Estratégias evolutivas para o ajuste de parâmetros de um modelo epidemiológico baseado em autômatos celulares probabilísticos

• Ano de defesa: 2021
• Autor(a) principal: Fraga, Larissa Maiara
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Ciência da Computação
• Palavras-chave em Português: Algoritmos Genéticos; Autômatos Celulares; Evolução Multi-estágio; Computação Evolutiva; Modelagem de sistemas dinâmicos; Ajuste de Parâmetros; Doença de Chagas; Genetic Algorithms; Cellular Automata; Multi-stage evaluation; Evolutionary Computing; Dynamic systems modeling; Parameter adjustment; Chagas Disease; Computação; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::MATEMATICA DA COMPUTACAO::MODELOS ANALITICOS E DE SIMULACAO; Algorítmos genéticos; Automato celular; Computação evolutiva; Chagas, Doença de
• Link de acesso: https://repositorio.ufu.br/handle/123456789/36370; http://doi.org/10.14393/ufu.di.2022.5042

Reliable models allow the simulation of critical processes that can serve as a basis for planning and defining public policies. Dynamic systems modeling is an important research tool that can predict and assess the impact of decisions taken by organizations and governments. Cellular automata have been used as an alternative for these types of modeling, as they are dynamic, discrete systems capable of describing complex phenomena from a set of simple rules and local iterations. Once the basic representation of the system is defined, one of the main difficulties in modeling is the adjustment of the various parameters that compose it. Since the evolutionary algorithms have been shown to be a powerful adaptive search technique, in this work we investigate two strategies based on genetic algorithms (GAs) to adjust parameters of a model, aiming to approximate the simulations results to the behavior found in the data to be modeled. The first strategy consists of applying a typical genetic algorithm to solve the problem, while the second adopts multiple stages of evolution, where each stage is responsible for adjusting a subset of the parameters. Both strategies adopt as a case study a model that describes the evolution of a population of insect vectors responsible for Chagas disease (SLIMI et al., 2009). This model is based on probabilistic cellular automata and was originally proposed to reproduce real data collected in a village in Mexico. Here, it is used to generate the dataset used as reference by the genetic algorithm. Experimental results show that the parameters defined in both evolutionary strategies reproduce a behavior similar to the reference model with respect of the number of insects and their spatial distribution.