Dissertations / Theses on the topic 'Self-organizing maps of Kohonen'

This thesis aims to cluster financial account time series by extracting global features from the time series and by using two different dimensionality reduction methods, Kohonen Self Organizing Maps and principal component analysis, to cluster the set of the time series by using K-means. The results are then used to further cluster a set of financial services provided by a financial institution, to determine if it is possible to find a set of services which coincide with the time series clusters. The results find several sets of services that are prevalent in the different time series clusters. The resulting method can be used to understand the dynamics between deposits variability and the customers usage of different services and to analyse whether a service is more used in different clusters.
Målet med denna uppsats är att klustra tidsserier över finansiella konton genom att extrahera tidsseriernas karakteristik. För detta används två metoder för att reducera tidsseriernas dimensionalitet, Kohonen Self Organizing Maps och principal komponent analys. Resultatet används sedan för att klustra finansiella tjänster som en kund använder, med syfte att analysera om det existerar ett urval av tjänster som är mer eller mindre förekommande bland olika tidsseriekluster. Resultatet kan användas för att analysera dynamiken mellan kontobehållning och kundens finansiella tjänster, samt om en tjänst är mer förekommande i ett tidsseriekluster.

I here present my version of the Kohonen Self-Organising Map (KSOM) as applied to the classification, or rather clustering, of astronomical data. The main body of this work is concerned with the grouping of period-folded stellar lightcurves and clustering based on the lightcurve shape alone. It has been found that the algorithm is an extremely stable grouping mechanism for data of low (3cr signal to noise) to good quality. With further analysis of the results it is possible to locate underpopulated samples of data that exist within the data. This can be successfully achieved for samples of 1%, or less, total population. Additionally the same algorithm has been applied to the extraction of planetary transit lightcurves from those of eclipsing binaries (chapter 5), and to the grouping of X-ray/optical data from the XMM-Subaru deep-field observations (chapter 6). In both cases the algorithm has shown itself to be quite capable of performing such tasks and as such I propose that it could become a very useful astronomical tool. In summary I also present a few ideas for further refinement of the results presented by the KSOM and how these may be used in future study.

This thesis presents a biologically inspired model of learning and development. This model decomposes the lifetime of a single learning system into a number of stages, analogous to the infant, juvenile, adolescent and adult stages of development in a biological system. This model is then applied to Kohonen's SOM algorithm.In order to better understand the operation of Kohonen's SOM algorithm, a theoretical analysis of self-organisation is performed. This analysis establishes the role played by lateral connections in organisation, and the significance of the Laplacian lateral connections common to many SOM architectures.This analysis of neighbourhood interactions is then used to develop three key variations on Kohonen's SOM algorithm. Firstly, a new scheme for parameter decay, known as Butterworth Step Decay, is presented. This decay scheme provides training times comparable to the best training times possible using traditional linear decay, but precludes the need for a priori knowledge of likely training times. In addition, this decay scheme allows Kohonen's SOM to learn in a continuous manner.Secondly, a method is presented for establishing core knowledge in the fundamental representation of a SOM. This technique is known as Syllabus Presentation. This technique involves using a selected training syllabus to reinforce knowledge known to be significant. A method for developing a training syllabus, known as Percept Masking, is also presented.Thirdly, a method is presented for preventing the loss of trained representations in a continuously learning SOM. This technique, known as Arbor Pruning, involves restricting the weight update process to prevent the loss of significant representations. This technique can be used if the data domain varies within a known set of dimensions. However, it cannot be used to control forgetfulness if dimensions are added to or removed from ++
the data domain.

Orientador: Marcio Luiz de Andrade Netto
Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação
Made available in DSpace on 2018-08-08T10:53:36Z (GMT). No. of bitstreams: 1 Peres_SarajaneMarques_D.pdf: 11492142 bytes, checksum: 7eeb4f81f681e145111b43ccfdde2b7e (MD5) Previous issue date: 2006
Resumo: Redes Neurais Artificiais Auto-Organizáveis (RNA-AO), introduzidas por Teuvo Kohonen na década de 60, constituem uma poderosa ferramenta para análise de dados, mais especificamente para análise de agrupamentos, visualização e aproximação de superfícies. Nesta tese definiu-se uma nova forma para determinar a dimensão topológica do espaço de saídada RNA-AO a partir da análise do conjunto de dados a ser explorado pela rede, realizada como apoio combinado da Teoria de Fractais e do Raciocínio Aproximado Fuzzy. Ao combinar essas duas teorias, concebeu-se uma nova medida de dimensão fractal, a medida de Dimensão Fractal Fuzzy Significativa (DFFS) de um conjunto de dados. Tanto o processo de determinação da DFFS quanto sua aplicação como inferência da dimensãotopológica para a RNA-AO foram validados neste trabalho. O primeiro por meio de sua aplicação ao problema de Tendência a Agrupamentos e o segundo por meio da análise de qualidade das RNAs-AO projetadas segundo tal inferência
Abstract: Self Organizing Maps (SOM), introduced by Teuvo Kohonen during the decade of 1960's, is a powerful tool for data analysis, mainly for clustering analysis and surface approximation. In this thesis, we have defined a new way to determine the output space topological dimension of the SOM using the analysis of the dataset to be explored by the map. This analysis is carried out with the combined support of the Fractal Theory and the Fuzzy Approximated Reasoning, deriving a new fractal dimension measure: the Meaningful Fractal Fuzzy Dimension - DFFS (of the Portuguese "Dimensão Fractal:..Fuzzy Significativa"). The DFFS determination process and its application as an inference to the SOM topological dimension have been both validated in this work. The former has been carried out through its application to the Clustering Tendency Analysis and the latter through the quality analysis of the SOM designed by such inference
Doutorado
Engenharia de Computação
Doutor em Engenharia Elétrica

The Data Mining techniques, based on the Kohonen self-organizing maps have been largely used for classifying signals in several areas of expertise. Generally, the SOM network (Self- Organizing Maps) is used to specify similarity relationships between objects by adopting cluster analysis. The computational cost, data preparation and mathematical modeling can influence the interpretation of results, in which those from the evaluation classes are among its limitations. The Kohonen maps do not permit detailed evaluation of the class of objects, which may however be defined by the class limits, in other words defining a measure that can link when an object belonging to a particular class can migrate from one class to another . To adopt this approach the solutions proposed in this Masters dissertation are designed to implement the Kohonen self-organizing maps and the fuzzy logic to generate neighborhoods between classes aimed at applying these techniques on a two-case study for classifying signals from potencial power systems and Biomedical output signals adopting an interpretation of the Kohonen nebula maps. The work is basically divided into three stages: the first which would be followed by a review of the data-mining techniques and fuzzy logic shown in literature; the second focuses on applying the classifier algorithm using artificial neural networks, specifically the usage of neural networks as SOM data mining techniques to enable the classification of signals while the third step demonstrates the SOM network fuzzy logic multidisciplinary approach as an alternative tool of the data-mining methods.
Fundação de Amparo a Pesquisa do Estado de Alagoas
As técnicas de mineração de dados baseadas nos mapas auto organizáveis de Kohonen têm sido bastante utilizada na classificação de sinais nas mais diversas áreas de conhecimento. Geralmente, a rede SOM (Self-Organizing Maps) é usada para especificar relações de similaridade entre objetos abordando análise de agrupamentos. O custo computacional, a preparação dos dados e modelagem matemática poderá influenciar na interpretação dos resultados, entre suas limitações encontram-se aquelas provenientes da avaliação das classes. Os mapas de Kohonen não permite avaliar de forma detalhada a classe dos objetos, os quais poderão está definidos pelo limite da classe, ou seja, definir uma medida que possa relacionar quando um objeto que pertença a uma classe particular possa migrar de uma classe para outra. Para adotar essa abordagem a solução proposta nesta dissertação de mestrado têm como objetivo aplicar os mapas auto-organizáveis de Kohonen e a lógica nebulosa para gerar as vizinhanças entre as classes visando aplicação dessas técnicas em dois estudos de casos na classificação dos sinais provenientes dos sistemas elétricos de potência e sinais biomédicos adotando uma interpretação nebulosa dos mapas de Kohonen. O trabalho se divide basicamente em três etapas: na primeira, será realizada uma revisão das técnicas de mineração de dados e da lógica nebulosa mostradas na literatura; na segunda, concentra-se aplicar o algoritmo classificador utilizando redes neurais artificiais, especificamente redes neurais SOM como técnica de mineração de dados para efetuar a classificação dos sinais; na terceira etapa demonstramos a abordagem multidisciplinar da rede SOM e da lógica nebulosa como uma ferramenta alternativa aos métodos de mineração de dados.

