Explorando o uso de regras de correspondência condicional na busca evolutiva por autômatos celulares classificadores de densidade binária

• Ano de defesa: 2020
• Autor(a) principal: Cardoso, Alberto Luis Libório
• Orientador(a): Oliveira, Pedro Paulo Balbi de
• Banca de defesa: Ruivo , Eurico Luiz Prospero , França, Fabricio Olivetti de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Presbiteriana Mackenzie
• Programa de Pós-Graduação: Engenharia Elétrica
• Departamento: Escola de Engenharia Mackenzie (EE)
• País: Brasil
• Palavras-chave em Português: autômatos celulares; algoritmos genéticos; computação emergente; vida artificial; density classification task
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS
• Link de acesso: https://dspace.mackenzie.br/handle/10899/28603

Cellular automata (CA) are discrete systems, fundamentally based upon local interactions, which, even though simple, may yield complex behaviour or universal computation. A classical problem to probe the computational capacity of CAs is the density classification task, whose objective is to decide the prevailing bit in an arbitrary binary sequence. Here we investigated the efficacy of a recent proposed representation of CA rules would have in that task, given that the new structure of the search space, induced by this new representation, might prove beneficial since new routes on that structure could lead to rules that perform well for such problem. Evolutionary searches using genetic algorithms were employed in different formulations of the density task, even in larger dimensionalities (more states) of the space, led to limited impact on the efficacy of the rules found. The results contrast with those found in the literature, pointing at limitations of the representation scheme employed applied to a density classification problem.