Aprender conceitos provenientes de fluxos de dados é uma tarefa significamente diferente do aprendizado tradicional em lote. No aprendizado em lote, existe uma premissa implicita que os conceitos a serem aprendidos são estáticos e não evoluem significamente com o tempo. Por outro lado, em fluxos de dados os conceitos a serem aprendidos podem evoluir ao longo do tempo. Esta evolução é chamada de mudança de conceito, e torna a criação de um conjunto fixo de treinamento inaplicável neste cenário. O aprendizado incremental é uma abordagem promissora para trabalhar com fluxos de dados. Contudo, na presença de mudanças de conceito, conceitos desatualizados podem causar erros na classificação de eventos. Apesar de alguns métodos incrementais baseados no modelo de misturas gaussianas terem sido propostos na literatura, nota-se que tais algoritmos não possuem uma política explicita de descarte de conceitos obsoletos. Nesse trabalho um novo algoritmo incremental para fluxos de dados com mudanças de conceito baseado no modelo de misturas gaussianas é proposto. O método proposto é comparado com vários algoritmos amplamente utilizados na literatura, e os resultados mostram que o algoritmo proposto é competitivo com os demais em vários cenários, superando-os em alguns casos. / Learning concepts from data streams differs significantly from traditional batch learning. In batch learning there is an implicit assumption that the concept to be learned is static and does not evolve significantly over time. On the other hand, in data stream learning the concepts to be learned may evolve over time. This evolution is called concept drift, and makes the creation of a fixed training set be no longer applicable. Incremental learning paradigm is a promising approach for learning in a data stream setting. However, in the presence of concept drifts, out dated concepts can cause misclassifications. Several incremental Gaussian mixture models methods have been proposed in the literature, but these algorithms lack an explicit policy to discard outdated concepts. In this work, a new incremental algorithm for data stream with concept drifts based on Gaussian Mixture Models is proposed. The proposed methodis compared to various algorithms widely used in the literature, and the results show that it is competitive with them invarious scenarios, overcoming them in some cases.

Esse trabalho oferece uma avaliação da abundância do papagaio-de-peito-roxo (Amazona vinacea) para 2016 e 2017, combinando contagens em dormitórios ao longo de toda a distribuição da espécie, em escala global, com amostragens replicadas em dormitórios na região oeste de Santa Catarina (WSC), em escala local, Brasil. As contagens em escala global resultaram em 3888 e 4066 indivíduos em 2016 e 2017, respectivamente. As estimativas para o WSC foram de 945 ± 50 e 1393 ± 40 para os mesmos dois anos. Não foi observada nenhuma evidência de crescimento populacional de 2016 para 2017, pois o acréscimo no número de indivíduos foi acompanhado por aumento do esforço amostral em ambas escalas. Quando extrapolamos a abundância no WSC para toda a área de distribuição da espécie, segundo a IUCN, e pressupondo densidade homogênea, obtivemos valores que estão acima da contagem na escala global, mas dentro da mesma ordem de magnitude. Nosso resultado oferece uma base sólida para afirmar que o tamanho populacional global de A. vinacea é de milhares de indivíduos, mas não dezenas de milhares. Realizamos um esforço sistemático para considerar as principais fontes de incerteza na estimativa de abundância da espécie. Cada contagem, tanto na escala local quanto na global, incluíram visitas em todos os dormitórios conhecidos dentro de um intervalo de 10 dias, evitando duplas contagens devido ao movimento dos papagaios entre dormitórios. No WSC, a abundância foi estimada usando um N-Mixture Model implementado em contexto Bayesiano. Apesar de nossa estimativa de tamanho populacional e de área de distribuição serem maiores do que as consideradas pela IUCN, sugerimos que A. vinacea permaneça na categoria “Em Perigo”, até que sejam realizados estudos sobre tendência populacional. / We offer an assessment of Vinaceous parrot (Amazona vinacea) abundance in 2016 and 2017, combining roost counts over the whole range of the species, with a replicated survey of roosts at the local scale, in western Santa Catarina state (WSC), Brazil. The whole range counts amounted to 3888 and 4066 individuals in 2016 and 2017, respectively. The WSC estimates were 945 ± 50 and of 1393 ± 40 individuals, for the same two years. We found no evidence of population growth from 2016 to 2017 because the increase in numbers is accompanied by an increase in observation effort both in WSC and at the whole-range scale. When extrapolating the WSC abundance estimate to the whole IUCN extant range of the species under the simplifying assumption of homogenous population density, we obtain values above the whole-range counts, but within the same order of magnitude. Such result offers a sound basis for putting the global population size of A. vinacea in the thousands of individuals, but not in the tens of thousands of individuals. We made a systematic effort to address key sources of uncertainty in parrot abundance estimation. Each count, at the local or whole-range scale, includes visits to all relevant roosts within less than ten days time to avoid double counting due to movement between roosts. At the local scale, we estimated abundance using an N-Mixture Model of replicated count data, implemented in a Bayesian framework. Even though we estimate a larger population size and a bigger geographic range that those currently reported by the IUCN, we suggest that A. vinacea should remain in the ‘Endangered’ IUCN threat category, pending further investigation of population trends.

