Attribute selection

The work presented in this manuscript is in the field of pattern recognition. Two main contributions are made in this manuscript. Firstly, the reduction of dimensionality which allows us to find relevant structures of lower dimensionality, hidden within the observations we have. Second, the application of hybrid metaheuristic methods to solve optimization problems. In our work, we have been interested in evolutionary bio-inspired methods such as genetic algorithms and artificial immune systems, and then in methods derived from swarm intelligence, which is an area of collective intelligence in its own right. The work described in this manuscript gives rise to applications mainly inspired by the collective behaviour of particles, ants, bees and fireflies for the selection of characteristics. Next, several hybridizations of previously used bio-inspired methods for attribute selection were proposed to reduce the number of traits and improve classification rates.