An important characteristic of influenza A is its ability to escape host immunity through antigenic drift. A novel influenza A strain that causes a pandemic confers full immunity to infected individuals, yet because of antigenic drift, these individuals have decreased immunity to drifted strains. We compute the required decrease in immunity so that a recurrence is possible. Models for influenza A must make assumptions on the host contact structure on which the disease spreads. By computing the reproduction number, we show that the classical random mixing assumption predicts an unrealistically large decrease of immunity before a recurrence is possible. We improve over the classical random mixing assumption by incorporating a contact network structure. A complication of contact networks is correlations induced by the initial pandemic. Thus, we provide a novel analytic derivation of such correlations and show that contact networks may require a dramatically smaller drop in immunity before recurrence. Hence, the key new insight is that on contact networks the establishment of a new strain is possible for much higher immunity levels of previously infected individuals than predicted by the commonly used random mixing assumption. This suggests that stable contacts like classmates, coworkers and family members are a crucial path for the spread of influenza in human population. / Graduate

Concept drift refers to a problem that is caused by a change in the data distribution in data mining. This leads to reduction in the accuracy of the current model that is used to examine the underlying data distribution of the concept to be discovered. A number of techniques have been introduced to address this issue, in a supervised learning (or classification) setting. In a classification setting, the target concept (or class) to be learned is known. One of these techniques is called “Ensemble learning”, which refers to using multiple trained classifiers in order to get better predictions by using some voting scheme. In a traditional ensemble, the underlying base classifiers are all of the same type. Recent research extends the idea of ensemble learning to the idea of using committees, where a committee consists of diverse classifiers. This is the main difference between the regular ensemble classifiers and the committee learning algorithms. Committees are able to use diverse learning methods simultaneously and dynamically take advantage of the most accurate classifiers as the data change. In addition, some committees are able to replace their members when they perform poorly. This thesis presents two new algorithms that address concept drifts. The first algorithm has been designed to systematically introduce gradual and sudden concept drift scenarios into datasets. In order to save time and avoid memory consumption, the Concept Drift Introducer (CDI) algorithm divides the number of drift scenarios into phases. The main advantage of using phases is that it allows us to produce a highly scalable concept drift detector that evaluates each phase, instead of evaluating each individual drift scenario. We further designed a novel algorithm to handle concept drift. Our Dynamic Committee for Concept Drift (DCCD) algorithm uses a voted committee of hypotheses that vote on the best base classifier, based on its predictive accuracy. The novelty of DCCD lies in the fact that we employ diverse heterogeneous classifiers in one committee in an attempt to maximize diversity. DCCD detects concept drifts by using the accuracy and by weighing the committee members by adding one point to the most accurate member. The total loss in accuracy for each member is calculated at the end of each point of measurement, or phase. The performance of the committee members are evaluated to decide whether a member needs to be replaced or not. Moreover, DCCD detects the worst member in the committee and then eliminates this member by using a weighting mechanism. Our experimental evaluation centers on evaluating the performance of DCCD on various datasets of different sizes, with different levels of gradual and sudden concept drift. We further compare our algorithm to another state-of-the-art algorithm, namely the MultiScheme approach. The experiments indicate the effectiveness of our DCCD method under a number of diverse circumstances. The DCCD algorithm generally generates high performance results, especially when the number of concept drifts is large in a dataset. For the size of the datasets used, our results showed that DCCD produced a steady improvement in performance when applied to small datasets. Further, in large and medium datasets, our DCCD method has a comparable, and often slightly higher, performance than the MultiScheme technique. The experimental results also show that the DCCD algorithm limits the loss in accuracy over time, regardless of the size of the dataset.