Dentro do cenário de projeto de circuitos elétricos orientados para o processamento de redes neurais artificiais, este trabalho se concentra no estudo da implementação de Mapas Auto-organizáveis (SOM, do inglês, Self-Organizing Maps) em chips FPGA. A pesquisa aqui realizada busca, fundamentalmente, responder à seguinte pergunta: como devem ser projetadas as arquiteturas computacionais de cada etapa de processamento do SOM para serem adequadamente executadas em FPGA? De forma mais detalhada, o trabalho investiga as possibilidades que diferentes circuitos de computação do SOM oferecem em relação à velocidade de processamento, ao consumo de recursos do FPGA e à consistência com o formalismo teórico que fundamenta esse modelo de rede neural. Tal objetivo de pesquisa é motivado por possibilitar o desenvolvimento de sistemas de processamento neural que exibam as características positivas típicas de implementações diretas em hardware, como o processamento embarcado e a aceleração computacional. CONTRIBUIÇÕES PRINCIPAIS No decorrer da investigação de tais questões, o presente trabalho gerou contribuições com diferentes graus de impacto. A contribuição mais essencial do ponto de vista de estruturação do restante da pesquisa é a fundamentação teórica das propriedades de computação do SOM em hardware. Tal fundamentação é importante pois permitiu a construção dos alicerces necessários para o estudo das diferentes arquiteturas de circuitos exploradas neste trabalho, de forma que estas permanecessem consistentes com as premissas teóricas que certificam o modelo de computação neural estudado. Outra contribuição avaliada como de grande impacto, e que se consolida como um objeto gerado pela pesquisa, é a proposta de um circuito processador para SOM em FPGA que possui o estado-da-arte em velocidade de computação, medido em CUPS (Connections Updated Per Second). Tal processador permite atingir 52,67 GCUPS, durante a fase de treinamento do SOM, um ganho de aproximadamente 100% em relação aos trabalhos publicados na literatura. A aceleração possibilitada pela exploração de processamentos paralelos em FPGA, desenvolvida neste trabalho, é de três a quatro ordens de grandeza em relação a execuções em software do SOM com a mesma configuração. A última contribuição considerada como de grande impacto é a caracterização da execução do SOM em FPGA. Tal avaliação se faz necessária porque os processos de computação dos modelos neurais em hardware, embora semelhantes, não são necessariamente idênticos aos mesmos processos executados em software. Desta forma, a contribuição deste ponto de pesquisa pode ser entendida como a análise do impacto das mudanças implementadas na computação do SOM em FPGA em relação à execução tradicional do algoritmo, feita pela avaliação dos resultados produzidos pela rede neural por medidas de erros topográficos e de quantização. Este trabalho também gerou contribuições consideradas como de médio impacto, que podem ser divididas em dois grupos: aplicações práticas e aportes teóricos. A primeira contribuição de origem prática é a investigação de trabalhos publicados na literatura envolvendo SOM cujas aplicações podem ser viabilizadas por implementações em hardware. Os trabalhos localizados nesse levantamento foram organizados em diferentes categorias, conforme a área de pesquisa - como, por exemplo, Indústria, Robótica e Medicina - e, em geral, eles utilizam o SOM em aplicações que possuem requisitos de velocidade computacional ou embarque do processamento, portanto, a continuidade de seus desenvolvimentos é beneficiada pela execução direta em hardware. As outras duas contribuições de médio impacto de origem prática são as aplicações que serviram como plataforma de teste dos circuitos desenvolvidos para a implementação do SOM. A primeira aplicação pertence à área de telecomunicações e objetiva a identificação de símbolos transmitidos por 16-QAM ou 64-QAM. Estas duas técnicas de modulação são empregadas em diversas aplicações com requisitos de mobilidade - como telefonia celular, TV digital em dispositivos portáteis e Wi-Fi - e o SOM é utilizado para identificar sinais QAM recepcionados com ruídos e distorções. Esta aplicação gerou a publicação de um artigo na revista da Springer, Neural Computing and Applications: Sousa; Pires e Del-Moral-Hernandez (2017). A segunda aplicação pertence à área de processamento de imagem e visa reconhecer ações humanas capturadas por câmeras de vídeo. O processamento autônomo de imagens executado por chips FPGA junto às câmeras de vídeo pode ser empregado em diferentes utilizações, como, por exemplo, sistemas de vigilância automática ou assistência remota em locais públicos. Esta segunda aplicação também é caracterizada por demandar arquiteturas computacionais de alto desempenho. Todas as contribuições teóricas deste trabalho avaliadas como de médio impacto estão relacionadas ao estudo das características de arquiteturas de hardware para computação do modelo SOM. A primeira destas é a proposta de uma função de vizinhança do SOM baseada em FPGA. O objetivo de tal proposta é desenvolver uma expressão computacional para ser executada no chip que constitua uma alternativa eficiente tanto à função gaussiana, tradicionalmente empregada no processo de treinamento do SOM, quanto à função retangular, utilizada de forma rudimentar nas primeiras pesquisas publicadas sobre a implementação do SOM em FPGA. A segunda destas contribuições é a descrição detalhada dos componentes básicos e dos blocos computacionais utilizados nas diferentes etapas de execução do SOM em FPGA. A apresentação dos detalhes da arquitetura de processamento, incluindo seus circuitos internos e a função computada por cada um de seus blocos, permite que trabalhos futuros utilizem os desenvolvimentos realizados nesta pesquisa. Esta descrição detalhada e funcional foi aceita para publicação no IEEE World Congress on Computational Intelligence (WCCI 2018): Sousa et al. (2018). A terceira contribuição teórica de médio impacto é a elaboração de um modelo distribuído de execução do SOM em FPGA sem o uso de uma unidade central de controle. Tal modelo permite a execução das fases de aprendizado e operação da rede neural em hardware de forma distribuída, a qual alcança um comportamento global de auto-organização dos neurônios apenas pela troca local de dados entre elementos de processamento vizinhos. A descrição do modelo distribuído, em conjunto com sua caracterização, está publicada em um artigo no International Joint Conference on Neural Networks do IEEE (IJCNN 2017): Sousa e Del-Moral-Hernandez (2017a). A última contribuição deste grupo de aporte teórico é a comparação entre diferentes modelos de execução do SOM em FPGA. A comparação tem a função de avaliar e contrastar três diferentes possibilidades de implementação do SOM: o modelo distribuído, o modelo centralizado e o modelo híbrido. Os testes realizados e os resultados obtidos estão publicados em um trabalho no International Symposium on Circuits and Systems do IEEE (ISCAS 2017): Sousa e Del-Moral-Hernandez (2017b). Finalmente, apresentam-se a seguir as contribuições avaliadas como de menor impacto, em comparação com as contribuições já descritas, ou ainda incipientes (e que possibilitam continuidades da pesquisa em trabalhos futuros), sendo relacionadas a seguir como contribuições complementares: * Pesquisa de literatura científica sobre o estado-da-arte da área da Engenharia de Sistemas Neurais Artificiais. * Identificação de grupos internacionais de pesquisa de execução do SOM em hardware, os quais foram reconhecidos por publicarem regularmente seus estudos sobre diferentes tipos de implementações e categorias de circuitos computacionais. * Enumeração das justificativas e motivações mais frequentes na literatura para o processamento de sistemas neurais de computação em hardware. * Comparação e contraste das características de microprocessadores, GPUs, FPGAs e ASICs (tais como, custo médio do componente, paralelismo computacional oferecido e consumo típico de energia) para contextualização do tipo de aplicações que a escolha pela pesquisa com o dispositivo FPGA possibilita. * Levantamento das propriedades de computação do SOM em hardware mais frequentemente utilizadas nas pesquisas publicadas na literatura, tais como, quantidade de bits usados nos cálculos, tipo de representação de dados e arquitetura típica dos circuitos de execução das diferentes etapas de processamento do SOM. * Comparação do consumo de área do FPGA e da velocidade de processamento entre a execução da função de vizinhança tradicional gaussiana e a função de vizinhança proposta neste trabalho (com resultados obtidos de aproximadamente 4 vezes menos área do chip e 5 vezes mais velocidade de operação). * Caracterização do aumento dos recursos consumidos no chip e da velocidade de operação do sistema, em relação à implementação do SOM com diferentes complexidades (quantidade de estágios decrescentes do fator de aprendizado e da abertura da função de vizinhança) e comparação destas propriedades da arquitetura proposta em relação aos valores publicados na literatura. * Proposta de uma nova métrica para caracterização do erro topográfico na configuração final do SOM após o treinamento.
In the context of design electrical circuits for processing artificial neural networks, this work focuses on the study of Self-Organizing Maps (SOM) executed on FPGA chips. The work attempts to answer the following question: how should the computational architecture be designed to efficiently implement in FPGA each one of the SOM processing steps? More specifically, this thesis investigates the distinct possibilities that different SOM computing architectures offer, regarding the processing speed, the consumption of FPGA resources and the consistency to the theory that underlies this neural network model. The motivation of the present work is enabling the development of neural processing systems that exhibit the positive features typically associate to hardware implementations, such as, embedded processing and computational acceleration. MAIN CONTRIBUITIONS In the course of the investigation, the present work generated contributions with different degrees of impact. The most essential contribution from the point of view of structuring the research process is the theoretical basis of the hardware-oriented SOM properties. This is important because it allowed the construction of the foundations for the study of different circuit architectures, so that the developments remained consistent with the theory that underpins the neural computing model. Another major contribution is the proposal of a processor circuit for implementing SOM in FPGA, which is the state-of-the-art in computational speed measured in CUPS (Connections Updated Per Second). This processor allows achieving 52.67 GCUPS, during the training phase of the SOM, which means a gain of 100%, approximately, in relation to other published works. The acceleration enabled by the FPGA parallel processing developed in this work reaches three to four orders of magnitude compared with software implementations of the SOM with the same configuration. The highlights made in the text indicate pieces of writing that synthesize the idea presented. The last main contribution of the work is the characterization of the FPGA-based SOM. This evaluation is important because, although similar, the computing processes of neural models in hardware are not necessarily identical to the same processes implemented in software. Hence, this contribution can be described as the analysis of the impact of the implemented changes, regarding the FPGA-based SOM compared to traditional algorithms. The comparison was performed evaluating the measures of topographic and quantization errors for the outputs produced by both implementations. This work also generated medium impact contributions, which can be divided into two groups: empirical and theoretical. The first empirical contribution is the survey of SOM applications which can be made possible by hardware implementations. The papers presented in this survey are classified according to their research area - such as Industry, Robotics and Medicine - and, in general, they use SOM in applications that require computational speed or embedded processing. Therefore, the continuity of their developments is benefited by direct hardware implementations of the neural network. The other two empirical contributions are the applications employed for testing the circuits developed. The first application is related to the reception of telecommunications signals and aims to identify 16-QAM and 64-QAM symbols. These two modulation techniques are used in a variety of applications with mobility requirements, such as cell phones, digital TV on portable devices and Wi-Fi. The SOM is used to identify QAM distorted signals received with noise. This research work was published in the Springer Journal on Neural Computing and Applications: Sousa; Pires e Del-Moral-Hernandez (2017). The second is an image processing application and it aims to recognize human actions captured by video cameras. Autonomous image processing performed by FPGA chips inside video cameras can be used in different scenarios, such as automatic surveillance systems or remote assistance in public areas. This second application is also characterized by demanding high performance from the computing architectures. All the theoretical contributions with medium impact are related to the study of the properties of hardware circuits for implementing the SOM model. The first of these is the proposal of an FPGA-based neighborhood function. The aim of the proposal is to develop a computational function to be implemented on chip that enables an efficient alternative to both: the Gaussian function (traditionally employed in the SOM training process) and the rectangular function (used rudimentary in the first published works on hardware-based SOMs). The second of those contributions is the detailed description of the basic components and blocks used to compute the different steps of the SOM algorithm in hardware. The description of the processing architecture includes its internal circuits and computed functions, allowing the future works to use the architecture proposed. This detailed and functional description was accepted for publication in the IEEE World Congress on Computational Intelligence (WCCI 2018): Sousa et al. (2018). The development of an FPGA distributed implementation model for the SOM composes the third of those contributions. Such a model allows an execution of the neural network learning and operational phases without the use of a central control unit. The proposal achieves a global self-organizing behavior only by using local data exchanges among the neighboring processing elements. The description and characterization of the distributed model are published in a paper in the IEEE International Joint Conference on Neural Networks (IJCNN 2017): Sousa e Del-Moral-Hernandez (2017a). The last contribution of this group is the comparison between different FPGA architectures for implementing the SOM. This comparison has the function of evaluating and contrasting three different SOM architectures: the distributed model, the centralized model and the hybrid model. The tests performed and the results obtained are published in an article in the IEEE International Symposium on Circuits and Systems (ISCAS 2017): Sousa e Del-Moral-Hernandez (2017b). Finally, the contributions assessed as having a minor impact, compared to contributions already described, or still incipient (and which allow the continuity of the research in possible future works), are presented as complementary contributions: * Research in the scientific literature on the state-of-the-art works in the field of Artificial Neural Systems Engineering. * Identification of the international research groups on hardware-based SOM, which were recognized for regularly publishing their studies on different types of implementations and categories of computational circuits. * Enumeration of the justifications and motivations often mentioned in works on hardware developments of neural computing systems. * Comparison and contrast of the characteristics of microprocessors, GPUs, FPGAs and ASICs (such as, average cost, parallelism and typical power consumption) to contextualize the type of applications enabled by the choice of FPGA as the target device. * Survey of literature for the most commonly hardware properties used for computing the SOM, such as the number of bits used in the calculations, the type of data representation and the typical architectures of the FPGA circuits. * Comparison of the FPGA resources consumption and processing speed between the execution of the traditional Gaussian neighborhood function and the proposed alternative neighborhood function (with obtained results of approximately 4 times less chip area and 5 times more computational speed). * Characterization of the increase in chip resources consumptions and the decrease in system speeds, according to the implementations of the SOM with different complexities (such as, the number of stages in learning factor and the width of the neighborhood function). Comparison of these properties between the proposed architecture and the works published in the literature. * Proposal of a new metric for the characterization of the topographic error in the final configuration of the SOM after the training phase.

