Proefschrift--Utrecht. / Title also in English. Summary in English. Vita. Includes bibliographical references.

Thesis (M.S.)--Boise State University, 2009. / Includes abstract. Includes bibliographical references (leaves 66-68).

Thesis (M.S.)--Boise State University, 2009. / Title from t.p. of PDF file (viewed April 28, 2010). Includes abstract. Includes bibliographical references (leaves 66-68).

Proefschrift--Utrecht. / Title also in English. Summary in English. Vita. Includes bibliographical references.

Thesis (Ph. D.)--State University of New York at Binghamton, Psychology Department, 2005. / Includes bibliographical references.

Thesis (Ph.D.)--Duquesne University, 2006. / Title from document title page. Abstract included in electronic submission form. Includes bibliographical references (p.146-151) and index.

Thesis (Ph. D.)--University of Oregon, 2006. / Typescript. Includes vita and abstract. Includes bibliographical references (leaves 129-137). Also available for download via the World Wide Web; free to University of Oregon users.

Made available in DSpace on 2014-06-11T19:23:31Z (GMT). No. of bitstreams: 0 Previous issue date: 2004Bitstream added on 2014-06-13T20:30:17Z : No. of bitstreams: 1 alves_dbm_me_prud.pdf: 1723505 bytes, checksum: 9db3e55e3fb351fffc293805469954c5 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / O Global Navigation Satellite System (GNSS), que congrega os vários sistemas de posicionamento por satélite existentes, tem como principal objetivo viabilizar o posicionamento de baixa, média e alta precisão. Dentre os sistemas de posicionamento que integram o GNSS, o Global Positioning System (GPS) tem grande destaque. Mas as observáveis GPS, tal como todas as outras observáveis envolvidas nos processos de medidas, estão sujeitas a erros aleatórios, sistemáticos e grosseiros. Os erros aleatórios são inevitáveis, sendo, portanto, considerados uma propriedade inerente das observações. Erros grosseiros (outliers) devem ser eliminados através do processo de controle de qualidade. Erros sistemáticos podem ser parametrizados ou eliminados por técnicas apropriadas de observação. Eles degradam a acurácia do posicionamento realizado com o GPS. Esses erros incluem erros da órbita dos satélites GPS, multicaminho, erros de refração atmosférica, dentre outros. Dessa forma, alguns trabalhos recentes têm utilizado o modelo semiparamétrico e o método dos mínimos quadrados com penalidades (MMQ com penalidades) para atenuar os efeitos desses erros residuais, utilizando dados de receptores de monofrequência. No modelo semiparamétrico as variáveis estimadas são divididas em uma parte paramétrica (coordenadas da estação e ambigüidades), que é de interesse do usuário, e uma parte não-paramétrica (funções de erros que variam suavemente com o tempo). Assim, devido ao número de incógnitas ser maior que o usual, é utilizado o MMQ com penalidades. Essa técnica utiliza uma spline cúbica natural, cuja suavidade é determinada pelo parâmetro suavizador, calculado pela validação cruzada generalizada. Nesse método, os erros são modelados como funções que variam suavemente com o tempo... / The Global Navigation Satellite System (GNSS), that encompasses several satellite positioning systems, has as main goal to make available the low, medium and high precision positioning. Among the positioning systems that integrate GNSS, the Global Positioning System (GPS) has a great importance. But the GPS observables, like all other observables involved in a measurement process, are subject to random, systematic and outliers errors. The random errors are inevitable, being, therefore, considered an inherent property of the observations. Outliers should be eliminated through the quality control process. Systematic errors can be modeled or eliminated by appropriate observation techniques. The systematic errors degrade the accuracy of the positioning accomplished by GPS. These errors are those related to GPS satellites orbits, multipath, atmospheric refraction among others. Thus, some authors have been using the semiparametric model and the penalised least squares technique to mitigate these residual errors, using single frequency receiver data. In a semiparametric model the estimated variables are divided into a parametric part (station coordinates and ambiguities), which is of interest to the users, and a nonparametric one (composed by error functions that vary smoothly with time). However, due to the unknowns number being larger than the usual, the penalised least squares is used. This technique uses a natural cubic spline, whose smoothness is determined by a smoothing parameter, computed by using the generalized cross validation. In this method, the errors are modeled as functions which vary smoothly in time. And more, the systematic errors functions, ambiguities and station coordinates are estimated simultaneously. As a result, the ambiguities and the station coordinates are estimated with better reliability and accuracy than the conventional least square method...(Complete abstract click electronic access below)