In this era of the Internet of Things and Big Data, a proliferation of connected devices continuously produce massive amounts of fast evolving streaming data. There is a need to study the relationships in such streams for analytic applications, such as network intrusion detection, fraud detection and financial forecasting, amongst other. In this setting, it is crucial to create data mining algorithms that are able to seamlessly adapt to temporal changes in data characteristics that occur in data streams. These changes are called concept drifts. The resultant models produced by such algorithms should not only be highly accurate and be able to swiftly adapt to changes. Rather, the data mining techniques should also be fast, scalable, and efficient in terms of resource allocation. It then becomes important to consider issues such as storage space needs and memory utilization. This is especially relevant when we aim to build personalized, near-instant models in a Big Data setting. This research work focuses on mining in a data stream with concept drift, using an online bagging method, with consideration to the memory utilization. Our aim is to take an adaptive approach to resource allocation during the mining process. Specifically, we consider metalearning, where the models of multiple classifiers are combined into an ensemble, has been very successful when building accurate models against data streams. However, little work has been done to explore the interplay between accuracy, efficiency and utility. This research focuses on this issue. We introduce an adaptive metalearning algorithm that takes advantage of the memory utilization cost of concept drift, in order to vary the ensemble size during the data mining process. We aim to minimize the memory usage, while maintaining highly accurate models with a high utility. We evaluated our method against a number of benchmarking datasets and compare our results against the state-of-the art. Return on Investment (ROI) was used to evaluate the gain in performance in terms of accuracy, in contrast to the time and memory invested. We aimed to achieve high ROI without compromising on the accuracy of the result. Our experimental results indicate that we achieved this goal.

The goal of this thesis is to explore some variations of neural networks. The thesis is mainly split into two parts: a variation of the shaping functions in neural networks and a variation of learning rules in neural networks. In the first part, we mainly investigate polynomial perceptrons - a perceptron with a polynomial shaping function instead of a linear one. We prove the polynomial perceptron convergence theorem and illustrate the notion by showing that a higher order perceptron can learn the XOR function through empirical experiments with implementation. In the second part, we propose three models (SMLP, SA, SA2) for stream learning and anomaly detection in streams. The main technique allowing these models to perform at a level comparable to the state-of-the-art algorithms in stream learning is the learning rule used. We employ mini-batch gradient descent algorithm and stochastic gradient descent algorithm to speed up the models. In addition, the use of parallel processing with multi-threads makes the proposed methods highly efficient in dealing with streaming data. Our analysis shows that all models have linear runtime and constant memory requirement. We also demonstrate empirically that the proposed methods feature high detection rate, low false alarm rate, and fast response. The paper on the first two models (SMLP, SA) is published in the 29th Canadian AI Conference and won the best paper award. The invited journal paper on the third model (SA2) for Computational Intelligence is under peer review.

Pas de résumé / In order to give a general statistical description of turbulence, one tries to identify universal statistical features, common to a wider class of turbulent flows. In 1988, Kraichnan and Panda discovered one such possibly universal feature, namely, the fact that Navier-Stokes turbulence tends to reduce the strength of the nonlinearity in its governing equations. In the flrst part of the manuscript we consider the strength of the nonlinear term and, more precisely, of its fluctuations in isotropic turbulence. In order to measure this strength, we compare to the case of a flow fleld with the same energy distribution where the modes are statistically independent, as is the case in Gaussian noise. It is shown that the turbulent flow self-organizes towards a state in which the nonlinearity is reduced, and it is discussed what the implications of this reduction are. Also, in two dimensions it is illustrated how this self-organization manifests itself through the appearance of well-deflned vortical flow structures. In the second part of the manuscript, we investigate the dynamics of the Hasegawa- Wakatani model, a model relevant in the study of magnetically conflned fusion plasmas. The two-dimensional version of this model is considered, which includes some key features of the turbulent dynamics of a tokamak-edge. We consider the limit of the model in which the nonlinearity is reduced with respect to the linear forces. For this weakly nonlinear, wave dominated regime, analytical predictions suggest the presence of a feedback loop in which energy is transferred to highly anisotropic zonal flows by nonlocal interactions. We confirm these predictions and we demonstrate a strong suppression of the turbulent radial particle flux. In wall bounded geometry, the same mechanism is observed and here also the flux is eflciently reduced by the turbulence-zonal flow interaction.