Work deal about self-organizing maps, especially about Kohonen self-organizing map. About creating of aplication, which realize creating and learning of self-organizing map. And about usage of self-organizing map for self-localization of robot.

Esse trabalho descreve o desenvolvimento de uma nova ferramenta quimioinformática designada de SISTEMATX que possibilitou a análise quimiotaxonômica da família Asteraceae, empregando novos parâmetros moleculares, bem como o estudo da relação quantitativa estrutura química atividade biológica de substâncias provenientes desse grupo vegetal. A família Asteraceae, uma das maiores entre as angiospermas, caracteriza-se quimicamente pela produção de sesquiterpenos lactonizados (SLs). Um total de 1111 (SLs), extraídos de 658 espécies, 161 gêneros, 63 subtribos e 15 tribos da família Asteraceae foram representados e cadastrados em duas dimensões no SISTEMATX e associados à respectiva origem botânica. A partir dessa codificação, o grau de oxidação e as estruturas em três dimensões de cada SL foram obtidos pelo sistema. Essas informações, associadas aos dados botânicos, foram exportadas para um arquivo texto, o qual permitiu a obtenção de vários tipos de descritores moleculares. Esses parâmetros moleculares foram correlacionados com o grau de oxidação médio por tribo e tiveram sua seleção realizada por regressão linear múltipla utilizando algoritmo genético. Equações com coeficientes estatísticos variando entre 0,725 ≤ r2 ≤ 0,981 e 0,647 ≤ Qcv2 ≤ 0,725 foram obtidas com apenas um descritor, possibilitando a identificação de algumas características estruturais relacionadas ao grau de oxidação. Não foi obtida nenhuma relação entre o grau de oxidação dos SL e a evolução das tribos da família Asteraceae. Os descritores moleculares também foram usados como dados de entrada para separar as ocorrências botânicas através de mapas auto-organizáveis (rede não supervisionada Kohonen). Os mapas gerados, com cada bloco de descritor, separaram as tribos da família Asteraceae com valores de índices de acerto total entre 66,7% e 83,6%. A análise desses resultados evidencia semelhanças entre as tribos Heliantheae, Helenieae, e Eupatorieae e, também, entre as tribos Anthemideae e Inuleae. Tais observações são coincidentes com as classificações sistemáticas propostas por Bremer, que utilizam principalmente dados morfológicos e, também, moleculares. A mesma abordagem foi utilizada para separar os ramos da tribo Heliantheae, segundo a classificação proposta por Stuessy, cuja separação é baseada no número de cromossomos das subtribos. Os mapas auto-organizáveis obtidos separam em duas regiões distintas os ramos A e C, com elevados índices de acerto total que variam entre 81,79% a 92,48%. Ambos os estudos demonstram que os descritores moleculares podem ser utilizados como uma ferramenta para classificação de táxons em níveis hierárquicos baixos, tais como tribos e subtribos. Adicionalmente, foi demonstrado que os marcadores químicos corroboram parcialmente com as classificações que empregam dados morfológicos e moleculares. Os descritores obtidos por fragmentos ou pela representação da estrutura dos SLs em duas dimensões foram suficientes para obtenção de resultados significativos, não sendo obtida melhora nos resultados com os descritores que utilizam a representação em três dimensões das estruturas. Paralelamente, um estudo adicional foi realizado relacionando a estrutura química, representada pelos mesmos descritores moleculares anteriormente mencionados, com a atividade citotóxica de 37 SLs frente às células tumorais da nasofaringe KB. Uma equação com índices estatísticos significativos (r2=0,826 e Qcv2=0,743) foi obtida. Os cinco descritores, selecionados a partir de uma equação estatisticamente mais significativa, representam uma descrição global de propriedades estéricas e características eletrônicas de cada molécula que auxiliaram na determinação de fragmentos estruturais importantes para a atividade citotóxica. Tal modelo permitiu verificar que os esqueletos carbônicos dos tipos guaianolídeo e pseudoguaianolídeo são encontrados nos SLs que apresentam maior atividade citotóxica.
This work describes the development of a new chemoinformatic tool named SISTEMATX that allowed the chemotaxonomic analysis of the Asteraceae family employing new molecular parameters, as well as the quantitative structure activity relationship study of compounds produced by this botanical group. The Asteraceae, one of the largest families among of angiosperms, is chemically characterized by the production of sesquiterpene lactones (SLs). A total of 1111 (SLs), extracted from 658 species, 161 genera, 63 subtribes and 15 tribes of the Asteraceae, were represented and registered in two dimensions in the SISTEMATX and associated with their botanical source. From this codification, the degree of oxidation and the structures in three dimensions of each SL were obtained by the system. These data linked with botanical origin were exported for a text file which allow the generation of several types of molecular descriptors. These molecular parameters were correlated with the average oxidation degree by tribe and were selected by multiple linear regressions using genetic algorithms. Equations with statistical coefficients varying between 0,725 ≤ r2 0,981 and 0,647 ≤ Qcv2 ≤ 0,725 were obtained with only one descriptor, making possible the identification of some structural characteristics related to the oxidation level. Any relationship between the degree of oxidation of SL and the tribes evolution of the family Asteraceae was not obtained. The molecular descriptors were also used as input data to separate the botanical occurrences through the self organizing-maps (unsupervised net Kohonen). The generated maps with each block descriptor, divide the Asteraceae tribes with total indexes values between 66,7% and 83,6%. The analysis of these results shows evident similarities among the Heliantheae, Helenieae and Eupatorieae tribes and, also, between the Anthemideae and Inuleae tribes. Those observations are in agreement with the systematic classifications proposed by Bremer, that use mainly morphologic and, also, molecular data. The same approach was utilized to separate the branches of the Heliantheae tribe, according to the Stuessys classification, whose division is based on the chromosome numbers of the subtribes. From the obtained self-organizing maps, two different areas (branches A and C) were separated with high hit indexes varying among 81,79% to 92,48%. Both studies demonstrate that the molecular descriptors can be used as a tool for taxon classification in low hierarchical levels such as tribes and subtribes. Additionally, was demonstrated that the chemical markers partially corroborate with the classifications that use morphologic and molecular data. Descriptors obtained by fragments or by the representation of the SL structures in two dimensions were sufficient to obtain significant results, and were not obtained better results with descriptors that utilize the structure representation in three dimensions. An additional study was accomplished relating the chemical structure, represented by the same molecular descriptors previously mentioned, with the cytotoxic activity of 37 SLs against tumoral cells derived from human carcinoma of the nasopharynx (KB). An equation with significant statistical indexes was obtained. The five descriptors, selected from the more statistical significant equation, shows a global description of sterical properties and electronic characteristics of each molecule that aid in the determination of important structural fragments for the cytotoxic activity. From the model can be verified that the carbon skeletons of the guaianolide and pseudoguaianolide types are encountered in the SLs that show the higher cytotoxic activity.

