This research identifies and investigates major issues in inducing accurate and comprehensible fuzzy rules from datasets. A review of the current literature on fuzzy rulebase induction uncovers two significant issues: A. There is a tradeoff between inducing accurate fuzzy rules and inducing comprehensible fuzzy rules; and, B. A common strategy for the induction of fuzzy rulebases, that of iterative rule learning where the rules are generated one by one and independently of each other, may not be an optimal one. FRANTIC, a system that provides a framework for exploring the claims above is developed. At the core lies a mechanism for creating individual fuzzy rules. This is based on a significantly modified social insect-inspired heuristic for combinatorial optimisation -- Ant Colony Optimisation. The rule discovery mechanism is utilised in two very different strategies for the induction of a complete fuzzy rulebase: 1. The first follows the common iterative rule learning approach for the induction of crisp and fuzzy rules; 2. The second has been designed during this research explicitly for the induction of a fuzzy rulebase, and generates all rules in parallel. Both strategies have been tested on a number of classification problems, including medical diagnosis and industrial plant fault detection, and compared against other crisp or fuzzy induction algorithms that use more well-established approaches. The results challenge statement A above, by presenting evidence to show that one criterion need not be met at the expense of the other. This research also uncovers the cost that is paid -- that of computational expenditure -- and makes concrete suggestions on how this may be resolved. With regards to statement B, until now little or no evidence has been put forward to support or disprove the claim. The results of this research indicate that definite advantages are offered by the second simultaneous strategy, that are not offered by the iterative one. These benefits include improved accuracy over a wide range of values for several key system parameters. However, both approaches also fare well when compared to other learning algorithms. This latter fact is due to the rule discovery mechanism itself -- the adapted Ant Colony Optimisation algorithm -- which affords several additional advantages. These include a simple mechanism within the rule construction process that enables it to cope with datasets that have an imbalanced distribution between the classes, and another for controlling the amount of fit to the training data. In addition, several system parameters have been designed to be semi-autonomous so as to avoid unnecessary user intervention, and in future work the social insect metaphor may be exploited and extended further to enable it to deal with industrial-strength data mining issues involving large volumes of data, and distributed and/or heterogeneous databases.

Made available in DSpace on 2016-06-02T19:05:54Z (GMT). No. of bitstreams: 1 3998.pdf: 3486824 bytes, checksum: f1c040adfdc7d0672bc93a058f8a413d (MD5) Previous issue date: 2011-06-28 / Financiadora de Estudos e Projetos / The goal of this work is to study, expand and evaluate the use of multiobjective genetic algorithms and the iterative rule learning approach in fuzzy system generation, especially, in fuzzy rule-based systems, both in automatic fuzzy rule generation from datasets and in fuzzy sets optimization. This work investigates the use of multi-objective genetic algorithms with a focus on the trade-off between accuracy and interpretability, considered contradictory objectives in the representation of fuzzy systems. With this purpose, we propose and implement an evolutive multi-objective genetic model composed of three stages. In the first stage uniformly distributed fuzzy sets are created. In the second stage, the rule base is generated by using an iterative rule learning approach and a multiobjective genetic algorithm. Finally the fuzzy sets created in the first stage are optimized through a multi-objective genetic algorithm. The proposed model was evaluated with a number of benchmark datasets and the results were compared to three other methods found in the literature. The results obtained with the optimization of the fuzzy sets were compared to the result of another fuzzy set optimizer found in the literature. Statistical comparison methods usually applied in similar context show that the proposed method has an improved classification rate and interpretability in comparison with the other methods. / O objetivo deste trabalho é estudar, expandir e avaliar o uso dos algoritmos genéticos multiobjetivo e a abordagem iterativa na geração de sistemas fuzzy, mais especificamente para sistemas fuzzy baseados em regras, tanto na geração automática da base de regras fuzzy a partir de conjuntos de dados, como a otimização dos conjuntos fuzzy. Esse trabalho investiga o uso dos algoritmos genéticos multiobjetivo com enfoque na questão de balanceamento entre precisão e interpretabilidade, ambos considerados contraditórios entre si na representação de sistemas fuzzy. Com este intuito, é proposto e implementado um modelo evolutivo multiobjetivo genético composto por três etapas. Na primeira etapa são criados os conjuntos fuzzy uniformemente distribuídos. Na segunda etapa é tratada a geração da base de regras usando a abordagem iterativa e um algoritmo genético multiobjetivo. Por fim, na terceira etapa os conjuntos fuzzy criados na primeira etapa são otimizados mediante um algoritmo genético multiobjetivo. O modelo desenvolvido foi avaliado em diversos conjuntos de dados benchmark e os resultados obtidos foram comparados com outros três métodos, que geram regras de classificação, encontrados na literatura. Os resultados obtidos após a otimização dos conjuntos fuzzy foram comparados com resultados de outro otimizador de conjuntos fuzzy encontrado na literatura. Métodos estatísticos de comparação usualmente aplicados em contextos semelhantes mostram uma melhor taxa de classificação e interpretabilidade do método proposto com relação a outros métodos.