As últimas décadas têm apresentado um crescente número de contribuições para o entendimento sobre quando e onde ocorreu o surgimento do Homo sapiens. Modelos baseados nessas evidências, geralmente, sugerem que a gênese dos humanos modernos ocorreu na África, há cerca de 200.000 anos antes do presente, de onde migraram para as outras partes do mundo. Análises da diversidade genética de populações atuais corroboram esse cenário, ao sugerir que, a partir de uma origem única, a espécie foi, gradativamente, perdendo variabilidade à medida que as populações divergiram, espacial e temporalmente, umas das outras e de suas ancestrais africanas. No que se refere especificamente à morfologia craniana, diversos autores sugerem a existência deste mesmo padrão de decréscimo da variabilidade em função do distanciamento em relação a África, embora seja, também, reconhecida entre os especialistas a partição da diversidade craniana humana entre dois padrões fundamentais: um representado pela morfologia similar àquela que caracterizou os primeiros Homo sapiens, antes que o processo de raciação, no sentido de diversificação, tivesse ocorrido, representado pela denominada \"morfologia generalizada\"; e outro representado pelas demais variações morfológicas, correspondendo às populações já diversificadas fora da África, denominada \"morfologia especializada\". Nesse sentido, o entendimento dos processos evolutivos envolvidos nos eventos de diferenciação morfológica gera bastante controvérsia entre os especialistas. Embora a maioria das informações já obtidas aponte para o fato de que a morfologia craniana evoluiu, majoritariamente, por processos estocásticos, algumas evidências sugerem que, ao menos em condições ambientais extremas, algumas regiões anatômicas cranianas específicas tenham uma parcela de sua variabilidade morfológica fixada por seleção natural. Nesse contexto, o objetivo primordial desta pesquisa é caracterizar a evolução da variação craniana humana, abordada a partir de dois tópicos centrais: 1) A investigação da composição, padrão de ocorrência, distribuição e estruturação da diversidade morfológica craniana humana; e, 2) A análise do contexto evolutivo da variação observada no crânio humano, em função de suas características de integração, modularidade e estase evolutiva investigadas a partir da exploração de seus padrões de variância e covariância. Para tanto, foram utilizadas as características métricas cranianas (24 variáveis do protocolo Howells) de 9.287 indivíduos, distribuídos em 161 populações autóctones de dispersão mundial. Apenas indivíduos morfologicamente íntegros constituíram o banco de dados, eliminando qualquer efeito devido à ocorrência de \"missing values\". Informações adicionais às séries presentes no banco de dados foram utilizadas para uma melhor caracterização geográfica e cronológica dessas populações, e que possibilitou o cálculo das distâncias geográficas entre elas e a estratificação dos dados sob diferentes critérios. Bancos de dados complementares, compostos por marcadores moleculares (mtDNA e microssatélites) também foram utilizados para a análise exploratória comparativa de questões específicas. Os resultados obtidos para as análises da composição, distribuição e estruturação da diversidade craniana humana mostram que grupos populacionais particulares, normalmente associados à alguma região geográfica específica, apresentam padrões de diversificação diversos daqueles observados para todas as populações analisadas de maneira conjunta, o que sugere a ocorrência de respostas evolutivas específicas associadas às condições particulares, como seleção, por exemplo. Em relação às investigações do contexto evolutivo da variação observada, inferida pelos padrões de correlação, covariância e modularidade investigados em diferentes agrupamentos populacionais, os resultados gerados demonstraram que, de maneira geral, os padrões de variância/covariância e a magnitude dos padrões de correlação entre os caracteres apresentam-se de maneira estável, com raras exceções ao estado de estase evolutiva predominante. Em suma, os resultados obtidos através das diferentes estratégias empregadas nesta tese reforçam a ideia de que a evolução da morfologia craniana é melhor explicada por um modelo que assuma a ocorrência de diferentes ditames evolutivos, como deriva genética e seleção natural, por exemplo, que, devido ao recente processo de diversificação da espécie apresentam, de maneira generalizada, em estado de estase / The last decades have seen a growing number of contributions to the understanding of when and where was the emergence of Homo sapiens. Models based on this evidence generally suggests that the genesis of modern humans occurred in Africa some 200,000 years before present, where migrated to other parts of the world. Analysis of genetic diversity of current populations corroborate this scenario, suggesting that, from a single source, the species was gradually losing variability as the populations diverged, spatially and temporally, from each other and from their African ancestors. With regard specifically to the cranial morphology, several authors suggest the existence of this same pattern of decreasing variability as a function of distance from Africa, although it is also recognized among experts partition the human cranial diversity between two fundamental patterns: one represented by morphology similar to that characterized the first Homo sapiens before the process raciação in the sense diversifying, occurred, represented by the so-called \"general morphology\" and the other represented by other morphological variations, corresponding to the populations already been diversified Africa, called \"specialized morphology.\" In this sense, understanding the evolutionary processes involved in the events of morphological differentiation generates a lot of controversy among experts. Although most of the information already obtained point to the fact that the cranial morphology evolved mostly by stochastic processes, some evidence suggests that, at least in extreme environmental conditions, some cranial specific anatomical regions have a portion of their morphological variability determined by natural selection. In this context, the primary objective of this research is to characterize the evolution of human cranial variation, approached from two themes: 1) The investigation of the composition, pattern of occurrence, distribution and structuring of human cranial morphological diversity, and, 2) analysis of the context of evolutionary change observed in the human skull, due to its characteristics of integration, modularity and evolutionary stasis investigated from the exploitation of their patterns of variance and covariance. For this, we used the metric cranial characteristics (24 variables protocol Howells) of 9287 individuals distributed in 161 indigenous peoples worldwide dispersion. Only morphologically intact individuals constituted the database, eliminating any effect due to the occurrence of \"missing values\". Additional information on these series in the database were used to better characterize geographic and chronological these populations, and that allowed the calculation of geographical distances between them and the stratification of the data under different criteria. Databases additional compounds by molecular markers (mtDNA and microsatellites) were also used for exploratory comparative analysis of specific issues. The results for the analyzes of the composition, structure and distribution of human cranial diversity show that particular population groups, usually associated with a specific geographic region, provide diversification patterns different from those observed for all populations analyzed jointly, suggesting the occurrence of specific evolutionary responses associated with particular conditions, such as selection, for example. Regarding investigations of evolutionary context of the variation observed, inferred by patterns of correlation, covariance and modularity investigated in different population groups, the results generated showed that, in general, the patterns of variance / covariance and magnitude of correlation patterns between characters are presented in a stable manner, with rare exceptions the state of evolutionary stasis predominant. In summary, the results obtained through the different strategies employed in this thesis reinforce the idea that the evolution of cranial morphology is best explained by a model that assumes the occurrence of different evolutionary dictates, as genetic drift and natural selection, for example, that due to the recent process of diversification of species present in a generalized way, in a state of stasis