O presente trabalho é resultado da colaboração de pesquisadores do Centro de Engenharia Nuclear (CEN) e de pesquisadores do Centro de Biotecnologia (CB), ambos pertencentes ao IPEN, para o desenvolvimento de uma metodologia que visa auxiliar os profissionais citogeneticistas fornecendo uma ferramenta que automatize parte da rotina necessária para a avaliação qualitativa e quantitativa de danos biológicos em termos de aberração cromossômica. A técnica citogenética, sobre a qual esta ferramenta é desenvolvida, é a técnica de aberrações cromossômicas. Nela, são realizadas preparações citológicas de linfócitos de sangue periférico para que metáfases sejam analisadas e fotografadas ao microscópio e, com base na morfologia dos cromossomos, anomalias sejam investigadas. Quando esta tarefa é realizada manualmente, os cromossomos são analisados visualmente um a um pelo profissional citogeneticista, logo, trata-se de um processo minucioso em virtude da variação geral na aparência do cromossomo, do seu tamanho pequeno e do grande número de cromossomos por célula. Para um diagnóstico confiável, é necessário que várias células sejam analisadas, tornando-se uma tarefa repetitiva e demorada. Neste contexto, foi proposto o uso dos mapas auto-organizáveis para o reconhecimento automático de padrões morfológicos referentes às imagens de cromossomos humanos. Para isso, foi desenvolvido um método de extração de características por meio do qual é possível classificar os cromossomos em: dicêntricos, anéis, acrocêntricos, submetacêntricos e metacêntricos, com acerto de 93,4 % em relação ao diagnóstico dado por um profissional citogeneticista.
This work is a joint collaboration between Nuclear Energy Research Institute (IPEN), Nuclear Engineering Center and Biotechnology Center to develop a methodology aiming to assist cytogenetic professionals by providing a tool to automate part of the required routine to perform qualitative and quantitative evaluation of biological damage in terms of chromosomal aberration. The cytogenetic technique upon which this tool was developed, is the chromosome aberrations technique, in which cytological preparations of peripheral blood lymphocyte metaphases are performed to be analyzed and photographed under a microscope in order to investigating chromosomal aberration. Performed manually, the chromosomes are analyzed visually one by one by a cytogenetic professional, so it is a painstaking process due to the great deal of variation in the appearance of each chromosome, their small sizes and not to mention the high density of chromosomes per cell. In order to obtain a reliable diagnosis it is necessary that many cells be analyzed, which makes this a repetitive and time consuming process. In this context, the use of self-organizing maps for the automatic recognition of patterns relating to morphological pictures of human chromosomes has been proposed. For this, we developed a feature extraction method by which is possible to classify chromosomes in: dicentrics, ring-shaped, acrocentric, submetacentric and metacentric with 93.4% accuracy compared to diagnostic given by a professional cytogeneticist.

Depuis une quinzaine d'années, l'Imagerie par Résonance Magnétique fonctionnelle (IRMf) permet d'extraire de l'information sur le fonctionnement cérébral et particulièrement sur la localisation des processus cognitifs. L'information contenue par les acquisitions en IRMf est extraite à l'aide du modèle linéaire général et du processus d'inférence statistique. Bien que cette méthode dite « classique » ait permis de valider la plupart des modèles lésionnels de manière non invasive, elle souffre de certaines limites. Pour résoudre ce problème, différentes techniques d'analyse ont émergé et proposent une nouvelle façon d'interpréter les données de la neuroimagerie. Nous présentons deux nouvelles méthodes multivariées basées sur les cartes de Kohonen. Nos méthodes analysent les données IRMf avec le moins d'a priori possibles. En parallèle, nous tentons d'extraire de l'information sur les réseaux neuronaux impliqués dans les émotions. La première de ces méthodes s'intéresse à l'information de spécialisation fonctionnelle et la seconde à l'information de connectivité fonctionnelle. Nous présentons les résultats qui en découlent, puis chacune des méthodes est comparée à l'analyse dite classique en termes d'informations extraites. De plus, notre attention s'est focalisée sur la notion de valence émotionnelle et nous tentons d'établir l'existence d'un éventuel réseau partagé entre valence positive et valence négative. La constance de ce réseau est évaluée à la fois entre modalités perceptives et entre catégories de stimuli. Chacune des méthodes proposées permet de corroborer l'information recueillie par la méthode classique, en apportant de nouvelles informations sur les processus étudiés. Du point de vue des émotions, notre travail met en lumière un partage du réseau cérébral pour les va-lences négative et positive ainsi qu'une constance de cette information dans certaines régions cérébrales entre modalités perceptives et entre catégories
In the last fifteen years, functional magnetic resonance imaging (fMRI) have been used to extract information about cognitive processes location. The information contained in fMRI acquisitions is usually extracted using the general linear model coupled to the statistical inference process. Although this classical method has validated noninvasively most of the lesional models, it suffers from some limitations. To solve this problem, various analysis techniques have emerged and propose a new way of interpreting neuroimaging data. In this thesis, we present two multivariate methods to analyze fMRI data with the least possible a priori. In parallel, we are trying to extract information about brain emotion processing. The first method focuses on the brain functional specialization and the second method on the brain functional connectivity. After results presentation, each method is compared to the so-called classical analysis in terms of extracted information. In addition, emphasis was put on the concept of emotional valence. We try to establish the existence of a possible split between positive and negative valence networks. The consistency of the network is evaluated across both perceptual modalities and stimuli categories. Each of the proposed methods are as accurate as the conventional method and provide new highlights on the studied processes. From the perspective of emotions, our work highlights a shared brain network for positive and negative valences and a consistency of this information in some brain regions across both perceptual modalities and stimuli categories

Made available in DSpace on 2016-04-18T21:39:46Z (GMT). No. of bitstreams: 3 Fabiano Rodrigues Ferreira1.pdf: 1311600 bytes, checksum: 0965dfbc230f89f36675c671139868c2 (MD5) Fabiano Rodrigues Ferreira2.pdf: 3285153 bytes, checksum: a89026ec6080ef90e4d32d6b1d3ecf78 (MD5) Fabiano Rodrigues Ferreira3.pdf: 3363149 bytes, checksum: 1aeb3cbfdd346437db3d3527f88d4d81 (MD5) Previous issue date: 2008-02-29
Fundo Mackenzie de Pesquisa
Brazilian educational scenario suffers from the lack of incentive to a musical apprenticeship that leads students to reflect about their own reality. Due to the actual hegemonic politics, educational processes, in general, are characterized by diminishing student s potential for reflection, in a society that priorizes a strict technicist teaching, as contemporary society is. As a result, students are often not able to stablish relationships between what was learned and their own lives. Thus, it is necessary to have some mechanisms that could help the adaptation to student s cultural context, leading to a meaningful ethnic learning. Learning objects concept can be understood as examples of technological resources that appear in a way to organize and structure digital educational data. Such concept, althought is a new paradigm into educational ambit, has been widely used on educational systems by constant & crescent deliver of learning objects by Internet. In this way, this work focuses an adaptive learning object architecture, applied to the learning process of Brazilian musical rhythms, as an example. Such objects are dynamically retrieved from repositories through techniques based on self-organizing maps. Objects are selected in order to create learning resources adequate to some desirable adaptivity factor, as previous knowledge, learning styles or cultural aspects.
O cenário educacional brasileiro sofre com a falta de incentivo a um aprendizado musical que realmente faça o educando refletir sobre sua realidade. Devido à política hegemônica atual, os processos educativos, em geral, estão imersos numa alienação descontextualizante e no assistencialismo. O poder de pensamento e reflexão do educando acaba diminuindo consideravelmente numa sociedade que preza mais pelo ensino puramente tecnicista do que pelo incentivo à reflexão, como é o caso da sociedade contemporânea. O resultado disso acaba sendo uma inorganicidade educacional que faz com que o aluno não faça relação daquilo que aprendeu com sua própria vida. Torna-se necessário, portanto, estabelecer mecanismos que auxiliem a adaptação ao contexto cultural do mesmo, levando a uma etnoaprendizagem significativa e contextualizada. Entendem-se os objetos de aprendizagem como exemplos de recursos tecnológicos que surgiram como forma de organizar e estruturar materiais educacionais digitais. Tal conceito, embora seja um paradigma novo no âmbito da educação tem sido amplamente utilizado nos sistemas educacionais atuais através da constante e crescente disponibilização dos mesmos pela Internet. Dessa forma, este trabalho enfoca uma arquitetura de objetos de aprendizagem digitais adaptativos com uma aplicação no processo de aprendizagem de ritmos musicais brasileiros, como exemplo de utilização. Tais objetos são dinamicamente recuperados a partir de repositórios, através de técnicas baseadas em mapas auto-organizáveis. Objetos são selecionados de maneira a criar recursos de aprendizagem que sejam adequados a algum fator de adaptabilidade desejável para o contexto, como conhecimentos prévios, estilos de aprendizagem ou aspectos culturais.

This thesis deals with analysis of T-wave Alternans (TWA), periodical changes of T wave in ECG signal. Presence of these alternans may predict higher risk of sudden cardiac death. From the several possible ways of TWA classification, the training algorithms of self organizing maps are used in this thesis. Result of the thesis is a program, which in the first step detects QRS complexes in the signal. Then, in the next step, gained reference points are used for T-waves detection. Detected waves are represented by a vector of significant points, which is used as an input for artificial neural network. Final output of the program is a decision about presence of TWA in the signal and its rate.

This thesis discusses the design and the utilization of the artificial neural networks as ECG classifiers and the detectors of heart diseases in ECG signal especially myocardial ischaemia. The changes of ST-T complexes are the important indicator of ischaemia in ECG signal. Different types of ischaemia are expressed particularly by depression or elevation of ST segments and changes of T wave. The first part of this thesis is orientated towards the theoretical knowledges and describes changes in the ECG signal rising close to different types of ischaemia. The second part deals with to the ECG signal pre-processing for the classification by neural network, filtration, QRS detection, ST-T detection, principal component analysis. In the last part there is described design of detector of myocardial ischaemia based on artificial neural networks with utilisation of two types of neural networks back – propagation and self-organizing map and the results of used algorithms. The appendix contains detailed description of each neural networks, description of the programme for classification of ECG signals by ANN and description of functions of programme. The programme was developed in Matlab R2007b.

