Header logo is


4 results (BibTeX)

2016


no image
Annales des Concours 2016 MP Mathématiques, Informatique

Batog, G., Dumont, J., Puyhaubert, V.

In corrigés des problèmes posés aux concours CCP, Centrale/Supélec, Mines/Ponts, X/ENS, 2016 (inbook)

slt

H&K Éditions [BibTex]

2016


H&K Éditions [BibTex]


no image
Consistent change-point detection with kernels

Garreau, D., Arlot, S.

December 2016, (Submitted on 14 Dec 2016 (v1), last revised 29 Jun 2017 (this version, v3)) (article)

Abstract
In this paper we study the kernel change-point algorithm (KCP) proposed by Arlot, Celisse and Harchaoui (2012), which aims at locating an unknown number of change-points in the distribution of a sequence of independent data taking values in an arbitrary set. The change-points are selected by model selection with a penalized kernel empirical criterion. We provide a non-asymptotic result showing that, with high probability, the KCP procedure retrieves the correct number of change-points, provided that the constant in the penalty is well-chosen; in addition, KCP estimates the change-points location at the optimal rate. As a consequence, when using a characteristic kernel, KCP detects all kinds of change in the distribution (not only changes in the mean or the variance), and it is able to do so for complex structured data (not necessarily in ℝd). Most of the analysis is conducted assuming that the kernel is bounded; part of the results can be extended when we only assume a finite second-order moment.

slt

arXiv:1612.04740 [BibTex]

arXiv:1612.04740 [BibTex]

2015


no image
Annales des Concours 2015 PC Mathématiques, Informatique

Batog, G., Dumont, J., Puyhaubert, V.

In corrigés des problèmes posés aux concours CCP, Centrale/Supélec, Mines/Ponts, X/ENS, 2015 (inbook)

slt

H&K Éditions [BibTex]

2015


H&K Éditions [BibTex]

2014


no image
Metric learning for temporal sequence alignment

Garreau, D., Lajugie, R., Arlot, S., Bach, F.

In Advances in Neural Information Processing Systems, pages: 1817-1825, 2014 (inproceedings)

Abstract
In this paper, we propose to learn a Mahalanobis distance to perform alignment of multivariate time series. The learning examples for this task are time series for which the true alignment is known. We cast the alignment problem as a structured prediction task, and propose realistic losses between alignments for which the optimization is tractable. We provide experiments on real data in the audio-toaudio context, where we show that the learning of a similarity measure leads to improvements in the performance of the alignment task. We also propose to use this metric learning framework to perform feature selection and, from basic audio features, build a combination of these with better alignment performance.

slt

Paper [BibTex]

2014


Paper [BibTex]