<p>Random numbers are required for cryptographic applications such as IT security products, smart cards etc. Hardwarebased random number generators are widely employed. Cryptographic algorithms are implemented on FieldProgrammable Gate Arrays (FPGAs). In this work a True Random Number Generator (TRNG) employed for spaceapplication was designed, investigated and evaluated. Several cryptographic requirements has to be satisfied for therandom numbers. Two different noise sources was designed and implemented on the FPGA. The first design wasbased on ring oscillators as a noise source. The second design was based on astable oscillators developed on a separatehardware board and interfaced with the FPGA as another noise source. The main aim of the project was to analyse theimportant requirement of independent noise source on a physical level. Jitter from the oscillators being the source forthe randomness, was analysed on both the noise sources. The generated random sequences was finally subjected tostatistical tests.</p>

Random numbers are required for cryptographic applications such as IT security products, smart cards etc. Hardwarebased random number generators are widely employed. Cryptographic algorithms are implemented on FieldProgrammable Gate Arrays (FPGAs). In this work a True Random Number Generator (TRNG) employed for spaceapplication was designed, investigated and evaluated. Several cryptographic requirements has to be satisfied for therandom numbers. Two different noise sources was designed and implemented on the FPGA. The first design wasbased on ring oscillators as a noise source. The second design was based on astable oscillators developed on a separatehardware board and interfaced with the FPGA as another noise source. The main aim of the project was to analyse theimportant requirement of independent noise source on a physical level. Jitter from the oscillators being the source forthe randomness, was analysed on both the noise sources. The generated random sequences was finally subjected tostatistical tests.

Submitted by Marilene Donadel (marilene.donadel@unioeste.br) on 2017-09-12T00:16:13Z No. of bitstreams: 1 Angelica_D_Lira_2017.pdf: 559734 bytes, checksum: 2f3500ed0adb917c1925679afb9dc768 (MD5) / Made available in DSpace on 2017-09-12T00:16:13Z (GMT). No. of bitstreams: 1 Angelica_D_Lira_2017.pdf: 559734 bytes, checksum: 2f3500ed0adb917c1925679afb9dc768 (MD5) Previous issue date: 2017-04-28 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Several laboratory procedures and statistical methods were developed to evaluate diet in fish. Thus, through a scientometric approach, we identified patterns in articles on fish feeding ecology. The database used for the survey was Thomson Reuters (ISI Web of Knowledge, apps.webofknowledge.com). In which, we observed a significant increase in the number of articles on feeding ecology of fish over time. Among the 1319 articles evaluated, the most frequent characteristics were: i) publication in the Journal of Fish Biology; ii) Brazilian and American researchers as the main authors; iii) three authors per article; iv) marine environment as more studied; v) population, the most discussed subject; vi) frequency of occurrence, the most used method for analysis of stomach and intestinal contents; vii) index of relative importance, the method most used to evaluate the importance of food items in the diet; viii) the use of the Kolmogorov-Smirnov and Levene statistical tests to evaluate the assumptions of normality and homogeneity of variances, respectively; ix) the use of the univariate analysis of variance hypotheses, the Kruskall-Wallis non-parametric univariate test, the multivariate analysis of permutational variance analysis, and the multivariate analysis of similarity tests; and x) the use of parametric ordering techniques cluster analysis and non-parametric non-metric multidimensional scaling. / Diversos procedimentos laboratoriais e métodos estatísticos foram desenvolvidos para avaliar a dieta em peixes. Desse modo, por meio de uma análise cienciométrica identificamos tendências e padrões nos trabalhos sobre ecologia trófica de peixes, desde aspectos mais gerais até aspectos relacionados ao uso de ferramentas estatísticas. A base de dados utilizada para o levantamento foi a Thomson Reuters (ISI Web of Knowledge, apps.webofknowledge.com). Na qual verificamos aumento significativamente crescente na quantidade de artigos sobre ecologia trófica de peixes com o tempo. Dentre os 1319 artigos avaliados, as características mais frequentes foram: i) publicação na revista Journal of Fish Biology; ii) pesquisadores brasileiros e norte-americanos como autores principais; iii) três autores por trabalho; iv) ambiente marinho; v) população, o assunto mais abordado; vi) frequência de ocorrência, o método mais utilizado para análise do conteúdo estomacal e intestinal; vii) índice de importância relativa, método mais utilizado para avaliar a importância dos itens alimentares na dieta; viii) a utilização dos testes estatísticos Kolmogorov-Smirnov e Levene para avaliar os pressupostos de normalidade e homogeneidade de variâncias, respectivamente; ix) a utilização dos testes de hipóteses univariado paramétrico de análise de variância, do teste univaridado não-paramétrico de Kruskall-Wallis, do teste multivariado de análise de variância permutacional e do teste multivariado de análise de similaridades; e x) a utilização das técnicas de ordenação paramétrica análise de agrupamento e a não-paramétrica de escalonamento multidimensional não métrico.