Made available in DSpace on 2016-03-15T19:38:09Z (GMT). No. of bitstreams: 1 Jose Antonio Corrales Torres.pdf: 1028407 bytes, checksum: b5574af2fd8d98d1a9fe03ff29b6aa07 (MD5) Previous issue date: 2008-03-05
In the contemporary society, information and knowledge grew in importance and have become the most valuable assets, space and time are less relevant and more vulnerable due to the increasing mobile technology. New procedures and processes were created towards security. The information classification is the primary requirement to adjust rules and procedures, the protection level and cost. The current process is manual, restricted by the knowledge of few people and subject to imperfections. This study suggests a method to classify the information, regarding its confidentiality, using groups generated by an Artificial Neural Network. The development of this method was supported by studies of methodologies applied to information protection, to the technology and business risk management, classification methodologies and control structures. The implementation made use of a Neural Network, based on the Self-Organization Maps (SOM) of Kohonen, due to its heavy specialization on groups handling. The study case objective was the implementation and it considered the information from universities, due to their various properties (administrative, pedagogic and scientific research). The analysis of the results indicated the similarity among the elements that composed the groups generated by the training of the Neural Network, complemented by calculations using the original weights. The viability of the application of the considered method to an organization was confirmed.
Na sociedade contemporânea, a informação e o conhecimento assumiram a importância de representar os ativos de maior valor, num cenário em que o espaço e o tempo, devido à tecnologia voltada à mobilidade, perderam a relevância e tornaram-se mais vulneráveis. Surgiram novos procedimentos e mecanismos destinados à segurança. A classificação das informações é o requisito fundamental para direcionar as medidas, o nível de proteção e o custo. Atualmente o processo é manual, restrito ao entendimento de algumas pessoas e sujeito a imperfeições. Este estudo propõe um método para classificar as informações, quanto à sua confidencialidade, em grupos gerados por uma Rede Neural Artificial. O desenvolvimento deste método foi pautado por estudos em metodologias destinadas à segurança das informações, ao gerenciamento de risco de negócio e tecnológico, metodologias para classificação e estruturas de controle. A implementação usou a Rede Neural, baseada nos Mapas Auto-Organizáveis (SOM) de Kohonen, devido à sua acentuada especialização no tratamento de grupos. O estudo de caso objetivou a implementação e contemplou as informações das universidades, em razão da diversidade de suas propriedades (administrativa, pedagógica e pesquisa científica). A análise dos resultados obtidos permitiu observar a semelhança dos elementos que compõe os grupos gerados pelo treinamento da Rede Neural, complementado por cálculos que utilizam os pesos iniciais. Mostrou-se a viabilidade da aplicação do método proposto para uma organização.

This paper deals with application of neural networks in image segmentation. First part is an introduction to image processing and neural networks, second part describes an implementation of segmentation system and presents results of experiments. The segmentation system enables to use different types of classifiers, various image features extraction and also to evaluate the success of segmentation. Two classifiers were created - a neural network (self-organizing map) and an algorithm K-means. Colour (RGB and HSV) and texture features and their combinations were used for classification. Texture features were extracted using a set of Gabor filters. Experiments with designed classifiers and feature extractors were carried out and results were compared.

Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2004.
Includes bibliographical references (leaves 61-62).
To take further steps along the path toward true artificial intelligence, systems must be built that are capable of learning about the world around them through observation and explanation. These systems should be flexible and robust in the style of the human brain and little precompiled knowledge should be given initially. As a step toward achieving this lofty goal, this thesis presents the self-organizing event map (SOEM) architcture. The SOEM architecture seeks to provide a way in which computers can be taught, through simple observation of the world, about typical events in a way that organized according to events that are observed by the system. In this manner, the event map produces clusters of similar events and provides an implicit representation of the regularity within the event space to which the system has been exposed. As part of this thesis, a test system that makes use of self-organizing event map architecture has been developed in conjunction with the Genesis Project at the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT. This system receives input through a natural-language text interface and, through repreated training cycles, becomes capable of discerning between typical and exceptional events. Clusters of similar events develop within the map and these clusters act as an implicit is flexible and robust. The self-organizing event map, as a data structure, stores a plane of event models that are continually updated and form of the more commonly used (and explicit) notion of scripts and capability lists. For example, a trained map may recognize that dogs often run, but never fly. Therefore if a new input is received that describes a flying dog, the map would be capable of identifying the event as exceptional
(cont.) (or simply erroneous) and that further attention should be paid. By using clusters of similarity as an implicit representation, the self-organizing event maps presented here more accurately mimic natural memory systems and do not suffer from being tied to the limitations of a specific explicit representation of regularity.
by Seth Ronald Tardiff.
M.Eng.

Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2006.
Includes bibliographical references (leaves 59-60).
In this thesis, I present a method for organizing and relating events represented in two domains: the transition-space domain, which focuses on change and the trajectory-space domain, which focuses on movement along paths. Particular events are described in both domains, and each description is fed into a self organizing map. After these self organizing maps have been trained with enough events, the maps are clustered independently. Then, after the two self organizing maps are clustered, the clusters in the two maps are themselves clustered, creating links between trajectory descriptions and the transition descriptions. Thus, I provide a method for relating events seen in multiple perspectives. After training with 1914 different sentences about motion, my implemented system noted that particular motions along a path are highly correlated with particular transitions. For example, "the bird flew to the top of a tree" is part of a trajectory cluster that is highly correlated with a transition cluster in which a motion appears and a distance first decreases and finally disappears.
by Estanislao L. Fidelholtz.
M.Eng.

Made available in DSpace on 2016-03-15T19:37:30Z (GMT). No. of bitstreams: 1 Mauro Ulisses Schneider.pdf: 1455677 bytes, checksum: 6eba28391f8f6910fbf5457a57119bd3 (MD5) Previous issue date: 2010-08-23
Fundo Mackenzie de Pesquisa
The use of biometric systems has been widely stimulated by both the government and private entities to replace or improve traditional security systems. Biometric systems are becoming increasingly indispensable to protecting life and property, mainly due to its robustness, reliability, difficult to counterfeit and fast authentication. In real world applications, the devices for image acquisition and the environment are not always controlled and may under certain circumstances produce noisy images or with large variations in tonality, texture, geometry, hindering segmentation and consequently the authentication of the an individual. To deal effectively with such problems, this dissertation investigates the possibility of using committee machines combined with digital image processing techniques for iris segmentation. The components employed in the composition of the committee machines are support vector clustering, k-means and self organizing maps. In order to evaluate the performance of the tools developed in this dissertation, the experimental results obtained are compared with related works reported in the literature. Experiments on publicity available UBIRIS database indicate that committee machine can be successfully applied to the iris segmentation.
A utilização de sistemas biométricos vem sendo amplamente; incentivados pelo governo e entidades privadas a fim de substituir ou melhorar os sistemas de segurança tradicionais. Os sistemas biométricos são cada vez mais indispensáveis para proteger vidas e bens, sendo robustos, confiáveis, de difícil falsificação e rápida autenticação. Em aplicações de mundo real, os dispositivos de aquisição de imagem e o ambiente nem sempre são controlados, podendo em certas circunstâncias produzir imagens ruidosas ou com grandes variações na tonalidade, textura, geometria, dificultando a sua segmentação e por conseqüência a autenticação do indivíduo. Para lidar eficazmente com tais problemas, nesta dissertação é estudado o emprego de comitês de máquinas em conjunto com técnicas de processamento de imagens digitais para a segmentação da íris. Os componentes estudados na composição do comitê de máquinas são agrupamento por vetores-suporte, k-means e mapas auto- organizáveis. Para a avaliação do desempenho das ferramentas desenvolvidas neste trabalho, os resultados obtidos são comparados com trabalhos relacionados na literatura. Foi utilizada a base de dados pública UBIRIS disponível na internet.