Soft gamma repeaters (SGRs) are neutron stars with ultra-strong magnetic fields, on the order of 1014 G. As the source of the strongest magnetic fields in the universe, they are ideal objects to study the behavior of matter and light in this extreme environment. SGRs emit recurrent short duration, 0.1s, bursts of soft gamma-rays/hard X-rays that are expected to be highly polarized in the 2-10 keV energy range. By measuring the polarization of these bursts we can learn about the strength and configuration of the magnetic fields, the geometry of the emission region and the mass/radius relationship of the neutron star. Using the archival RXTE/PCA data we analyzed ~3 Ms of observations for SGR1806-20 and SGR1900+14. Over 5000 bursts were detected from the sources and each distribution of burst fluence was found to be well fit by a power law with an exponent of 1.60±0.02 for SGR1806-20 and 1.64±0.03 for SGR1900+14. The power law form holds over 4 magnitudes of fluence and the exponents were found to be independent of the level of burst activity. The exponent values suggest that SGR bursts are associated with a self-organized critical system, similar to earthquakes. To measure the polarization of SGR bursts a wide-field-of-view, large area detector is needed. To accomplish this we designed and tested a negative ion time projection chamber (NITPC) X-ray polarimeter which uses nitromethane (CH3NO¬2) as an electronegative gas additive. Utilizing a double gas electron multiplier (GEM) NITPC with CO2+CH3NO2 as a gas mixture we successfully measured gas gains, imaged photoelectron tracks and measured distributions of their length, measured drift velocity of negative ions in various electric fields, and measured modulation from polarized and unpolarized X-ray sources between 3 and 8 keV. Based on the lab instrument results and our SGR burst fluence analysis we propose an instrument appropriately sized for a NASA Small Mission Explorer Mission (SMEX) that would be capable of measuring the polarization of hundreds of bursts from an SGR in a state of high burst activity.