The content of this thesis is the design and statistical tests of two di erent hardware random number generators. It also includes an overview of the sources of entropy, algorithms used to correct deviations from the normal distribution and the description of statistical tests.

L’apport disproportionné aux taux de criminalité des membres des gangs de rue est, nul doute, une proposition empirique robuste. De nombreuses études ont conclu que l’association aux gangs de rue est un facteur de risque supplémentaire à celui que constitue déjà la fréquentation de pairs délinquants au nombre des meilleurs prédicteurs de la délinquance avec les antécédents criminels et les traits antisociaux de la personnalité. Pourtant, la contribution spécifique de l’association aux gangs de rue à l’explication de la délinquance est largement méconnue. Au nombre des variables les plus souvent citées pour l’expliquer figure néanmoins le concept de l’adhésion à la culture de gang qui n’a toutefois jamais été spécifiquement opérationnalisé. Le but de la thèse est d’étudier la contribution spécifique de l’adhésion d’un contrevenant à la culture des gangs de rue à l’explication de la délinquance. Plus précisément, elle a comme objectifs de définir la culture des gangs de rue, d’opérationnaliser l’adhésion à la culture des gangs de rue, d’examiner la fidélité de la mesure de l’adhésion à la culture de gang et d’étudier sa relation avec la nature, la variété et la fréquence des conduites délinquantes de contrevenants placés sous la responsabilité des centres jeunesse et des services correctionnels du Québec. Trois articles scientifiques, auxquels un chapitre régulier est joint, ont servi la démonstration de la thèse. D’abord, le premier article présente les démarches relatives au développement de la première Mesure de l’adhésion à la culture de gang, la MACg. Plus précisément, l’article présente la recension des écrits qui a permis de proposer une première définition de la culture de gang et d’opérationnaliser le concept. Il fait aussi état de la démarche de la validation de la pertinence de son contenu et des données préliminaires qui révèlent la très bonne cohérence interne de la MACg. Cette première étude est suivie de la présentation, dans le cadre d’un chapitre régulier, des résultats de l’examen de la cotation des principaux indicateurs de la culture de gang. Cette démarche constitue un complément nécessaire à l’examen de la validité apparente de la MACg. Les résultats révèlent des degrés de concordance très satisfaisants entre les observations de divers professionnels des centres jeunesse et des services correctionnels du Québec qui ont été invités à coter les indicateurs de la culture de gang à partir de deux histoires fictives d’un contrevenant mineur et d’un second d’âge adulte. Puis, le deuxième article présente les résultats d’un premier examen de la fidélité de la MACg à l’aide du modèle de Rasch de la Théorie de la réponse aux items. Ses résultats soutiennent l’unidimensionnalité de la MACg et sa capacité à distinguer des groupes d’items et de personnes le long d’un continuum de gravité d’adhésion à la culture de gang. Par contre, le fonctionnement différentiel et le mauvais ajustement de certains items sont observés, ainsi que l’inadéquation de la structure de réponses aux items (de type Likert) privilégiée lors de l’élaboration de la MACg. Une version réaménagée de cette dernière est donc proposée. Enfin, le troisième et dernier article présente les résultats de l’examen de la relation entre la délinquance et l’adhésion d’un contrevenant à la culture de gang telle que mesurée par la MACg. Les résultats soutiennent l’apport unique de l’adhésion d’un contrevenant à la culture de gang à la diversité et à la fréquence des conduites délinquantes auto-rapportées par des contrevenants placés sous la responsabilité des centres jeunesse et des services correctionnels du Québec. Le score à l’échelle originale et réaménagée de la MACg s’avère, d’ailleurs, un facteur explicatif plus puissant que l’âge, la précocité criminelle, les pairs délinquants et la psychopathie au nombre des meilleurs prédicteurs de la délinquance. L’étude met aussi en lumière l’étroite relation entre une forte adhésion à la culture de gang et la présence marquée de traits psychopathiques annonciatrice de problèmes particulièrement sérieux. Malgré ses limites, la thèse contribuera significativement aux développements des bases d’un nouveau modèle explicatif de l’influence de l’association aux gangs de rue sur les conduites des personnes. La MACg pourra aussi servir à l’évaluation des risques des hommes contrevenants placés sous la responsabilité