Modelo neuro-evolutivo de coordenação adaptativa em ambientes dinâmicos

Em ambientes dinâmicos e complexos, a política ótima de coordenação não pode ser derivada analiticamente, mas, deve ser aprendida através da interação direta com o ambiente. Geralmente, utiliza-se aprendizado por reforço para prover coordenação em tais ambientes. Atualmente, neuro-evolução é um dos métodos de aprendizado por reforço mais proeminentes. Em vista disto, neste trabalho, é proposto um modelo de coordenação baseado em neuro-evolução. Mais detalhadamente, desenvolveu-se uma extensão do método neuro-evolutivo conhecido como Enforced Subpopulations (ESP). Na extensão desenvolvida, a rede neural que define o comportamento de cada agente é totalmente conectada. Adicionalmente, é permitido que o algoritmo encontre, em tempo de treinamento, a quantidade de neurônios que deve estar presente na camada oculta da rede neural de cada agente. Esta alteração, além de oferecer flexibilidade na definição da topologia da rede de cada agente e diminuir o tempo necessário para treinamento, permite também a constituição de grupos de agentes heterogêneos. Um ambiente de simulação foi desenvolvido e uma série de experimentos realizados com o objetivo de avaliar o modelo proposto e identificar quais os melhores valores para os diversos parâmetros do modelo. O modelo proposto foi aplicado no domínio das tarefas de perseguição-evasão.

Resultantes, equações polinomiais e o teorema de Bezout

A presente dissertação aborda uma técnica para determinar as soluções de sistemas de equações polinomiais. Esta técnica que é puramente algébrica, interliga tópicos da Matemática, como a Geometria Algébrica e a Álgebra Computacional. Mais especificamente, estudamos a teoria de Resultantes e suas aplicações. Começamos com a motivação de encontrar as raízes comuns de dois polinômios a uma variável, em seguida é estendida para o caso mais geral de várias variáveis. Estudamos detalhadamente como obter fórmulas para o cálculo do Resultante, como por exemplo a fórmula de Macaulay e de Poisson. A técnica para resolver sistemas de equações polinomiais é então apresentada. Terminamos apresentando uma prova de um caso particular do Teorema de Bezout, como aplicação da teoria de Resultantes. Este teorema é muito importante, pois fornece um número de soluções de um sistema de equações polinomiais.

Classificação de imagens de ambientes coralinos: uma abordagem empregando uma combinação de classificadores e máquina de vetor de suporte

The use of the maps obtained from remote sensing orbital images submitted to digital processing became fundamental to optimize conservation and monitoring actions of the coral reefs. However, the accuracy reached in the mapping of submerged areas is limited by variation of the water column that degrades the signal received by the orbital sensor and introduces errors in the final result of the classification. The limited capacity of the traditional methods based on conventional statistical techniques to solve the problems related to the inter-classes took the search of alternative strategies in the area of the Computational Intelligence. In this work an ensemble classifiers was built based on the combination of Support Vector Machines and Minimum Distance Classifier with the objective of classifying remotely sensed images of coral reefs ecosystem. The system is composed by three stages, through which the progressive refinement of the classification process happens. The patterns that received an ambiguous classification in a certain stage of the process were revalued in the subsequent stage. The prediction non ambiguous for all the data happened through the reduction or elimination of the false positive. The images were classified into five bottom-types: deep water; under-water corals; inter-tidal corals; algal and sandy bottom. The highest overall accuracy (89%) was obtained from SVM with polynomial kernel. The accuracy of the classified image was compared through the use of error matrix to the results obtained by the application of other classification methods based on a single classifier (neural network and the k-means algorithm). In the final, the comparison of results achieved demonstrated the potential of the ensemble classifiers as a tool of classification of images from submerged areas subject to the noise caused by atmospheric effects and the water column

Método Fuzzy para auxílio ao diagnóstico de câncer de mama em ambiente inteligente de telediagnóstico colaborativo para apoio à tomada de decisão

Breast cancer, despite being one of the leading causes of death among women worldwide is a disease that can be cured if diagnosed early. One of the main techniques used in the detection of breast cancer is the Fine Needle Aspirate FNA (aspiration puncture by thin needle) which, depending on the clinical case, requires the analysis of several medical specialists for the diagnosis development. However, such diagnosis and second opinions have been hampered by geographical dispersion of physicians and/or the difficulty in reconciling time to undertake work together. Within this reality, this PhD thesis uses computational intelligence in medical decision-making support for remote diagnosis. For that purpose, it presents a fuzzy method to assist the diagnosis of breast cancer, able to process and sort data extracted from breast tissue obtained by FNA. This method is integrated into a virtual environment for collaborative remote diagnosis, whose model was developed providing for the incorporation of prerequisite Modules for Pre Diagnosis to support medical decision. On the fuzzy Method Development, the process of knowledge acquisition was carried out by extraction and analysis of numerical data in gold standard data base and by interviews and discussions with medical experts. The method has been tested and validated with real cases and, according to the sensitivity and specificity achieved (correct diagnosis of tumors, malignant and benign respectively), the results obtained were satisfactory, considering the opinions of doctors and the quality standards for diagnosis of breast cancer and comparing them with other studies involving breast cancer diagnosis by FNA.

