For about three years, my research at LIMOS hold on the development of new clustering methods for categorical data. Our work has been mainly motivated by two real-world applications of the new methods on Health data. In the first case study in collaboration with institut Analgesia, the goal has been to use the developped methods to determine the profiles of chronic pain patients so that the institute can given the latter a personalised follow-up. In the second case study in collaboration with a research laboratory of the Simon Fraser University in Canada, the goal has been to use the new methods to study the influcence of lifestyle activities on the cognitive health of older adults.

We proposed two clustering methods, one in the framework of the fuzzy sets theory and the second in the framework of belief functions theory in order to model the uncertainty of objects assigments into clusters. Our contributions conducted to three peer-reviewed publications: 2 FUZZ-IEEE and 1 EGC.

JURY:

Examiners:

Pr. Vincent Barra, Université Clermont Auvergne, France

Pr. Ana Belén Ramos-Guajardo, Universidad de Oviedo, Espagne

Pr. Manuel Ojeda Aciego, Universidad de Málaga, Espagne

Reviewers:

Dr. Marie-Jeanne Lesot, Sorbonne Université, France

Dr. Benjamin Quost, Université Technologique de Compiègne, France

Director of thesis

Pr Mephu Nguifo Engelbert, Université Clermont Auvergne, France

Supervisors:

Dr. Antoine Violaine, Université Clermont Auvergne, France

Pr. Moreno Sylvain, Simon Fraser University, Canada.