du système de justice pénale et à l’amélioration de la qualité des interventions qui leur sont dédiées. / The disproportionate contribution of street gang members to a city's crime rate is, no doubt, a strong empirical proposition. Numerous studies have concluded that street gang association is an additional risk factor tied to the association with delinquent peers which constitutes among the best predictors of delinquency along with criminal history and antisocial personality traits. However, the specific role that street gang association plays in explaining delinquency is largely unknown. Nevertheless, among the variables most often cited to explain this is the concept of adherence to the gang culture which has however never been specifically operationalized. The aim of this thesis is to study the specific contribution of an offender's adherence to the street gang culture to an explanation of delinquency. More specifically, its objectives are to define the street gang culture, operationalize adherence to the street gang culture, examine the reliability of the measurement of adherence to the street gang culture, and study its relationship with the nature, variety and frequency of delinquent behaviour of offenders placed under the responsibility of youth centres and correctional services in Quebec. Three scientific papers, to which a regular chapter has been attached, were used to demonstrate the thesis. The first paper presents the steps taken toward the development of the first Measurement of Adherence to the Gang Culture, the MACg. More specifically, the paper reviews the literature that led us to propose an initial definition of the gang culture and operationalize the concept. It also describes the approach to validating the relevance of its content and the preliminary data that demonstrate the measurement's strong internal consistency. This first study is followed by an examination of scoring of the main indicators of the gang culture, presented as a regular chapter. This approach is a necessary complement to the examination of the face validity of the MACg. The results indicate very satisfying levels agreement between observations of the various Quebec youth centre and correctional services professionals who were invited to score indicators of the gang culture based on two fictional stories of a juvenile offender and an adult offender.   Then, the second paper presents results of an initial study of the reliability of the MACg using the Rasch model approach to item response theory. Its findings support the one-dimensional nature of the MACg and its ability to distinguish groups of items and persons along a severity continuum of adherence to the gang culture. However, the differential functioning and poor fit of certain items were observed, in addition to the inadequacy of the type of response structure (Likert scale) chosen during the development of the MACg. Therefore, an adjusted version of the measurement is proposed. The third and final paper presents the results of a study examining the relationship between delinquency and an offender's adherence to the gang culture as measured by the MACg. The results support the unique contribution of an offender's adherence to the gang culture to the diversity and frequency of self-reported delinquent behaviour by offenders placed under the responsibility of youth centres and correctional services in Quebec. Moreover, the score on the original and adjusted MACg scales turned out to be a more powerful explanatory factor than age, criminal precocity, delinquent peers and psychopathy when it came to the best predictors of delinquency. The study also highlights the close relationship between a strong adherence to the gang culture and the marked presence of psychopathic traits which were warning signs of particularly serious problems. Despite its limitations, this thesis will significantly contribute to developing the foundation of a new model to explain the influence of street gang association on human behaviour. The MACg can also aid the risk assessment of male offenders placed under the responsibility of the criminal justice system and improve the quality of the treatments available to them.

The topic of this diploma thesis is Infectious diseases and migration. Its aim is to assess the distribution of selected infectious diseases in the world and in Europe, with a focus on Czech context. While from the literature it can be concluded that the spread of serious infectious diseases is conditioned by migration, based on the available data it cannot be confirmed that immigration currently constitutes a significant threat for Czech citizens.