Acoustic analysis of animal vocalizations has been widely used to identify the presence of individual species, classify vocalizations, identify individuals, and determine gender. In this work automatic identification of speaker and gender of mice from ultrasonic vocalizations and speaker identification of meerkats from their Close calls is investigated. Feature extraction was implemented using Greenwood Function Cepstral Coefficients (GFCC), designed exclusively for extracting features from animal vocalizations. Mice ultrasonic vocalizations were analyzed using Gaussian Mixture Models (GMM) which yielded an accuracy of 78.3% for speaker identification and 93.2% for gender identification. Meerkat speaker identification with Close calls was implemented using Gaussian Mixture Models (GMM) and Hidden Markov Models (HMM), with an accuracy of 90.8% and 94.4% respectively. The results obtained shows these methods indicate the presence of gender and identity information in vocalizations and support the possibility of robust gender identification and individual identification using bioacoustic data sets.

Microarray technology permit us to study the expression levels of thousands of genes simultaneously. The technique has a wide range of applications including identification of genes that change their expression in cells due to disease or drug stimuli. The dissertation is addressing statistical methods for the selection of differentially expressed genes in two experimental conditions. We propose two different methods for the selection of differentially expressed genes. The first method is a classical approach, where we consider a common distribution for the summary measure of equally expressed genes. To estimate this common distribution, the Johnson system of distribution is used. The advantage of using Johnson system is that, there is no need of a parametric assumption for gene expression data. In contrast to other classical methods, in the proposed method, there is a sharing of information across the genes by the assumption of a common distribution for the summary measure of equally expressed genes. The second method is the gene selection using a mixture model approach and Baye's theorem. This approach also uses the Johnson System of distribution for the estimation of distribution of summary measure. Johnson system of distribution has the flexibility of covering a wide variety of distributional shapes. This system provides a unique distribution corresponding to each pair of mathematically possible values of skewness and kurtosis. The significant flexibility of Johnson system is very useful in characterizing the complicated data set like microarray data. In this dissertation we propose a novel algorithm for the estimation of the four parameters of the Johnson system.

Model-based clustering is based on a finite mixture of distributions, where each mixture component corresponds to a different group, cluster, subpopulation, or part thereof. Gaussian mixture distributions are most often used. Criteria commonly used in choosing the number of components in a finite mixture model include the Akaike information criterion, Bayesian information criterion, and the integrated completed likelihood. The best model is taken to be the one with highest (or lowest) value of a given criterion. This approach is not reasonable because it is practically impossible to decide what to do when the difference between the best values of two models under such a criterion is ‘small’. Furthermore, it is not clear how such values should be calibrated in different situations with respect to sample size and random variables in the model, nor does it take into account the magnitude of the likelihood. It is, therefore, worthwhile considering a model-averaging approach. We consider an averaging of the top M mixture models and consider applications in clustering and classification. In the course of model averaging, the top M models often have different numbers of mixture components. Therefore, we propose a method of merging Gaussian mixture components in order to get the same number of clusters for the top M models. The idea is to list all the combinations of components for merging, and then choose the combination corresponding to the biggest adjusted Rand index (ARI) with the ‘reference model’. A weight is defined to quantify the importance of each model. The effectiveness of mixture model averaging for clustering is proved by simulated data and real data under the pgmm package, where the ARI from mixture model averaging for clustering are greater than the one of corresponding best model. The attractive feature of mixture model averaging is it’s computationally efficiency; it only uses the conditional membership probabilities. Herein, Gaussian mixture models are used but the approach could be applied effectively without modification to other mixture models. / Paul McNicholas

