Modelos descriptivos basados en aprendizaje supervisado para el tratamiento de big data y flujos continuos de datos

Supervised by:
  1. Cristóbal José Carmona del Jesús Director
  2. Pedro González García Co-director

Defence university: Universidad de Jaén

Fecha de defensa: 28 April 2020

  1. Francisco Herrera Triguero Chair
  2. Antonio Jesús Rivera Rivas Secretary
  3. Isaac Triguero Velázquez Committee member

Type: Thesis

Teseo: 647474 DIALNET


In this thesis the subgroup discovery and emerging pattern mining tasks for the resolution of complex problems, such as big data and data stream mining, among others, are analysed in depth. Different methods and tools are proposed in order to extract descriptive knowledge from these types of environments. In addition, different open problems in this area are highlighted. In particular, for subgroup discovery an analysis of the influence of data noise on the main evolutionary fuzzy systems developed is presented; a software package for the R platform with the main algorithms based on evolutionary fuzzy systems is proposed; and an initial analysis of the behaviour of the main approaches adapted to multi-instance problems, a complex problem on the rise, is shown. With respect to emerging pattern mining, a review of the main approaches developed in the task from a descriptive point of view is presented, together with three developments based on evolutionary fuzzy systems: one focused on improving the quality of the extracted knowledge from a descriptive point of view; another focused on performing this extraction in the big data domain and a last method focused on the context of data stream mining.