Após o ajustamento de observações pelo método dos mínimos quadrados (MMQ) ter sido realizado, é possível a detecção e a identificação de erros não aleatórios nas observações, por meio de testes estatísticos. A teoria da confiabilidade faz uso de medidas adequadas para quantificar o menor erro detectável em uma observação, e a sua influência sobre os parâmetros ajustados, quando não detectado. A teoria de confiabilidade convencional foi desenvolvida para os procedimentos de teste convencionais, como o data snooping, que pressupõem que apenas uma observação está contaminada por erros grosseiros por vez. Recentemente foram desenvolvidas medidas de confiabilidade generalizadas, relativas a testes estatísticos que pressupõem a existência, simultânea, de múltiplas observações com erros (outliers). Outras abordagens para o controle de qualidade do ajustamento, alternativas a estes testes estatísticos, também foram propostas recentemente, como por exemplo, o método QUAD (Quasi-Accurate Detection of outliers method). Esta pesquisa tem por objetivo fazer um estudo sobre o controle de qualidade do ajustamento de observações geodésicas, por meio de experimentos em uma rede GPS (Global Positioning System), utilizando tanto os métodos convencionais quanto o atual estado da arte. Desta forma, foram feitos estudos comparativos entre medidas de confiabilidade convencionais e medidas de confiabilidade generalizadas para dois outliers simultâneos, bem como estudos comparativos entre o procedimento data snooping e testes estatísticos para a identificação de múltiplos outliers. Também se investigou como a questão das variâncias e covariâncias das observações, bem como a geometria/configuração da rede GPS em estudo, podem influenciar nas medidas de confiabilidade, tanto na abordagem convencional, quanto na abordagem generalizada. Por fim, foi feito um estudo comparativo entre o método QUAD e os testes estatísticos para a identificação de erros. / After the adjustment of observations has been carried out by Least Squares Method (LSM), it is possible to detect and identify non-random errors in the observations using statistical tests. The reliability theory makes use of appropriate measures to quantify the minimal detectable bias (error) in an observation, and its influence on the adjusted parameters, if not detected. The conventional reliability theory has been developed for conventional testing procedures such as data snooping, which assumes that only one observation is contaminated by errors at a time. Recently, generalized measures of reliability has been developed, relating to statistical tests that assumes the existence, simultaneous, of multiple observations with errors (outliers). Other approaches to the quality control of the adjustment, alternatives to these statistical tests, were also proposed recently, such as the QUAD method (Quasi-Accurate Detection of outliers method). The goal of this research is to make a study about the quality control of the adjustment of geodetic observations, by means of experiments in a GPS (Global Positioning System) network, using both conventional methods and the current state of the art. In this way, comparisons were made between conventional reliability measures and generalized measures of reliability for two outliers, as well as comparisons between the data snooping procedure and statistical tests to identify multiple outliers. It was also investigated how the variances and covariances of the observations, as well as the geometry/configuration of the GPS network in study, can influence the measures of reliability, both in the conventional approach and in the generalized approach. Finally, a comparison was made between the QUAD method and the statistical tests to identify outliers (errors).

Submitted by Jorge Silva (jorgelmsilva@ufrrj.br) on 2017-08-16T19:52:28Z No. of bitstreams: 1 2017 - Givanildo de Gois.pdf: 10268407 bytes, checksum: 6b0d9ed5afee22baa701ac9aa599b1e9 (MD5) / Made available in DSpace on 2017-08-16T19:52:28Z (GMT). No. of bitstreams: 1 2017 - Givanildo de Gois.pdf: 10268407 bytes, checksum: 6b0d9ed5afee22baa701ac9aa599b1e9 (MD5) Previous issue date: 2017-02-22 / CAPES / Few studies were based on the areas of Physical and Statistical Climatology applied to the state of Rio de Janeiro (ERJ), focused on natural disasters (droughts, floods and desertification) based on climatic data and orbital products. The time series used were 47 years (1967 to 2013) and 71 years (1943 to 2013). Both series come from 100 existing stations belonging to