Orientador: Evandro Pereira Prado / Resumo: Diversos fatores podem interferir nos depósitos da pulverização em plantas sendo o tamanho de gotas e a ponta de pulverização uma das mais importantes nesse processo. Diante disso, objetivou-se avaliar como o tamanho de gotas e a ponta de pulverização afetarão a quantidade e qualidade dos depósitos da pulverização em plantas de amendoim (Arachis hypogaea L.) nos diferentes estádios de desenvolvimento. A verificação da deposição em plantas de amendoim nos estádios de desenvolvimento dos primeiros ramos (V1) e nos estádios de alongamento de ginóforos (R2) em função do tamanho de gotas (fina, média, grossa e muito grossa – experimento 1) e da ponta de pulverização (TXA 8002 VK - Cônico, TTJ 60 11002 - Plano duplo, AIXR 110025 - Plano com indução de ar, DGTJ 11002 - Plano duplo de deriva reduzida, XR 8003 - Plano de faixa ampliada – experimento 2) foram realizadas em delineamento de blocos inteiramente casualizados com 4 tratamentos e 40 repetições (experimento 1) e 5 tratamentos e 40 repetições (experimento 2), pela pulverização do corante alimentício Azul Brilhante na concentração de 1,5 g. L-1. Os valores de deposição obtidos nos experimentos foram submetidos à análise de variâncias e as médias comparadas pelo teste de Scott-Knott. A uniformidade de distribuição dos depósitos foi ajustada a regressão logística. No primeiro experimento (tamanho de gotas × estádios fenológicos) as médias de deposição no estádio V1 foram todas significativamente diferentes ente si, e no estádio R... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Several factors can interfere with spray deposits on plants. It is known that the droplet size and the spray tip are some of the most important in this process. Therefore, the objective was to evaluate how the droplet size and the spray tip will affect the quantity and quality of spray deposits on peanut (Arachis hypogaea L.) plants at different stages of development. The verification of deposition in peanut plants in the stages of development of the first branches (V1) and in the stages of elongation of gynophores (R2), depending on the droplet size (fine, medium, coarse and very coarse - experiment 1) and the spray tip (TXA 8002 VK - Conical, TTJ 60 11002 - Double plane, AIXR 110025 - Air induction plane, DGTJ 11002 - Double reduced drift plane, XR 8003 - Extended range plane - experiment 2), were carried out in a completely randomized block design with 4 treatments and 40 repetitions (experiment 1) and 5 treatments and 40 repetitions (experiment 2), by spraying the Brilliant Blue dye food at concentration of 1.5 g L-1. The deposition values obtained in the experiments were subjected to analysis of variances and their averages were compared by the Scott-Knott test. The uniformity of deposit distribution was adjusted to logistic regression. In the first experiment (droplet size × phenological stages), the deposition averages in the V1 stage were all significantly different from each other; in the R2 stage, the treatments with fine and medium droplets differed from the treatm... (Complete abstract click electronic access below) / Mestre

In tasks with more than one path to succeed, it is possible that participants’ strategies vary and therefore, participants should not be analyzed as a homogeneous group. This thesis investigates individual differences in a two-dimensional categorization task with redundancy (i.e., a task where any of the two dimensions by itself suffices for perfect performance). Individual differences in learned knowledge and used knowledge are considered and studied. Participants first performed a categorization task with redundancy (training phase), and afterward were asked to do categorizations in which the previously redundant knowledge becomes decisive (testing phase). Using the data from the testing phase, dimension(s) learned by each participant were determined and the response patterns of each participant in the training phase was used to determine which dimension(s) were used. The used knowledge was assessed using two separate analyses, both of which look at accuracy and response time patterns, but in different ways. Analysis 1 uses iterative decision bound modeling and RT-distance hypothesis and Analysis 2 uses the stochastic version of general recognition theory. In Analysis 1, more errors and slower response times close to a decision bound perpendicular to a dimension indicate that a participant is using that dimension. Analysis 2 goes a step further and in addition to determining which dimension(s) are used, specifies in what way they were used (i.e., identifying the strategy of each participant). Possible strategies are described heuristically (unidimensional, time efficient and conservative) and then each heuristic is translated into a drift diffusion model by the unique way that strategy is assumed to affect trial-by-trial difficulty of the task. Finally, a model selection criterion is used to pick the strategy that is used by each participant.

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第22263号 / 理博第4577号 / 新制||理||1657(附属図書館) / 京都大学大学院理学研究科地球惑星科学専攻 / (主査)教授 田口 聡, 教授 秋友 和典, 准教授 藤 浩明 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM