Thi Phuong Thao Tran - Interpretable time series kernel analytics by pre-image estimation

Vendredi 18 Septembre 2020 - 14:00
Organisé par : 
Thi Phuong Thao Tran
Intervenant : 
Thi Phuong Thao Tran
Équipes : 

 

Jury committee:

  • Mohamed Nadif, Professor, Paris Descartes University, Reviewer.
  • Christophe Marsala, Professor, Sorbonne University, Reviewer.
  • Patrick Gallinari, Professor, Sorbonne University, Examiner.
  • Sihem Amer-Yahia, Research director CNRS, Grenoble Alpes University, Examiner.
  • Ahlame Douzal, Assistant Professor,  Grenoble Alpes University, Thesis director.
  • Paul Honeine, Professor, University of Rouen Normandie, Thesis co-director.

 

Kernel methods are known to be effective to analyse complex objects by implicitly embedding them into some feature space. To interpret and analyse the obtained results, it is often required to restore in the input space the results obtained in the feature space by using pre-image estimation methods. This work proposes a pre-image estimation method for time series kernel analytics that consists of two steps. In the first step, a time warp function, driven by distance constraints in the feature space, is defined to embed time series in a metric space where analytics can be performed conveniently. In the second step, the time series pre-image estimation is cast as learning a linear (or a nonlinear) transformation that ensures a local isometry between the time series embedding space and the feature space.The proposed method is compared to state of the art through three major tasks that require pre-image estimation: 1) time series averaging, 2) time series reconstruction and denoising, and 3) time series representation learning. The extensive experiments conducted on 33 publicly-available datasets show the benefits of the pre-image estimation for time series kernel analytics.