ANA, CPRM, INMET, SERLA and LIGHT. The temporal series (raw data) were faulty and were filled with TRMM satellite 3B43 product (1998 to 2013) and INMET climatological norm (1947 to 1997). The series were submitted to descriptive, exploratory, parametric (Shapiro-Wilks-SW and Barlett-B), non-parametric tests (Mann-Kendall-MK, Sen-Se, Pettitt and SOCUM), Box Cox transformation and analysis Grouping (AA). In addition, monthly data from the Enhanced Vegetation Index (EVI2) between 2001- 2012 with the objective of verifying the trend of increase and decrease of vegetation in the ERJ by non-parametric tests and future scenarios by the Markov Chain. Estimation of the monthly mean air temperature (Tmi) in the ERJ based on observed series and reanalyses or the composition of both, and being adjusted to the three-base linear multiple regression (RLM) and linear regression models: Bases 1 Reanalysis / NCEP), 2 (INMET / NCDC) and 3 (Reanalysis / NCEP and INMET / NCDC) between 1948 and 2015. Descriptive analysis showed a probability of occurrence above 75%. The SW and B tests presented a low significance level for p-value (? < 5%) and the hypothesis of normality and homogeneity of variances in the stations was rejected. The Box Cox transformation was effective in stabilizing the residue normality and homogeneity of variance of the monthly rainfall series of the Middle Para?ba and Serrana regions, except for the Northwest Fluminense. The high variability of ? (0.326 to 0.565) is due to the fact that most of the stations are in the Sierra do Mar slope facing the mainland, where rainfall is influenced by the interaction of topography with local and synoptic systems, only one season In the Serra do Mar slope facing the Atlantic Ocean with influence of the coastal environment and the mesoscale and synoptic systems, in the series of 71 years. Based on AA were chosen 11 stations with normality or homogeneity of variance characterized two homogeneous pluviometric groups (G1 and G2) in the ERJ. For the trend analysis, the MK test and method were shown the presence of non significant trends of rainfall increase in the annual and seasonal scales. Pettitt and SOCUM were efficient in identifying the years of possible non-significant or insignificant abrupt changes in the 71-year time series. The SOCUM test identified 39 ENSO events in groups G1 and G2. The highest percentage in the neutral events (48.72%) and the lowest in the moderate El Ni?o and La Ni?a weak and strong (5.13%). An insignificant trend of vegetation growth is observed at 75%, followed by a significant downward trend of 25% of the ERJ Government regions. The Pettitt test showed the existence of abrupt changes not significant (NS), both growth and vegetation decrease in 6 regions and significant (S) decrease in 2 regions. The predictions of changes ranging from 1 to 2 years at constant intervals (3 to 10 years) were observed in all future scenarios. Bases 1 and 2 presented the highest number of significant coefficients, according to the F test for (p-value <0.05), the exception was Base 3. The latitude variable (?1) was more significant, followed by altitude (?3 ) In all Bases. Significant values of r2 (> 0.80) and r (> 0.90) in Base 2 and Base 1 with r2 (> 0.50) and r (> 0.70). The adjusted RLM models explained most of the spatial variability of Tmi for the ERJ. The parametric tests of SW and B applied to the monthly rainfall series without treatment and to the reduced variable the normal distribution standardized to 95% of probability point to the hypotheses of non-normality and neither homogeneity of the time series. The high sensitivity of the rainfall series to the B test were observed in the eight Government regions, due to the rigor of the test. The lambda coefficients of the Box Cox transformation applied to the monthly rainfall series for data without treatment and the reduced variable of the standardized normal distribution do not present efficiency in the stabilization of the homogeneity of the variances. Confirmed by the test of B, in 99.58% and 100% of the events. The efficiency found only in the stabilization of normality in 81.33% and 81.58% of the monthly cumulative frequencies of data without treatment and the reduced variable. Moderate performance of SPI methods with untreated and Box Cox versus SPI-transformed data with reduced-box data transformed by Box Cox is evident in SPI-1, which shows the presence of significant variations of statistical parameters in the Norte, Costa Verde, Baixada Litor?nea e Metropolitana shortage, followed