Nesta dissertação está proposta uma metodologia para fins de detecção de mudanças do uso do solo em imagens multitemporais de sensoriamento remoto. Em lugar de classificar os pixels de imagens que cobrem uma cena, em duas classes exaustivas e mutuamente excludentes (mudança, não-mudança), propõe-se adotar uma abordagem do tipo fuzzy, na qual são estimados os graus de pertinência às classes mudança e não-mudança. Com este objetivo adota-se aqui uma abordagem em nível de sub-pixel na estimação dos graus de pertinência para cada pixel. Esta abordagem se mostra mais adequada para fins de modelagem do que ocorre em cenas naturais, onde as alterações que acontecem ao longo de um período de tempo tendem a apresentar uma variação contínua em lugar de discreta. Em uma segunda etapa, os graus de pertinência estimados recebem um ajustamento adicional por meio da introdução de informações de contexto espacial. A metodologia proposta foi testada por meio de três experimentos, um empregando uma imagem sintética e dois utilizando imagens reais. A partir da análise quantitativa dos resultados e comparação com estudos semelhantes, comprova-se a adequação da metodologia proposta. / In this dissertation it is proposed a new methodology to land use change detection in remote sensing multitemporal image data. Rather than applying a rigid labeling of the pixels in the image data into two classes (change, no-change), we propose estimating the degrees of membership to classes change and no-change in a fuzzy-like fashion. To this end, a sub-pixel approach is implemented to detect the degree of change in every pixel. This methodology aims at modeling natural scenes in a more realistic way, since changes in natural scenes tend to occur in a continuum rather than in a sharp distinctive way. In a second step, the estimated values for the degrees of membership are further refined by means of spatial context information. Three experiments were performed to test the proposed methodology, one employing synthetic data and two using real image data. From the quantitative analysis of the results and from similar studies we can prove the adequacy of the proposed methodology.

Esse trabalho oferece uma avaliação da abundância do papagaio-de-peito-roxo (Amazona vinacea) para 2016 e 2017, combinando contagens em dormitórios ao longo de toda a distribuição da espécie, em escala global, com amostragens replicadas em dormitórios na região oeste de Santa Catarina (WSC), em escala local, Brasil. As contagens em escala global resultaram em 3888 e 4066 indivíduos em 2016 e 2017, respectivamente. As estimativas para o WSC foram de 945 ± 50 e 1393 ± 40 para os mesmos dois anos. Não foi observada nenhuma evidência de crescimento populacional de 2016 para 2017, pois o acréscimo no número de indivíduos foi acompanhado por aumento do esforço amostral em ambas escalas. Quando extrapolamos a abundância no WSC para toda a área de distribuição da espécie, segundo a IUCN, e pressupondo densidade homogênea, obtivemos valores que estão acima da contagem na escala global, mas dentro da mesma ordem de magnitude. Nosso resultado oferece uma base sólida para afirmar que o tamanho populacional global de A. vinacea é de milhares de indivíduos, mas não dezenas de milhares. Realizamos um esforço sistemático para considerar as principais fontes de incerteza na estimativa de abundância da espécie. Cada contagem, tanto na escala local quanto na global, incluíram visitas em todos os dormitórios conhecidos dentro de um intervalo de 10 dias, evitando duplas contagens devido ao movimento dos papagaios entre dormitórios. No WSC, a abundância foi estimada usando um N-Mixture Model implementado em contexto Bayesiano. Apesar de nossa estimativa de tamanho populacional e de área de distribuição serem maiores do que as consideradas pela IUCN, sugerimos que A. vinacea permaneça na categoria “Em Perigo”, até que sejam realizados estudos sobre tendência populacional. / We offer an assessment of Vinaceous parrot (Amazona vinacea) abundance in 2016 and 2017, combining roost counts over the whole range of the species, with a replicated survey of roosts at the local scale, in western Santa Catarina state (WSC), Brazil. The whole range counts amounted to 3888 and 4066 individuals in 2016 and 2017, respectively. The WSC estimates were 945 ± 50 and of 1393 ± 40 individuals, for the same two years. We found no evidence of population growth from 2016 to 2017 because the increase in numbers is accompanied by an increase in observation effort both in WSC and at the whole-range scale. When extrapolating the WSC abundance estimate to the whole IUCN extant range of the species under the simplifying assumption of homogenous population density, we obtain values above the whole-range counts, but within the same order of magnitude. Such result offers a sound basis for putting the global population size of A. vinacea in the thousands of individuals, but not in the tens of thousands of individuals. We made a systematic effort to address key sources of uncertainty in parrot abundance estimation. Each count, at the local or whole-range scale, includes visits to all relevant roosts within less than ten days time to avoid double counting due to movement between roosts. At the local scale, we estimated abundance using an N-Mixture Model of replicated count data, implemented in a Bayesian framework. Even though we estimate a larger population size and a bigger geographic range that those currently reported by the IUCN, we suggest that A. vinacea should remain in the ‘Endangered’ IUCN threat category, pending further investigation of population trends.