This thesis investigates two source memory errors: observation inflation, where observed actions are misremembered as being performed; and self-action inflation in which self-performed actions are misremembered as having been performed by somebody else. It has been proposed that these inflations occur because of overlapping brain activity during observation and performance. This has been attributed to mirror neurone activity. To test this, observation and self-action inflations are investigated for different types of actions (meaningful, meaningless and communicative) known to evoke different mirror neurone activity. Different age groups (young adult, and elderly) were studied as were the effects of relative ethnicity between observer and performer. The Remember-Know-Guess paradigm was used. This showed that people make inflations with high qualitative details and confidence. As anticipated, elderly participants made significantly more observation inflations than young adults. Across both age groups, significantly more inflations occurred for communicative and meaningful actions than for meaningless actions supporting the idea that mirror neurones may be involved in formation of inflations. However when the effects of relative ethnicity were included in the paradigm it was found that significantly more observation inflations were formed after observing different ethnicity actors. It has been hypothesised that if mirror neurone involvement is involved in observation inflations then the highest number of inflations are expected for the same ethnicity condition because of the overlap between participant and performer. This thesis therefore suggests a less simplistic explanation of the underlying mechanisms responsible for these types of memory error.

A realização de previsões adequadas nas indústrias oportuniza o correto dimensionamento de diversos aspectos da gestão da produção. Um dos métodos empregados no intuito de melhorar a precisão das previsões é conhecido como combinação de previsões. Ao longo dos anos, foram publicados estudos de combinação a fim de comparar os métodos já existentes e indicar entre estes, qual o mais acurado. No entanto não há unanimidade em suas conclusões. Entre as combinações existentes, o método da média aritmética é reconhecido como um dos mais utilizados, enquanto que o método da variância mínima é por vezes apresentado como mais acurado e permite em sua formulação a consideração ou não da correlação entre os erros das previsões individuais. No intuito de identificar, para previsões em séries reais industriais, se existe diferença entre a acurácia das previsões individuais e de suas combinações é que este estudo está sendo proposto. A modelagem individual abordada é a ARIMA e a RNA e as medidas empregadas para a escolha do método mais preciso são MAPE, MAE e MSE. O trabalho está estruturado em três artigos, nos quais se realizam comparações entre técnicas de previsão individual e suas combinações. O Artigo 1 aborda a comparação entre as técnicas de previsão individual e as combinações por média aritmética e variância mínima simplificada. O Artigo 2, por sua vez, apresenta um estudo comparativo das técnicas de previsão individuais e as combinações por média aritmética e variância mínima, utilizando a correlação entre os erros na obtenção dos pesos de cada previsão. Por fim, um comparativo entre as três combinações que contemplam este estudo é explicitado no Artigo 3. Como principal resultado, destaca-se o desempenho superior obtido por meio dos métodos de combinação por variância mínima, em especial o método simplificado. / The adequate forecasting in industries allows the correct sizing of many aspects of production management. A method used to improve the precision of forecasts is the combination of predictions. Over time, many studies were conducted to evaluate the existent methods and to indicate which one is the most precise. However, there is no unanimity in those studies conclusions. Among the combination methods, the arithmetic average is recognized as the broadly applied, while the minimum variance is sometimes presented as more accurate allowing to consider the correlation between the errors of individual forecasts or not. This study proposes to identify, in real industrial predictions series, if there are differences between accuracy of individual forecasts and their combinations. The individual predictions are performed by ARIMA and ANN models, and the measures used to choose the best method are MAPE, MAE and MSE. This dissertation is structured as three articles, in which a series of comparisons between individual prediction techniques and their combinations. Article 1 addresses the comparison between the individual prevision techniques and the combination methods of mean arithmetic and simplified minimum variance. Article 2 presents a comparative study between the individual prevision techniques and the combination methods of mean arithmetic and minimum variance, considering errors correlated. The comparison between the three combinations presented in the previous articles is explained in Article 3. As main result of the dissertation, it is highlighted the superior performance obtained with the minimum variance combined methods, specially the simplified method.