by low performance of the r2 coefficient in the ERJ regions. SPI-12 shows a significant high dispersion of the coefficient r, followed by a low to very low performance, and a low coefficient r2. This indicates poor accuracy of SPI index estimates in both methods. The EPE and RMSE errors do not present significant variations, in the durations of 1 and 12 months. A high variation of the rec coefficients with the index d in the SPI-1 month is verified, a poor performance of the methods with data without treatment and with transformed by Box Cox versus SPI with data of the reduced variable transformed by Box Cox, for the SPI-12 Was verified in the ERJ regions. The temporal / annual analyzes of SPI-1 and 12 in the regions show a high variability and greater intensity of SPI-1, unlike SPI-12. SPI-1 and SPI-12 in the regions show similarity in the behavior of SPI-1 and SPI-12, where the highest and lowest frequencies of droughts categorized as moderately, extremely and extremely dry were recorded in the 70, 80, 90, 2000 and in the period 2010/2013, except for the 60. Events of ENOS were observed in the study period. The Pettitt test identified the years of changes in the SPI-12 index, in 1977 (El Ni?o weak), 1984 (La Ni?a weak), 1989 (Neutral), 1992 (Neutral) and 2002 (Moderate El Ni?o). The prevailing category was close to normal in the Norte Fluminense, Baixadas Litor?neas and Costa Verde regions, followed in the other regions of Government in some parts (SW), (SSW) (SE) and (NE). The moderately dry category occurred in the regions, Metropolitana, Centro Sul Fluminense, M?dio Para?ba, Serrana and NoroesteFluminense, and the other parts in the (SW), (NW) and (NNE) portions of the ERJ. In short, the application of statistical, parametric and non-parametric tests, chain of markov, multivariate analysis are efficient tools in the evaluation of natural disasters in the ERJ. / H? poucos estudos baseados nas ?reas de Climatologia F?sica e Estat?stica aplicadas ao estado do Rio de Janeiro (ERJ), voltados para cat?strofes naturais (secas, enchentes e desertifica??o) baseado em dados clim?ticos e produtos orbitais. As s?ries temporais usadas foram de 47 anos (1967 a 2013) e 71 anos (1943 a 2013). oriundas de 100 esta??es pertencentes ? ANA, CPRM, INMET, SERLA e LIGHT. Estas s?ries (dados brutos) apresentavam falhas e foram preenchidas com produto 3B43 do sat?lite TRMM (1998 a 2013) e com as normais climatol?gicas do INMET (1947 a 1997). Para tanto elas foram submetidas ? an?lise descritiva, explorat?ria, testes param?tricos (Shapiro-Wilks-SW e Barlett-B), n?o param?tricos (Mann-Kendall-MK, M?todo de Sen -Se, Pettitt e SOCUM), transforma??o Box Cox e an?lise de agrupamento (AA). Al?m disso, foram usados dados mensais do Enhanced Vegetation Index (EVI2) entre 2001-2012 com objetivo de verificar tend?ncia de aumento e diminui??o da vegeta??o no ERJ pelos testes n?o param?tricos e os cen?rios futuros pela Cadeia de Markov. A estimativa da temperatura m?dia mensal do ar (Tmi) no ERJ, foi baseada em s?ries observadas de rean?lises ou atrav?s da composi??o de ambas e, sendo ajustadas aos modelos de regress?o linear m?ltipla (RLM) e simples (RLS) baseado em tr?s bases: Bases 1 (Rean?lise/NCEP), 2 (INMET/NCDC) e 3 (Rean?lise/NCEP e INMET/NCDC) entre 1948 a 2015. A an?lise descritiva mostrou uma probabilidade de ocorr?ncia acima de 75%, os testes SW e B apresentaram um baixo n?vel de signific?ncia para p-valor (? < 5%) e rejeitou-se a hip?tese de normalidade e homogeneidade de vari?ncias nas esta??es. A transforma??o Box Cox foi eficaz na estabiliza??o da normalidade dos res?duos e homogeneidade de vari?ncia da s?rie temporal de chuva mensal das regi?es M?dio Para?ba e Serrana, com exce??o do Noroeste Fluminense. A alta variabilidade de ? (0,326 a 0,565) ? devido ? maioria das esta??es encontram-se na vertente da Serra do Mar voltada para o continente, onde o regime de chuva ? influenciado pela intera??o da topografia com sistemas locais e sin?ticos e tendo apenas uma esta??o na vertente da Serra do Mar voltada para o Oceano Atl?ntico com influ?ncia do ambiente costeiro e dos sistemas de mesoescala e sin?ticos, na s?rie de 71 anos. Com base na AA foram escolhidas 11 esta??es com normalidade ou homogeneidade de vari?ncia, caracterizando dois grupos homog?neos pluviom?tricos (G1 e G2) no ERJ. Para a an?lise de tend?ncia, o teste MK e m?todo Se mostraram a presen?a de tend?ncias n?o significativas de aumento das chuvas nas escalas anual e sazonal, enquanto que o Pettitt e o SOCUM foram eficientes quanto ? identifica??o dos anos de poss?veis mudan?as abruptas n?o significativas ou insignificantes na s?rie temporal de 71 anos. O teste de SOCUM identificou 39 eventos de ENOS nos grupos G1 e G2, os maiores percentuais nos eventos neutros (48,72%) e os menores nos El Ni?o moderado e La Ni?a fraca e forte (5,13%). Outro resultado encotrado foi a exist?ncia de uma tend?ncia insignificante de crescimento da vegeta??o em 75%, seguido de uma tend?ncia significativa de diminui??o em 25% das regi?es pol?ticas do ERJ. J? o teste de Pettitt mostrou a exist?ncia de mudan?as bruscas n?o significativas (NS), ambos de crescimento e diminui??o da vegeta??o em 6 regi?es e significativas (S) de diminui??o em 2 regi?es. Os progn?sticos de mudan?as com varia??o de 1 a 2 anos em intervalos constante (3 a 10 anos) foram observados em todos os cen?rios futuros. As Bases 1 e 2 apresentaram o maior n?mero de coeficientes significativos, segundo O teste F para p-valor < 0,05, com exce??o para a Base 3. A vari?vel latitude (?1) foi mais significativa, seguido da altitude (?3) em todas as Bases. Foram encontrados valores significativos de r2 (>0,80) e r (> 0,90) na Base 2 e na Base 1 com r2 (>0,50) e r (>0,70). Os modelos de RLM ajustados explicaram a maior parte da variabilidade espacial da Tmi para o ERJ, enquanto que os testes param?tricos de SW e B aplicados a s?rie temporal mensal de chuva sem tratamento e ? vari?vel reduzida a distribui??o normal padronizada a 95% de probabilidade apontaram para as hip?teses de n?o-normalidade e n?o-homogeneidade da s?rie temporal. A alta sensibilidade da s?rie temporal de chuva ao teste B foram constatada nas oito regi?es pol?ticas do ERJ, devido ao rigor do teste. Os coeficientes de lambda da transforma??o Box Cox aplicada ?s s?ries temporais mensais de chuva para dados sem tratamento e a vari?vel reduzida da distribui??o normal padronizada n?o apresentam efici?ncia na estabiliza??o da homogeneidade das vari?ncias. Confirmado pelo teste de B, em 99,58% e 100% dos eventos repetitivamente. A efici?ncia constatada apena na estabiliza??o da normalidade em 81,33% e 81,58% das frequ?ncias acumuladas mensais dos dados sem tratamento e da vari?vel reduzida. Al?m disso observa-se que o desempenho moderado dos m?todos do SPI com dados sem tratamento e com os transformados pela Box Cox versus SPI com dados da vari?vel reduzida transformada pela Box Cox fica evidente no SPI-1, que mostra a presen?a de varia??es significativas dos par?metros estat?sticos nas regi?es Norte, Costa Verde, Baixada Litor?nea e Metropolitana, seguidos de baixo desempenho do coeficiente r2 nas regi?es do ERJ. J? o SPI-12 mostrou uma alta dispers?o significativa do coeficiente r, seguido de um desempenho baixo a muito baixo, e baixos valores do coeficiente r2, indicando fraca precis?o das estimativas dos ?ndices SPI em ambos os m?todos. Os erros EPE e RMSE n?o apresentaram varia??es significativas, nas dura??es de 1 e 12 meses. Contudo costatase uma alta varia??o dos coeficientes r e c com o ?ndice d no SPI-1 m?s, ressaltasse que um p?ssimo desempenho dos m?todos com dados sem tratamento e com transformados pela Box Cox versus SPI com dados da vari?vel reduzida transformada pela Box Cox, para o SPI-12 foi verificado nas regi?es do ERJ. Al?m disso, as an?lises temporal/anual dos SPI-1 e 12 nas regi?es mostra alta variabilidade e maior intensidade do SPI-1, ao contr?rio do SPI-12. No tocante a an?lise temporal dos SPI?1 e SPI?12 nas regi?es do ERJ verifica-se similaridade quanto ao comportamento dos SPI?1 e SPI?12, onde as maiores e menores frequ?ncias de eventos de secas categorizadas como moderadamente, muito e extremamente seco foram registradas nas d?cadas 1970, 1980, 1990, 2000 e no per?odo 2010/2013, com exce??o da d?cada de 1960. Eventos de ENOS foram observados no per?odo de estudo. O teste de Pettitt identificaram os anos de mudan?as do ?ndice SPI-12, em 1977 (El Ni?o fraco), 1984 (La Ni?a fraca), 1989 (Neutro), 1992 (Neutro) e 2002 (El Ni?o moderado). Prevaleceu a categoria pr?ximo ao normal nas regi?es Norte Fluminense, Baixadas Litor?neas e Costa Verde, seguido nas demais regi?es de Governo em algumas por??es (SW), (SSW) (SE) e (NE). A categoria moderadamente seca ocorreu nas regi?es, Metropolitana, Centro Sul Fluminense, M?dio Para?ba, Serrana e Noroeste Fluminense enas demais nas por??es (SW), (NW) e (NNE) do ERJ. Em suma, a aplica??o dos testes estat?sticos, param?tricos e n?o-param?tricos, cadeia de markov, an?lise multivariada s?o ferramentas eficientes na avalia??o das cat?strofes naturais no ERJ.