Dans le cadre de cette thèse, nous nous intéressons à plusieurs problématiques liées au traitement de données 3D. La première concerne la détection et le suivi de personnes dans des séquences d'images de profondeur. Nous proposons une amélioration d'une méthode existante basée sur une étape de segmentation, puis de suivi des personnes. La deuxième problématique abordée est la détection et la modélisation de têtes dans un nuage de points 3D. Pour cela, nous adoptons une approche probabiliste basée sur un nouveau modèle de mélange sphérique. La dernière application traitée est liée à la restauration d'images de profondeur présentant des données manquantes. Nous proposons pour cela d'utiliser une méthode d'approximation de surface par Dm-splines d'interpolation avec changements d'échelle pour approximer et restaurer les données. Les résultats présentés illustrent l'efficacité des algorithmes développés. / In this dissertation, we are interested in several issues related to 3D data processing. The first one concerns people detection and tracking in depth map sequences. We propose an improvement of an existing method based on a segmentation stage followed by a tracking module. The second issue is head detection and modelling in 3D point clouds. In order to do this, we adopt a probabilistic approach based on a new spherical mixture model. The last considered application deals with the restoration of deteriorated depth maps. To solve this problem, we propose to use a surface approximation method based on interpolation Dm-splines with scale transforms to approximate and restore the image. Presented results illustrate the efficiency of the developed algorithms.

Nesta dissertação está proposta uma metodologia para fins de detecção de mudanças do uso do solo em imagens multitemporais de sensoriamento remoto. Em lugar de classificar os pixels de imagens que cobrem uma cena, em duas classes exaustivas e mutuamente excludentes (mudança, não-mudança), propõe-se adotar uma abordagem do tipo fuzzy, na qual são estimados os graus de pertinência às classes mudança e não-mudança. Com este objetivo adota-se aqui uma abordagem em nível de sub-pixel na estimação dos graus de pertinência para cada pixel. Esta abordagem se mostra mais adequada para fins de modelagem do que ocorre em cenas naturais, onde as alterações que acontecem ao longo de um período de tempo tendem a apresentar uma variação contínua em lugar de discreta. Em uma segunda etapa, os graus de pertinência estimados recebem um ajustamento adicional por meio da introdução de informações de contexto espacial. A metodologia proposta foi testada por meio de três experimentos, um empregando uma imagem sintética e dois utilizando imagens reais. A partir da análise quantitativa dos resultados e comparação com estudos semelhantes, comprova-se a adequação da metodologia proposta. / In this dissertation it is proposed a new methodology to land use change detection in remote sensing multitemporal image data. Rather than applying a rigid labeling of the pixels in the image data into two classes (change, no-change), we propose estimating the degrees of membership to classes change and no-change in a fuzzy-like fashion. To this end, a sub-pixel approach is implemented to detect the degree of change in every pixel. This methodology aims at modeling natural scenes in a more realistic way, since changes in natural scenes tend to occur in a continuum rather than in a sharp distinctive way. In a second step, the estimated values for the degrees of membership are further refined by means of spatial context information. Three experiments were performed to test the proposed methodology, one employing synthetic data and two using real image data. From the quantitative analysis of the results and from similar studies we can prove the adequacy of the proposed methodology.

abstract: Data imbalance and data noise often coexist in real world datasets. Data imbalance affects the learning classifier by degrading the recognition power of the classifier on the minority class, while data noise affects the learning classifier by providing inaccurate information and thus misleads the classifier. Because of these differences, data imbalance and data noise have been treated separately in the data mining field. Yet, such approach ignores the mutual effects and as a result may lead to new problems. A desirable solution is to tackle these two issues jointly. Noting the complementary nature of generative and discriminative models, this research proposes a unified model fusion based framework to handle the imbalanced classification with noisy dataset. The phase I study focuses on the imbalanced classification problem. A generative classifier, Gaussian Mixture Model (GMM) is studied which can learn the distribution of the imbalance data to improve the discrimination power on imbalanced classes. By fusing this knowledge into cost SVM (cSVM), a CSG method is proposed. Experimental results show the effectiveness of CSG in dealing with imbalanced classification problems. The phase II study expands the research scope to include the noisy dataset into the imbalanced classification problem. A model fusion based framework, K Nearest Gaussian (KNG) is proposed. KNG employs a generative modeling method, GMM, to model the training data as Gaussian mixtures and form adjustable confidence regions which are less sensitive to data imbalance and noise. Motivated by the K-nearest neighbor algorithm, the neighboring Gaussians are used to classify the testing instances. Experimental results show KNG method greatly outperforms traditional classification methods in dealing with imbalanced classification problems with noisy dataset. The phase III study addresses the issues of feature selection and parameter tuning of KNG algorithm. To further improve the performance of KNG algorithm, a Particle Swarm Optimization based method (PSO-KNG) is proposed. PSO-KNG formulates model parameters and data features into the same particle vector and thus can search the best feature and parameter combination jointly. The experimental results show that PSO can greatly improve the performance of KNG with better accuracy and much lower computational cost. / Dissertation/Thesis / Doctoral Dissertation Industrial Engineering 2014