Saeid Soheily-Khah - Generalized k-means clustering for temporal data

14:00
Vendredi
7
Oct
2016
Organisé par : 
Saeid Soheily-Khah
Intervenant : 
Saeid Soheily-Khah
Équipes : 
Information détaillée : 

 

Membres du jury

  • M. Mohamed NADIF,  professeur à l’Université de Paris Descartes, rapporteur
  • M. Paul HONEINE,  professeur à l’Université de Rouen Normandie, rapporteur
  • M. Pierre-François MARTEAU,  professeur à l’Université Bretagne Sud, examinateur
  • M. Jose Antonio VILAR FERNANDEZ, maître de conférences à l’Universidade da Coruña, examinateur
  • Mme. Ahlame DOUZAL,  maître de conférences (HDR) à l’Université Grenoble Alpes, directeur de thèse
  • M. Eric GAUSSIER,  professeur à l’Université Grenoble Alpes, co-directeur de thèse

 

Résumé : 

Temporal data naturally arise in various emerging applications, such as sensor networks, human mobility or internet of things. Clustering is an important task, usually applied a priori to pattern analysis tasks, for summarization, group and prototype extraction; it is all the more crucial for dimensionality reduction in a big data context. Clustering temporal data under time warp measures is challenging because it requires aligning multiple temporal data simultaneously. To circumvent this problem, costly k-medoids and kernel k-means algorithms are generally used. This work investigates a different approach to temporal data clustering through weighted and kernel time warp measures and a tractable and fast estimation of the representative of the clusters that captures both global and local temporal features. A wide range of public and challenging datasets, encompassing images, traces and ecg data that are non-isotropic (i.e., non-spherical), not well-isolated and linearly non-!
separable, is used to evaluate the efficiency of the proposed temporal data clustering. The results of this comparison illustrate the benefits of the method proposed, which outperforms the baselines on all datasets. A deep analysis is conducted to study the impact of the data specifications on the effectiveness of the studied clustering methods.