametric) information about pre- and post-change data distributions. Results of comparison provide evidence that change-point detection based on Inductive Conformal Martingales is an efficient tool, capable to work under quite general conditions unlike traditional approaches.

Cite this Paper

BibTeX
@InProceedings{pmlr-v60-volkhonskiy17a, title = {Inductive Conformal Martingales for Change-Point Detection}, author = {Volkhonskiy, Denis and Burnaev, Evgeny and Nouretdinov, Ilia and Gammerman, Alexander and Vovk, Vladimir}, booktitle = {Proceedings of the Sixth Workshop on Conformal and Probabilistic Prediction and Applications}, pages = {132--153}, year = {2017}, editor = {Gammerman, Alex and Vovk, Vladimir and Luo, Zhiyuan and Papadopoulos, Harris}, volume = {60}, series = {Proceedings of Machine Learning Research}, month = {13--16 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v60/volkhonskiy17a/volkhonskiy17a.pdf}, url = {https://proceedings.mlr.press/v60/volkhonskiy17a.html}, abstract = {We consider the problem of quickest change-point detection in data streams. Classical change-point detection procedures, such as CUSUM, Shiryaev-Roberts and Posterior Probability statistics, are optimal only if the change-point model is known, which is an unrealistic assumption in typical applied problems. Instead we propose a new method for change-point detection based on Inductive Conformal Martingales, which requires only the independence and identical distribution of observations. We compare the proposed approach to standard methods, as well as to change-point detection oracles, which model a typical practical situation when we have only imprecise (albeit parametric) information about pre- and post-change data distributions. Results of comparison provide evidence that change-point detection based on Inductive Conformal Martingales is an efficient tool, capable to work under quite general conditions unlike traditional approaches.} }
Endnote
%0 Conference Paper %T Inductive Conformal Martingales for Change-Point Detection %A Denis Volkhonskiy %A Evgeny Burnaev %A Ilia Nouretdinov %A Alexander Gammerman %A Vladimir Vovk %B Proceedings of the Sixth Workshop on Conformal and Probabilistic Prediction and Applications %C Proceedings of Machine Learning Research %D 2017 %E Alex Gammerman %E Vladimir Vovk %E Zhiyuan Luo %E Harris Papadopoulos %F pmlr-v60-volkhonskiy17a %I PMLR %P 132--153 %U https://proceedings.mlr.press/v60/volkhonskiy17a.html %V 60 %X We consider the problem of quickest change-point detection in data streams. Classical change-point detection procedures, such as CUSUM, Shiryaev-Roberts and Posterior Probability statistics, are optimal only if the change-point model is known, which is an unrealistic assumption in typical applied problems. Instead we propose a new method for change-point detection based on Inductive Conformal Martingales, which requires only the independence and identical distribution of observations. We compare the proposed approach to standard methods, as well as to change-point detection oracles, which model a typical practical situation when we have only imprecise (albeit parametric) information about pre- and post-change data distributions. Results of comparison provide evidence that change-point detection based on Inductive Conformal Martingales is an efficient tool, capable to work under quite general conditions unlike traditional approaches.
APA
Volkhonskiy, D., Burnaev, E., Nouretdinov, I., Gammerman, A. & Vovk, V.. (2017). Inductive Conformal Martingales for Change-Point Detection. Proceedings of the Sixth Workshop on Conformal and Probabilistic Prediction and Applications, in Proceedings of Machine Learning Research 60:132-153 Available from https://proceedings.mlr.press/v60/volkhonskiy17a.html.

Related Material