Full text
| |
| Author(s): |
Fabio Augusto Faria
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2010-03-19 |
| Examining board members: |
Ricardo da Silva Torres;
Adriano Alonso Veloso;
Anderson de Rezende Rocha
|
| Advisor: | Ricardo da Silva Torres |
| Abstract | |
Learning techniques have been used in several applications (medicine, biology, surveillance systems, e.g.) This work aims to evaluate the use of the Genetic Programming (GP) learning technique for image retrieval and classification tasks. This technique is a problem-solving system that follows principles of inheritance and evolution, inspired by the idea of Natural Selection. The space of all possible solutions is investigated using a set of optimization techniques that imitate the theory of evolution. The main contributions of this work are: proposal of classifier implementation using GP to combine visual evidences (image descriptors) to be used in image classification tasks; comparison of GP with other learning techniques in content-based image retrieval tasks (AU) | |