Atributos visuais de textura e forma foram investigados para a recuperação baseada em conteúdo de imagens mamográficas (CBIR). Para a similaridade de imagens, foi considerada a estrutura de densidade mamária, representada principalmente pelos tecidos fibro-glandulares. A pesquisa consistiu de três etapas: (1) Preparação e processamento das imagens; (2) Extração e seleção de atributos visuais de textura e forma; (3) Implementação de um sistema de recuperação de imagem. A primeira etapa consistiu dos processos de retirada de ruído do fundo da imagem, segmentação da região da mama, detecção da região de músculo peitoral, localização do mamilo e da segmentação da região de tecidos fibro-glandulares. Utilizou-se a equação de Difusão Anisotrópica com filtro de Wiener para retirada e suavização de ruídos encontrados na imagem e preservação da borda da mama. Para a segmentação da região da mama, foram utilizadas as técnicas de limiarização de Princípio de Máxima Entropia, Método de Preservação de Momento, Método de Otsu, Método interativo de Ridler & Carvard, Método de Reddi e Método da Matriz de Co-ocorrência. A melhor imagem foi escolhida numa tarefa supervisionada. A detecção automática da região do músculo peitoral foi feita com a combinação do operador de Canny e a transformada de Radon como detector de linha. A posição do mamilo foi detectada com a transformada de Radon como detector de direção de densidade. A segmentação da região de tecidos fibro-glandulares foi feita também com as técnicas de limiarização do Princípio de Máxima Entropia, Método de Preservação de Momento, e Método de Otsu. Momentos Estatísticos extraídos do Histograma, Medida de Granulometria, Momentos Estatísticos extraídos do Domínio de Radon, Momento de Hu, e Textura de Haralick foram investigados como atributos de textura. Medida de Área, Circularidade e Razão de Diâmetro foram investigados como atributos de forma. A rede de Mapas Auto-Organizáveis de Kohonen foi utilizada como sistema de recuperação de imagem. Foram utilizadas, neste trabalho, 1080 imagens do projeto de Banco de Imagens do HCFMRP-USP, módulo Mamografia. O treinamento e teste foram feitos com a técnica de \"leaving-one-out\" e os melhores resultados obtidos foram: Taxa de precisão de 91,07% para a combinação dos cinco grupos de atributos de Forma, Estatísticos Extraídos do Histograma, Momento de Hu, Espectral no Domínio de Radon e de Medida de Granulometria; taxa de precisão e revocação do coeficiente de correlação médio representadas pela área sob a curva com valor de 0,02351 dos grupos de atributos de forma, de Textura de Haralick e Momento de Hu. Os resultados obtidos indicaram a relevância de nosso trabalho e seu potencial de utilização para a recuperação baseada em conteúdo de imagens mamográficas.
Visual texture based on texture and shape features were investigated for content-based mammographic images retrieval (CBIR). For similarity of images, the mammary density structures were considered, mainly represented by fibro-glandular tissues. This research consisted of three stages: (1) Images preparation and processing; (2) Extraction and selection of the visual features; (3) Implementation of a retrieval system. The first stage consisted of noisy removing from the image background, breast region segmentation, pectoral muscle region detection, nipple localization and the fibro-glandular tissues region segmentation. The equation of Anisotropic Diffusion was used with Wiener filter for noisy removing with the breast region edge preservation. For the breast region segmentation, the Thresholding techniques were used of Maximum Entropy Principle, Moment Preserving Method, Otsu Method, Ridler & Carvard Method, Reddi Method and Co-occurrence Matrix Method. The better image was chosen in a supervised task. The automatic pectoral muscle region detection was made with the Canny operator and Radon Transform combination as straight line detector. The nipple position was detected with the Radon Transform as density direction detector. The fibro-glandular tissues region was also defined with the thresholding techniques of the Maximum Entropy Principle, Moment Preserving Method, and Otsu Method. The Statistical Moments extracted from the Histogram, Measured of Granulometry, Statistical Moments extracted in Radon Domain, Moment of Hu, and Haralick Textures were investigated as texture features. Area, Circularity and Diameter Ratio were investigated as shape features. The Self-Organizing Maps of Kohonen was used as image retrieval system. One thousand and eighty images of the HCFMRP-USP Database Project, Mammography Module, were used in this work. The training and test processes were realized with the \"leaving-one-out\" technique and the best results obtained were: The precision rate of 91,07% for the combination of the five following features group: Shape, Statistical Moments extracted of the Histogram, Moment of Hu, Statistical Moments extracted in Radon Domain and Measure of Granulometry; precision and revocation rates of the average coefficient of correlation represented by the area under the curve with value of 0,02351 for the three following features group: Shape, Haralick Textures and Moment de Hu. The results obtained indicated the relevance of our work for the content-based mammographic images retrieval.

Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2011.
Cataloged from PDF version of thesis.
Includes bibliographical references (p. 77).
Today, powerful programs readily parse English text; understanding, however, is another matter. In this thesis, I take a step toward understanding by introducing CLARIFY, a program that disambiguates words. CLARIFY identifies patterns in observed word contexts, and uses these patterns to select the optimal word sense for any specific situation. CLARIFY learns successful patterns by manipulating an accelerated Self-Organizing Map to save these example contexts and then references them to perform further context based disambiguation within the language. Through this process and after training on 125 examples, CLARIFY can now decipher that shrimp in the sentence "The shrimp goes to the store. " is a small-person, not relying on a literal definition of each word as a separate element but looking at the sentence as a fluid solution of many elements, thereby making the inference crustacean absurd. CLARIFY is implemented in 1500 lines of Java.
by Martin Marcel Couturier.
M.Eng.

Neste trabalho, um novo enfoque para detecção e diagnóstico de falhas (DDF) em robôs manipuladores é apresentado. Um robô com falhas pode causar sérios danos e pode colocar em risco o pessoal presente no ambiente de trabalho. Geralmente, os pesquisadores têm proposto esquemas de DDF baseados no modelo matemático do sistema. Contudo, erros de modelagem podem ocultar os efeitos das falhas e podem ser uma fonte de alarmes falsos. Aqui, duas redes neurais artificiais são utilizadas em um sistema de DDF para robôs manipuladores. Um perceptron multicamadas treinado por retropropagação do erro é usado para reproduzir o comportamento dinâmico do manipulador. As saídas do perceptron são comparadas com as variáveis medidas, gerando o vetor de resíduos. Em seguida, uma rede com função de base radial é usada para classificar os resíduos, gerando a isolação das falhas. Quatro algoritmos diferentes são empregados para treinar esta rede. O primeiro utiliza regularização para reduzir a flexibilidade do modelo. O segundo emprega regularização também, mas ao invés de um único termo de penalidade, cada unidade radial tem um regularização individual. O terceiro algoritmo emprega seleção de subconjuntos para selecionar as unidades radiais a partir dos padrões de treinamento. O quarto emprega o mapa auto-organizável de Kohonen para fixar os centros das unidades radiais próximos aos centros dos aglomerados de padrões. Simulações usando um manipulador com dois graus de liberdade e um Puma 560 são apresentadas, demostrando que o sistema consegue detectar e diagnosticar corretamente falhas que ocorrem em conjuntos de padrões não-treinados.
In this work, a new approach for fault detection and diagnosis in robotic manipulators is presented. A faulty robot could cause serious damages and put in risk the people involved. Usually, researchers have proposed fault detection and diagnosis schemes based on the mathematical model of the system. However, modeling errors could obscure the fault effects and could be a false alarm source. In this work, two artificial neural networks are employed in a fault detection and diagnosis system to robotic manipulators. A multilayer perceptron trained with backpropagation algorithm is employed to reproduce the robotic manipulator dynamical behavior. The perceptron outputs are compared with the real measurements, generating the residual vector. A radial basis function network is utilized to classify the residual vector, generating the fault isolation. Four different algorithms have been employed to train this network. The first utilizes regularization to reduce the flexibility of the model. The second employs regularization too, but instead of only one penalty term, each radial unit has a individual penalty term. The third employs subset selection to choose the radial units from the training patterns. The forth algorithm employs the Kohonen’s self-organizing map to fix the radial unit center near to the cluster centers. Simulations employing a two link manipulator and a Puma 560 manipulator are presented, demonstrating that the system can detect and isolate correctly faults that occur in nontrained pattern sets.

SILVA, Ana Cristina Cunha da. O uso de redes neurais auto-organizáveis na análise da transferência de conhecimentos prosódico em aprendizes brasileiros de língua inglesa. 2010, 201f. Tese (Doutorado em Linguística) – Universidade Federal do Ceará, Departamento de Letras Vernáculas, Programa de Pós-graduação em Linguística, Fortaleza-CE, 2010.
Submitted by nazareno mesquita (nazagon36@yahoo.com.br) on 2012-06-28T13:08:58Z No. of bitstreams: 1 2010_tese_ACCSilva.pdf: 2172197 bytes, checksum: 036ba2cdc331410f0516a0ba2abe520d (MD5)
Approved for entry into archive by Maria Josineide Góis(josineide@ufc.br) on 2013-10-10T13:22:45Z (GMT) No. of bitstreams: 1 2010_tese_ACCSilva.pdf: 2172197 bytes, checksum: 036ba2cdc331410f0516a0ba2abe520d (MD5)
Made available in DSpace on 2013-10-10T13:22:45Z (GMT). No. of bitstreams: 1 2010_tese_ACCSilva.pdf: 2172197 bytes, checksum: 036ba2cdc331410f0516a0ba2abe520d (MD5) Previous issue date: 2010
The objective of this dissertation was to investigate how the prosodic knowledge is organized in an early stage of L2 acquisition in Brazilian learners of English with the help of a connectionist neural network. The approach proposed in this research is first, to quantify the utterances of L2 learners in the form of LPC coefficients and other linguistic/phonetics features that can represent the phenomenon studied here (Transfer of the prosodic knowledge from Portuguese to English). This process is called speech feature extraction, an important step in the connectionist approach to speech processing. Second, since certain features of the lexical item or sentence produced by each learner are determined, these data are entered into the neural network to analyze the statistical properties (regularities) of the set of speakers as a whole. Third, visualization tools are used to analyze how the network organizes speakers and what information is most relevant to this process of group formation (e.g. proficiency level, a certain characteristic or property of speech, among others). The network is known as Self-Organizing Map (Self-Organizing Map, SOM). The SOM organizes speakers for similarity degree in well-defined groups (clusters). Application of SOM in this context is therefore innovative. The SOM network is implemented in Matlab environment using the SOMtoolbox package, which is a set of programming routines developed by the research group in Finland, also the inventors of the SOM. The simulation results indicate that SOM might be used more frequently to assess the degree of distance that a group of learners is to the group of native speakers. Thus, a neural network might be used as a tool in the context of determining the level of foreign language proficiency.
O objetivo desta tese foi investigar como o conhecimento prosódico está organizado em um estágio inicial de aquisição de L2 em aprendizes brasileiros de inglês com a ajuda de uma rede neural conexionista. A abordagem proposta neste trabalho consiste primeiramente em "quantificar" as elocuções dos aprendizes de L2 na forma de coeficientes LPC e outras características linguísticas/fonéticas que possam representar o fenômeno aqui estudado (Transferência do Conhecimento Prosódico do Português para o inglês). A este processo dá-se o nome de "extração de características" da fala (feature extraction), uma importante etapa na abordagem conexionista do processamento da fala. Em segundo lugar, uma vez determinadas as características do item lexical ou da frase produzida por cada aprendiz, são inseridos esses dados na rede neural a fim de analisar as propriedades (regularidades) estatísticas do conjunto de falantes como um todo. Em terceiro, utiliza-se ferramentas de visualização para analisar como a rede organiza os falantes e quais informações são mais relevantes para este processo de formação de grupos (e.g. nível de proficiência, uma certa característica ou propriedade da fala, entre outros). A rede utilizada é conhecida como Mapa Auto-Organizável (Self-Organizing Map, SOM). A rede SOM organiza os falantes por grau de similaridade em grupos bem definidos (clusters). A aplicação da rede SOM neste contexto é, portanto, inovadora. A rede SOM é implementada no ambiente Matlab usando o pacote Som toolbox, que é um conjunto de rotinas de programação desenvolvidas pelo grupo de pesquisa da Finlândia, também inventores da rede SOM. Os resultados das simulações apontam que a rede SOM pode vir a ser usada mais frequentemente para avaliar o grau de distância a que um grupo de aprendizes está do grupo de falantes nativos. Dessa forma, uma rede neural pode vir a ser aplicada como ferramenta no contexto de determinação de nível de proficiência em língua estrangeira.