Técnicas neuronais de agrupamentos aplicadas à detecção de anomalias em mamogramas digitais

This work proposes the development of an intelligent system for analysis of digital mammograms, capable to detect and to classify masses and microcalcifications. The digital mammograms will be pre-processed through techniques of digital processing of images with the purpose of adapting the image to the detection system and automatic classification of the existent calcifications in the suckles. The model adopted for the detection and classification of the mammograms uses the neural network of Kohonen by the algorithm Self Organization Map - SOM. The algorithm of Vector quantization, Kmeans it is also used with the same purpose of the SOM. An analysis of the performance of the two algorithms in the automatic classification of digital mammograms is developed. The developed system will aid the radiologist in the diagnosis and accompaniment of the development of abnormalities

Especificação de um bloco funcional implementando rede neural para redes industriais foundations fieldbus

This work proposes the specification of a new function block according to Foundation Fieldbus standards. The new block implements an artificial neural network, which may be useful in process control applications. The specification includes the definition of a main algorithm, that implements a neural network, as well as the description of some accessory functions, which provide safety characteristics to the block operation. Besides, it also describes the block attributes emphasizing its parameters, which constitute the block interfaces. Some experimental results, obtained from an artificial neural network implementation using actual standard functional blocks on a laboratorial FF network, are also shown, in order to demonstrate the possibility and also the convenience of integrating a neural network to Fieldbus devices

Separação cega de fontes lineares e não lineares usando algoritmo genético, redes neurais artificiais RBF e negentropia de Rényi como medida de independência

Conventional methods to solve the problem of blind source separation nonlinear, in general, using series of restrictions to obtain the solution, often leading to an imperfect separation of the original sources and high computational cost. In this paper, we propose an alternative measure of independence based on information theory and uses the tools of artificial intelligence to solve problems of blind source separation linear and nonlinear later. In the linear model applies genetic algorithms and Rényi of negentropy as a measure of independence to find a separation matrix from linear mixtures of signals using linear form of waves, audio and images. A comparison with two types of algorithms for Independent Component Analysis widespread in the literature. Subsequently, we use the same measure of independence, as the cost function in the genetic algorithm to recover source signals were mixed by nonlinear functions from an artificial neural network of radial base type. Genetic algorithms are powerful tools for global search, and therefore well suited for use in problems of blind source separation. Tests and analysis are through computer simulations

Análise de um controlador baseado no Jacobiano estimado da planta através de uma rede neural

This work presents an analysis of the control law based on an indirect hybrid scheme using neural network, initially proposed for O. Adetona, S. Sathanathan and L. H. Keel. Implementations of this control law, for a level plant of second order, was resulted an oscillatory behavior, even if the neural identifier has converged. Such results had motivated the investigation of the applicability of that law. Starting from that, had been made stability mathematical analysis and several implementations, with simulated plants and with real plants, for analyze the problem. The analysis has been showed the law was designed being despised some components of dynamic of the plant to be controlled. Thus, for plants that these components have a significant influence in its dynamic, the law tends to fail

Medição da densidade óssea através de ondas eletromagnéticas e classificação por MVS

This work holds the purpose of presenting an auxiliary way of bone density measurement through the attenuation of electromagnetic waves. In order to do so, an arrangement of two microstrip antennas with rectangular configuration has been used, operating in a frequency of 2,49 GHz, and fed by a microstrip line on a substrate of fiberglass with permissiveness of 4.4 and height of 0,9 cm. Simulations were done with silica, bone meal, silica and gypsum blocks samples to prove the variation on the attenuation level of different combinations. Because of their good reproduction of the human beings anomaly aspects, samples of bovine bone were used. They were subjected to weighing, measurement and microwave radiation. The samples had their masses altered after mischaracterization and the process was repeated. The obtained data were inserted in a neural network and its training was proceeded with the best results gathered by correct classification on 100% of the samples. It comes to the conclusion that through only one non-ionizing wave in the 2,49 GHz zone it is possible to evaluate the attenuation level in the bone tissue, and that with the appliance of neural network fed with obtained characteristics in the experiment it is possible to classify a sample as having low or high bone density

Cálculos AB initio de propriedades estruturais, eletrônicas e ópticas de colmpostos CaXo3 (X=Si, Ge, Sn)

In this work we present a study of structural, electronic and optical properties, at ambient conditions, of CaSiO3, CaGeO3 and CaSnO3 crystals, all of them a member of Ca-perovskite class. To each one, we have performed density functional theory ab initio calculations within LDA and GGA approximations of the structural parameters, geometry optimization, unit cell volume, density, angles and interatomic length, band structure, carriers effective masses, total and partial density of states, dielectric function, refractive index, optical absorption, reflectivity, optical conductivity and loss function. A result comparative procedure was done between LDA and GGA calculations, a exception to CaSiO3 where only LDA calculation was performed, due high computational cost that its low symmetry crystalline structure imposed. The Ca-perovskite bibliography have shown the absence of electronic structure calculations about this materials, justifying the present work

JFloat: uma biblioteca de ponto flutuante para a linguagem Java com suporte a arredondamento direcionado

This work presents JFLoat, a software implementation of IEEE-754 standard for binary floating point arithmetic. JFloat was built to provide some features not implemented in Java, specifically directed rounding support. That feature is important for Java-XSC, a project developed in this Department. Also, Java programs should have same portability when using floating point operations, mainly because IEEE-754 specifies that programs should have exactly same behavior on every configuration. However, it was noted that programs using Java native floating point types may be machine and operating system dependent. Also, JFloat is a possible solution to that problem

Aplicação de redes neurais na estimação da temperatura interna de transformadores de distribuição imersos em óleo

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Previsão de cargas elétricas através de uma Rede Neural Híbrida Back-ART Fuzzy

Pós-graduação em Engenharia Elétrica - FEIS

Reconhecimento de padrões lexicais por meio de redes neurais

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)