No description available.

No description available.

Este trabalho descreve experimentos realizados com Redes Neurais Artificiais e algoritmos de aprendizado simbólico. Também são investigados dois algoritmos de extração de conhecimento de Redes Neurais Artificiais. Esses experimentos são realizados com três bases de dados com o objetivo de comparar os desempenhos obtidos. As bases de dados utilizadas neste trabalho são: dados de falência de bancos brasileiros, dados do jogo da velha e dados de análise de crédito. São aplicadas sobre os dados três técnicas para melhoria de seus desempenhos. Essas técnicas são: partição pela menor classe, acréscimo de ruído nos exemplos da menor classe e seleção de atributos mais relevantes. Além da análise do desempenho obtido, também é feita uma análise da dificuldade de compreensão do conhecimento extraído por cada método em cada uma das bases de dados. / This work describes experiments carried out witch Artificial Neural Networks and symbolic learning algorithms. Two algorithms for knowledge extraction from Artificial Neural Networks are also investigates. This experiments are performed whit three data set with the objective of compare the performance obtained. The data set used in this work are: Brazilians banks bankruptcy data set, tic-tac-toe data set and credit analysis data set. Three techniques for data set performance improvements are investigates. These techniques are: partition for the smallest class, noise increment in the examples of the smallest class and selection of more important attributes. Besides the analysis of the performance obtained, an analysis of the understanding difficulty of the knowledge extracted by each method in each data bases is made.

It is proposed to continue operating the graphite moderated Advanced Gas-cooled Reactor (AGR) fleet past its design life. Nuclear graphite's properties change in reactor and our limited mechanistic understanding of the relationship between graphite structure, across different lengthscales, and its properties limits our ability to predict its future behaviour. An improved understanding of the relationship between graphite's structural features, the relationship between features across different lengthscales and their effect on material properties would all contribute to a mechanistic understanding of graphite behaviour. Thermal expansion generates thermal strains and stresses in the graphite core during thermal transients, such as during reactor start-up and shut-down. Thermal expansion is a function of graphite crystal thermal expansion, crystallographic preferred orientation and microstructure, although the exact relationship between these is not understood. It is also altered by neutron irradiation. This thesis investigates graphite microstructure, virgin and irradiated, and its crystallographic preferred orientation, specifically as they pertain to thermal expansion. The microstructure of British nuclear graphites PGA and Gilsocarbon, used in the Magnox and AGR fleet respectively, have been investigated using scanning electron microscopy (SEM). Trepanned AGR graphite, that is, graphite drilled from the reactor brick during routine inspection is examined. These samples are from the 2012 Hinkley Point B inspection campaign and are taken from several points through the brick thickness. This provides a 'snap shot' of current AGR graphite condition. Deep trepan samples removed from further into the brick thickness are observed for the first time. Neutron damage was observed in Magnox graphite, irradiated in an inert environment in the material test reactor programme INEEL. The spatial variation in texture of PGA and Gilsocarbon, and the change in such texture after prestress was observed using synchrotron x-ray diffraction. Numerical models were used to identify the required texture change to produce changes in CTE, observed by other authors, during in-situ stress. PGA filler lamellae are arranged in parallel arrays and Gilsocarbon's smaller platelets are arranged in bunched clusters. Severe radiolytic oxidation is observed at all trepan locations, with oxidation decreasing away from the fuel. Radiolytic oxidation occurs at platelet edges. Texture measurements have indicated that PGA graphite exhibits significant spatial variation in texture. Gilsocarbon exhibits less variation but the variation observed is large enough to cause increased thermal stresses. Texture measurements of prestressed graphite have indicated that texture changes also vary spatially. Texture results and SEM observations indicate that spatial variation in texture is caused by spatial variation in microstructure. Changes to the filler particle during prestress may alter local texture. These results indicate there is a link between nuclear graphite's microstructure and its texture. The texture, a function of lamellae or platelet arrangement, determines its thermal expansion. Spatial variations in microstructure formed during manufacturing leads to spatial variations in CTE and possibly other texture sensitive properties, such as dimensional change. Deformation of the lamellae or platelets during stress; thermal creep or irradiation creep is expected to contribute to the observed change in properties associated with these stimuli.

Este trabalho descreve experimentos realizados com Redes Neurais Artificiais e algoritmos de aprendizado simbólico. Também são investigados dois algoritmos de extração de conhecimento de Redes Neurais Artificiais. Esses experimentos são realizados com três bases de dados com o objetivo de comparar os desempenhos obtidos. As bases de dados utilizadas neste trabalho são: dados de falência de bancos brasileiros, dados do jogo da velha e dados de análise de crédito. São aplicadas sobre os dados três técnicas para melhoria de seus desempenhos. Essas técnicas são: partição pela menor classe, acréscimo de ruído nos exemplos da menor classe e seleção de atributos mais relevantes. Além da análise do desempenho obtido, também é feita uma análise da dificuldade de compreensão do conhecimento extraído por cada método em cada uma das bases de dados. / This work describes experiments carried out witch Artificial Neural Networks and symbolic learning algorithms. Two algorithms for knowledge extraction from Artificial Neural Networks are also investigates. This experiments are performed whit three data set with the objective of compare the performance obtained. The data set used in this work are: Brazilians banks bankruptcy data set, tic-tac-toe data set and credit analysis data set. Three techniques for data set performance improvements are investigates. These techniques are: partition for the smallest class, noise increment in the examples of the smallest class and selection of more important attributes. Besides the analysis of the performance obtained, an analysis of the understanding difficulty of the knowledge extracted by each method in each data bases is made.

No description available.