As redes ópticas passivas (PON), em virtude da oferta de maior largura de banda a custos relativamente baixos, vêm se destacando como possível candidata para suprir a demanda dos novos serviços como, tráfego de voz, vídeo, dados e de serviços móveis, exigidos pelos usuários finais. Uma importante candidata, para realizar o controle de acesso nas PONs, é a técnica de acesso múltiplo por divisão de código óptico (OCDMA), por apresentar características relevantes, como maior segurança e capacidade flexível sob demanda. No entanto, agentes físicos externos, como as variações de temperatura ambiental no enlace, exercem uma influência considerável sobre as condições de operação das redes ópticas. Especificamente, nas OCDMA-PONs, os efeitos da variação de temperatura ambiental no enlace de transmissão, afetam o valor do pico do autocorrelação do código OCDMA a ser detectado, degradando a qualidade de serviço (QoS), além do aumento da taxa de erro de bit (BER) do sistema. O presente trabalho apresenta duas novas propostas de técnicas, utilizando sistemas inteligentes, mais precisamente, controladores lógicos fuzzy (FLC) aplicados nos transmissores e nos receptores das OCDMA-PONs, com o objetivo de mitigar os efeitos de variação de temperatura. Os resultados das simulações mostram que o desempenho da rede é melhorado quando as abordagens propostas são empregadas. Por exemplo, para a distância de propagação de 10 km e variações de temperatura de 20°C, o sistema com FLC, suporta 40 usuários simultâneos com a BER = 10-9, enquanto que, sem FLC, acomoda apenas 10. Ainda neste trabalho, é proposta uma nova técnica de classificação de códigos OCDMA, com o uso de redes neurais artificiais, mais precisamente, mapas auto-organizáveis de Kohonen (SOM), importante para que o sistema de gerenciamento da rede possa oferecer uma maior segurança para os usuários. Por fim, sem o uso de técnica inteligente, é apresentada, uma nova proposta de código OCDMA, cujo formalismo desenvolvido, permite generalizar a obtenção de código com propriedades distintas, como diversas ponderações e comprimentos de códigos.
Passive optical networks (PON), due to the provision of higher bandwidth at relatively low cost, have been excelling as a possible candidate to meet the demand of new services, such as voice traffic, video, data and mobile services, as required by end users. An important candidate to perform access control in PONs, is the Optical Code-Division Multiple-Access (OCDMA) technique, due to relevant characteristics, such as improved security and flexible capacity on demand. However, external physical agents, such as variations in environmental temperature on the Fiber Optic Link, have considerable influence on the operating conditions of optical networks. Specifically, in OCDMA-PONs, the effects of environmental temperature variation in the transmission link affect the peak value on the autocorrelation of the OCDMA code to be detected, degrading the quality of service (QoS), in addition to increasing the Bit Error Rate (BER) of the system. This thesis presents two new proposals of techniques using intelligent systems, more precisely, Fuzzy Logic Controllers (FLC) applied on the transmitters and receivers of OCDMA-PONs, in order to mitigate the effects of temperature variation. The simulation results show that the network performance is improved when the proposed approaches are employed. For example, for the propagation distance of 10 kilometers and temperature variations of 20°C, the FLC system supports 40 simultaneous users at BER = 10-9, whereas without the FLC, the system can accommodate only 10. Furthermore, in this work is proposed a new technique of OCDMA codes classification, using Artificial Neural Networks (ANN), more precisely, the Self-Organizing Maps (SOM) of Kohonen, important for the network management system to provide increased security for users. Finally, without the use of intelligent technique, it is presented a new proposal of OCDMA code, whose formalism developed, allows to generalize the code acquisition with distinct properties, such as different weights and length codes.

Thesis (Ph. D.) -- University of Maryland, College Park, 2004.
Thesis research directed by: Computer Science. Title from t.p. of PDF. Includes bibliographical references. Published by UMI Dissertation Services, Ann Arbor, Mich. Also available in paper.

The purpose of this thesis is to examine how semantic relations in a document collection can be visualized with a Kohonen self-organizing map. It can be seen as a map of the information space which can be used to support information retrieval. The study makes a comparison of the possible differences between a map that is based on morphologically unprocessed text and a map where the text has been lemmatized. The text that is being processed is the definitions of all the existing instances in WordNet of a random selection of indexing terms from the Times-collection. The purpose is to see if morphological processing somehow can show different semantic connections between term instances based on their definitions. Eventually some different cartographic and display methods are compared to examine their strengths and weaknesses when used as possible applications of information visualization. The results show only marginal advantage of visualization based on lemmatized text. The lemmatization brings together new instances of words but the semantic relations are far from unambiguous. The conclusion drawn from this study is that the authors didn't find any single visualization method that can show all aspects in a map. Different combinations of three and two dimensional methods might be required to get a better picture of an information space.
Uppsatsnivå: D

The ever-increasing density of computer storage devices has allowed the average user to store enormous quantities of multimedia content, and a large amount of this content is usually music. Current search techniques for musical content rely on meta-data tags which describe artist, album, year, genre, etc. Query-by-content systems, however, allow users to search based upon the actual acoustical content of the songs. Recent systems have mainly depended upon textual representations of the queries and targets in order to apply common string-matching algorithms and are often confined to a single query style (e.g., humming). These methods also lose much of the information content of the song which limits the ways in which a user may search. We present a query-by-content system which supports querying in several styles using a Self-Organizing Map as its basis. The results from testing our system show that it performs better than random orderings and is, therefore, a viable option for musical query-by-content.

We use ERA-Interim reanalysis data of 2 meter temperature to perform a pattern analysis of the Arctic temperatures exploiting an artificial neural network called Self Organizing-Map (SOM). The SOM method is used as a cluster analysis tool where the number of clusters has to be specified by the user. The different sized SOMs are analyzed in terms of how the size changes the representation of specific features. The results confirm that the larger the SOM is chosen the larger will be the root mean square error (RMSE) for the given SOM, which is followed by the fact that a larger number of patterns can reproduce more specific features for the temperature.
Wir benutzten das künstliche neuronale Netzwerk Self Organizing-Map (SOM), um eine Musteranalyse von ERA-Interim Reanalysedaten durchzuführen. Es wurden SOMs mit verschiedener Musteranzahl verglichen. Die Ergebnisse zeigen, dass SOMs mit einer größeren Musteranzahl deutlich spezifischere Muster produzieren im Vergleich zu SOMs mit geringen Musteranzahlen. Dies zeigt sich unter anderem in der Betrachtung der mittleren quadratischen Abweichung (RMSE) der Muster zu den zugeordneten ERA Daten.

With the growth of computer networks and the advancement of hardware technologies, unprecedented access to data volumes become accessible in a distributed fashion forming heterogeneous data sources. Understanding and combining these data into data warehouses, or merging remote public data into existing databases can significantly enrich the information provided by these data. This problem is called data integration: combining data residing at different sources, and providing the user with a unified view of these data. There are two issues with making use of remote data sources: (1) discovery of relevant data sources, and (2) performing the proper joins between the local data source and the relevant remote databases. Both can be solved if one can effectively identify semantically-related attributes between the local data sources and the available remote data sources. However, performing these tasks manually is time-consuming because of the large data sizes and the unavailability of schema documentation; therefore, an automated tool would be definitely more suitable. Automatically detecting similar entities based on the content is challenging due to three factors. First, because the amount of records is voluminous, it is difficult to perceive or discover information structures or relationships. Second, the schemas of the databases are unfamiliar; therefore, detecting relevant data is difficult. Third, the database entity types are heterogeneous and there is no existing solution for extracting a richer classification result from the processing of two different data types, or at least from textual and numerical data. We propose to utilize self-organizing maps (SOM) to aid the visual exploration of the large data volumes. The unsupervised classification property of SOM facilitates the integration of completely unfamiliar relational database tables and attributes based on the contents. In order to accommodate heterogeneous data types found in relational databases, we extended the term frequency – inverse document frequency (TF-IDF) measure to handle numerical and textual attribute types by unified vectorization processing. The resulting map allows the user to browse the heterogeneously typed database attributes and discover clusters of documents (attributes) having similar content. iii The discovered clusters can significantly aid in manual or automated constructions of data integrity constraints in data cleaning or schema mappings for data integration.

This thesis explored and visualized the relationships of documents data, based on the technique of self-organizing maps (SOM), a subtype of artificial neural network for visualizing high-dimensional data in low-dimensional views. The source data for this thesis are the full Extensible Markup Language (XML) texts of A Standard Corpus of Present Day Edited American English. The first step is transforming these XML files to produce a term-document matrix, including stop word removal, stemming, tf-idf (term frequency–inverse document frequency) weighting, global filtering; here rows of this matrix represent documents as n-dimensional vectors. Secondly, these vectors are clustered and visualized by SOM consisting of neurons, each neuron relatives to a set of documents with a certain number of same terms. Then a network has been constructed from SOM, with vertices set of neurons and documents, lines set of linkages between neurons and documents. Finally this network exports to the Pajek for analysis and final visualization.

