Ph.D. Student: Amelia Zafra
Advisors: Sebastián Ventura
Defended on: October 2009
Keywords: multiple instance learning, genetic programming
Digital version 


This work focuses on the design of grammatical genetic programming models for solving different paradigm of learning applications with multiple instances.

First, we review the status of art of this learning. Following this review, we find that almost all learning paradigms used in machine learning have been extended to this paradigm, but there are no proposals of Evolutionary Algorithms (EAs) in this learning framework. EAs are a good alternative in different learning paradigms which have been applied, the large number of publications appeared since its appearance is an evidence of this popularity. In this work grammatical genetic programming methods both mono-and multi-objective are introduced for the resolution of different applications. In first place, an experimental study using benchmark data sets is carried out to demonstrate their effectiveness with respect to the most relevant proposals done over the years. Then, the models are applied over two real problems: web index page recommendation and prediction of a student’s academic performance considering the work developed in the educational platform; these problems approached from a traditional supervised learning contain many missing values making difficult the correct classification. Using MIL, we seek a more flexible representation to solve them.


The development of this thesis was supported by:

  • Spanish Ministry of Science and Technology, projects TIN2008-06681-C05-02 and TIN2008-06681-C06-03.
  • Regional Government of Andalusia, project P08-TIC-3720.
  • Spanish Ministry of Education under the FPU program (AP2005-01746)


  1. A. Zafra, C. Romero, S. Ventura and E. Herrera-Viedma. Multi-instance genetic programming for web index recommendation. Expert Syst. Appl., vol. 36(9), pp. 11470-11479. 2009.
  2. A. Zafra and S. Ventura. G3P-MI: A genetic programming algorithm for multiple instance learning. Information Sciences, vol. 180(23), pp. 4496-4513. 2010.
  1. A. Zafra and S. Ventura. Multi-objective Genetic Programming for Multiple Instance Learning. ECML 2007, pp. 790-797. 2007.
  2. A. Zafra, S. Ventura, E. Herrera-Viedma, C. Romero. Multiple Instance Learning with Genetic Programming for Web Mining.IWANN 2007, pp. 919-927. 2007.
  3. A. Zafra, E.L. Gibaja and S. Ventura. Multiple Instance Learning with MultiObjective Genetic Programming for Web Mining.HIS 2008, pp. 513-518. 2008.
  4. A. Zafra and S. Ventura. Predicting Student Grades in Learning Management Systems with Multiple Instance Learning Genetic Programming. EDM 2009, pp. 309-318. 2009.
  5. A. Zafra and S. Ventura. Comparison of Multi-objective Grammar-Guided Genetic Programming Methods to Multiple Instance Learning. ISDA 2009, pp. 1120-1125. 2009.
  6. A. Zafra, C. Romero and S. Ventura. Predicting Academic Achievement Using Multiple Instance Genetic Programming.HAIS 2009, pp. 450-458. 2009.
  7. A. Zafra. Multi-Instance Learning with MultiObjective Genetic Programming. Encyclopedia of Data Warehousing and Mining 2009, pp. 1372-1379. 2009.