Improving search in geometric semantic genetic programming

• Ano de defesa: 2016
• Autor(a) principal: Luiz Otavio Vilas Boas Oliveira
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Minas Gerais
• Palavras-chave em Português: Programação genética (Computação); Computação; Operadores semânticos geométricos; Operadores de dispersão; seleção de instâncias; Programação genética semântica
• Link de acesso: https://hdl.handle.net/1843/ESBF-AFWG3P

Making Genetic Programming methods semantic-aware has been the focus of many works in the past years. Among these methods, Geometric Semantic GP (GSGP) acts on the syntax of the parent programs producing offspring respecting a semantic criterion. In this thesis we focus on the open issues of GSGP. We investigate the impact of the geometric semantic crossover with different distance functions and the possibility of optimally adjusting its coefficients. We also present the Sequential Symbolic Regression, an attempt to control the exponential growth of the individuals caused by the use of this operator. In addition, we propose a Geometric Dispersion framework to construct operators that move individuals to less dense areas of the search space. Last, we present a study of the impact of selecting training instances in order to reduce the semantic space dimensionality. All methods proposed showed GSGP search can be improved by adding simple and effective mechanisms to its current operators