In response to increasing numbers of cases involving digital media, and the increasing sizes of and number of pieces of media in those cases, forensic investigators are relying increasingly on triage techniques for prioritizing which media to review. This thesis describes a framework for clustering documents aquired during a digital forensics investigation on a self organizing(aka Kahonen) map allowing new documents to be categorized relative to existing documents. Furthermore the presented algorithm avoids the need to work with source documents but with sdhash fingerprints allowing a fifty-fold reduction in data required. To test the methodology, document fingerprints are regenerated from the SOM and compared.

Artificial Intelligence Lab, Department of MIS, University of Arizona
The Kohonen Self-Organizing Map (SOM) is an unsupervised learning technique for summarizing high-dimensional data so that similar inputs are, in general, mapped close to one another. When applied to textual data, SOM has been shown to be able to group together related concepts in a data collection and to present major topics within the collection with larger regions. Research in which properties of SOM were validated, called the Proximity and Size Hypotheses,is presented through a user evaluation study. Building upon the previous research in automatic concept generation and classification, it is demonstrated that the Kohonen SOM was able to perform concept clustering effectively, based on its concept precision and recall7 scores as judged by human experts. A positive relationship between the size of an SOM region and the number of documents contained in the region is also demonstrated.

Thesis (Ph. D.)--University of Florida, 2001.
Title from first page of PDF file. Document formatted into pages; contains ix, 117 p.; also contains graphics. Vita. Includes bibliographical references (p. 109-116).

Die Kartographie ist eine Wissenschaft, die in ihrem Charakter starke interdisziplinäre Züge aufweist. Sie zeigt sich in den verschiedensten Facetten und wird darum in den unterschiedlichsten Wissenschaften angewandt. Markantester Charakter ist, schon per Definition, die Modellierung von geowissenschaftlichen Ereignissen und Sachverhalten. „A unique facility for the creation and manipulation of visual or virtual representations of geospace – maps – to permit the exploration, analysis, understanding and communication of information about that space.“(ICA 2003) Aus dieser Definition wird die Charakteristik einer Kommunikationswissenschaft (Brassel) deutlich. Gerade seit dem Paradigmenwechsel der 1970er Jahre fließen zahlreiche weitere Aspekte wie Informatik, Semiotik und Psychologie in das Verständnis von Kartographie ein. Dadurch wird die Karte nicht mehr als reines graphisches Mittel verstanden, sondern als Träger und Übermittler von Informationen verstanden. Der Kartennutzer und dessen Verständnis von Karten rücken dabei immer weiter in den Vordergrund und werden „Ziel“ der kartographischen Verarbeitung. Aus diesem Verständnis heraus, möchte ich in der folgenden Arbeit einen relativ neuen Einfluss und Aspekt der Kartographie vorstellen. Es handelt sich um das Modell der Self-Organizing Maps (SOM), welches erstmalig Anfang der 1980er Jahre von Teuvo Kohonen vorgestellt wurde und deshalb auch, von einigen Autoren, als Kohonenmaps bezeichnet wird. Dem Typus nach, handelt es sich dabei um künstliche neuronale Netze, welche dem Nervensystem des menschlichen Gehirns nachempfunden sind und damit allgemein als eine Art selbständiger, maschineller Lernvorgang angesehen werden können. Im Speziellen sind Self-Organizing Maps ein unüberwachtes Lernverfahren, das in der Lage ist völlig unbekannte Eingabewerte zu erkennen und zu verarbeiten. Durch diese Eigenschaft eignen sie sich als optimales Werkzeug für Data Mining sowie zur Visualisierung von hochdimensionalen Daten. Eine Vielzahl von Wissenschaftlern hat diesen Vorteil bereits erkannt und das Modell in ihre Arbeit einbezogen oder auf dessen Verwendbarkeit analysiert. Deshalb möchte in dieser Arbeit, einige dieser Verwendungsmöglichkeiten und den daraus resultierenden Vorteil für die Kartographie aufzeigen.

Modern high-throughput technologies such as microarrays, next generation sequencing and mass spectrometry provide huge amounts of data per measurement and challenge traditional analyses. New strategies of data processing, visualization and functional analysis are inevitable. This thesis presents an approach which applies a machine learning technique known as self organizing maps (SOMs). SOMs enable the parallel sample- and feature-centered view of molecular phenotypes combined with strong visualization and second-level analysis capabilities. We developed a comprehensive analysis and visualization pipeline based on SOMs. The unsupervised SOM mapping projects the initially high number of features, such as gene expression profiles, to meta-feature clusters of similar and hence potentially co-regulated single features. This reduction of dimension is attained by the re-weighting of primary information and does not entail a loss of primary information in contrast to simple filtering approaches. The meta-data provided by the SOM algorithm is visualized in terms of intuitive mosaic portraits. Sample-specific and common properties shared between samples emerge as a handful of localized spots in the portraits collecting groups of co-regulated and co-expressed meta-features. This characteristic color patterns reflect the data landscape of each sample and promote immediate identification of (meta-)features of interest. It will be demonstrated that SOM portraits transform large and heterogeneous sets of molecular biological data into an atlas of sample-specific texture maps which can be directly compared in terms of similarities and dissimilarities. Spot-clusters of correlated meta-features can be extracted from the SOM portraits in a subsequent step of aggregation. This spot-clustering effectively enables reduction of the dimensionality of the data in two subsequent steps towards a handful of signature modules in an unsupervised fashion. Furthermore we demonstrate that analysis techniques provide enhanced resolution if applied to the meta-features. The improved discrimination power of meta-features in downstream analyses such as hierarchical clustering, independent component analysis or pairwise correlation analysis is ascribed to essentially two facts: Firstly, the set of meta-features better represents the diversity of patterns and modes inherent in the data and secondly, it also possesses the better signal-to-noise characteristics as a comparable collection of single features. Additionally to the pattern-driven feature selection in the SOM portraits, we apply statistical measures to detect significantly differential features between sample classes. Implementation of scoring measurements supplements the basal SOM algorithm. Further, two variants of functional enrichment analyses are introduced which link sample specific patterns of the meta-feature landscape with biological knowledge and support functional interpretation of the data based on the ‘guilt by association’ principle. Finally, case studies selected from different ‘OMIC’ realms are presented in this thesis. In particular, molecular phenotype data derived from expression microarrays (mRNA, miRNA), sequencing (DNA methylation, histone modification patterns) or mass spectrometry (proteome), and also genotype data (SNP-microarrays) is analyzed. It is shown that the SOM analysis pipeline implies strong application capabilities and covers a broad range of potential purposes ranging from time series and treatment-vs.-control experiments to discrimination of samples according to genotypic, phenotypic or taxonomic classifications.

Computing systems are becoming more and more complex and are assuming more and more responsibilities in all sectors of human activity. Science and technology information present a rich resource, essential for managing research and development programs. Many of today's applications are built as distribution systems. The Internet is one of the best-known distribution systems and is used by nearly everyone today. With a great deal of available data on the net in different languages, it is essential to use efficient methods to extract useful information from the data. Fortunately, the parallel growth of information and of analytical tools offer the promise of advanced decision aids to support research and development more effectively. Data mining, information retrieval and other information-based technologies especially nowadays, are receiving increased attention. The importance of English is well established in every field. Likewise, Arabic is also a major natural language, spoken by over 250 millions people in 21 Arab countries as the first language, and in Islamic countries it is used as a second language. It is one of the languages of the Semitic family and thus preserves the complexity of this group. Arabic is highly derivated, as well as being an inflected language, so it requires good stemming for effective text mining. Yet no standard approach to stemming has emerged. This work investigates some of the issues involved in achieving bilingual text mining from large bodies of electronic Arabic-English datasets. The main aim of this thesis is to address the above issues and provide the best framework. To address this aim, this thesis evaluates the current proposed preprocessing and SOM clustering algorithms. Our proposed MLTextMAES approach has the ability to perform the four main stages of standard text mining, taking into account pre-processing, clustering (via SOM) and test of quality. Thus we have employed SOM as a tool for the clustering of documents into groups with similar categories. To the author's knowledge there is no significant literature available regarding the SOM technique applied to Arabic-English text mining. The model is found to be useful in strategic decision-making settings. The results indicate that SOM is a feasible tool for multilingual languages, and presents several advantages over current methods. Our experimental results show improved clustering performance when using Arabic-English language documents for our datasets.

We used the self-organizing maps (SOMs) method on eight models that participated in the Coupled model intercomparison project phase 5 (CMIP5) and two different greenhouse gases (GHG) concentration experiments. The SOMs were created from the winter 500 hPa horizontal temperature flux for each model. The clustering by the SOM revealed that in addition to the three flux pathways found in reanalyses (Pacific, Atlantic and Siberian/continental pathway), superpositions of these occur for the free running climate models, which develop their dynamic more freely than the reanalyses. It was found that the general structure of fluxes is indirectly dependent on the GHG concentrations, as the derived results from SOM patterns are different between the two GHG concentrations. It is suggested that flux patterns change from stable cyclonic motion over the north pole to flux pathways that feature more meridional fluxes through the North Atlantic and North Pacific into the Arctic.
Die Methode der Self-Organizing Maps (SOMs) wurde auf acht CMIP5-Modelle mit jeweils zwei verschiedenen Treibhausgasszenarien angwendet. Die SOMs wurden für jedes Modell und jede der beiden Modelläufe für den horizontalen Temperaturfluss in 500 hPa im Winter erstellt. Zusätzlich zu den aus der Analyse von Reanalyse-Daten erwarteten drei Transportwegen (pazifisch, atlantisch und sibirisch/kontinental) wurden Überlagerungen dieser gefunden. Es konnte gezeigt werden, dass die grundsätzliche Struktur der Transporte indirekt abhängig von der Treibhausgaskonzentration ist. Die Ergebnisse deuten darauf hin, dass sich die generelle Struktur des atmosphärischen Transports von einer stabilen zyklonalen Bewegung über dem Nordpol sich zu Transporten verschiebt, welche meridionale Transporte über den Nortdatlantik und den Nordpazifik in die Arktis führen.