The field of machine learning studies algorithms that infer predictive models from data. Predictive models are applicable for many practical tasks such as spam filtering, face and handwritten digit recognition, and personalized product recommendation. In general, they are used to predict a target label for a given data instance. In order to make an informed decision about the deployment of a predictive model, it is crucial to know the model’s approximate performance. To evaluate performance, a set of labeled test instances is required that is drawn from the distribution the model will be exposed to at application time. In many practical scenarios, unlabeled test instances are readily available, but the process of labeling them can be a time- and cost-intensive task and may involve a human expert. This thesis addresses the problem of evaluating a given predictive model accurately with minimal labeling effort. We study an active model evaluation process that selects certain instances of the data according to an instrumental sampling distribution and queries their labels. We derive sampling distributions that minimize estimation error with respect to different performance measures such as error rate, mean squared error, and F-measures. An analysis of the distribution that governs the estimator leads to confidence intervals, which indicate how precise the error estimation is. Labeling costs may vary across different instances depending on certain characteristics of the data. For instance, documents differ in their length, comprehensibility, and technical requirements; these attributes affect the time a human labeler needs to judge relevance or to assign topics. To address this, the sampling distribution is extended to incorporate instance-specific costs. We empirically study conditions under which the active evaluation processes are more accurate than a standard estimate that draws equally many instances from the test distribution. We also address the problem of comparing the risks of two predictive models. The standard approach would be to draw instances according to the test distribution, label the selected instances, and apply statistical tests to identify significant differences. Drawing instances according to an instrumental distribution affects the power of a statistical test. We derive a sampling procedure that maximizes test power when used to select instances, and thereby minimizes the likelihood of choosing the inferior model. Furthermore, we investigate the task of comparing several alternative models; the objective of an evaluation could be to rank the models according to the risk that they incur or to identify the model with lowest risk. An experimental study shows that the active procedure leads to higher test power than the standard test in many application domains. Finally, we study the problem of evaluating the performance of ranking functions, which are used for example for web search. In practice, ranking performance is estimated by applying a given ranking model to a representative set of test queries and manually assessing the relevance of all retrieved items for each query. We apply the concepts of active evaluation and active comparison to ranking functions and derive optimal sampling distributions for the commonly used performance measures Discounted Cumulative Gain and Expected Reciprocal Rank. Experiments on web search engine data illustrate significant reductions in labeling costs. / Maschinelles Lernen befasst sich mit Algorithmen zur Inferenz von Vorhersagemodelle aus komplexen Daten. Vorhersagemodelle sind Funktionen, die einer Eingabe – wie zum Beispiel dem Text einer E-Mail – ein anwendungsspezifisches Zielattribut – wie „Spam“ oder „Nicht-Spam“ – zuweisen. Sie finden Anwendung beim Filtern von Spam-Nachrichten, bei der Text- und Gesichtserkennung oder auch bei der personalisierten Empfehlung von Produkten. Um ein Modell in der Praxis einzusetzen, ist es notwendig, die Vorhersagequalität bezüglich der zukünftigen Anwendung zu schätzen. Für diese Evaluierung werden Instanzen des Eingaberaums benötigt, für die das zugehörige Zielattribut bekannt ist. Instanzen, wie E-Mails, Bilder oder das protokollierte Nutzerverhalten von Kunden, stehen häufig in großem Umfang zur Verfügung. Die Bestimmung der zugehörigen Zielattribute ist jedoch ein manueller Prozess, der kosten- und zeitaufwendig sein kann und mitunter spezielles Fachwissen erfordert. Ziel dieser Arbeit ist die genaue Schätzung der Vorhersagequalität eines gegebenen Modells mit einer minimalen Anzahl von Testinstanzen. Wir untersuchen aktive Evaluierungsprozesse, die mit Hilfe einer Wahrscheinlichkeitsverteilung Instanzen auswählen, für die das Zielattribut bestimmt wird. Die Vorhersagequalität kann anhand verschiedener Kriterien, wie der Fehlerrate, des mittleren quadratischen Verlusts oder des F-measures, bemessen werden. Wir leiten die Wahrscheinlichkeitsverteilungen her, die den Schätzfehler bezüglich eines gegebenen Maßes minimieren. Der verbleibende Schätzfehler lässt sich anhand von Konfidenzintervallen quantifizieren, die sich aus der Verteilung des Schätzers ergeben. In vielen Anwendungen bestimmen individuelle Eigenschaften der Instanzen die Kosten, die für die Bestimmung des Zielattributs anfallen. So unterscheiden sich Dokumente beispielsweise in der Textlänge und dem technischen Anspruch. Diese Eigenschaften beeinflussen die Zeit, die benötigt wird, mögliche Zielattribute wie das Thema oder die Relevanz zuzuweisen. Wir leiten unter Beachtung dieser instanzspezifischen Unterschiede die optimale Verteilung her. Die entwickelten Evaluierungsmethoden werden auf verschiedenen Datensätzen untersucht. Wir analysieren in diesem Zusammenhang Bedingungen, unter denen die aktive Evaluierung genauere Schätzungen liefert als der Standardansatz, bei dem Instanzen zufällig aus der Testverteilung gezogen werden. Eine verwandte Problemstellung ist der Vergleich von zwei Modellen. Um festzustellen, welches Modell in der Praxis eine höhere Vorhersagequalität aufweist, wird eine Menge von Testinstanzen ausgewählt und das zugehörige Zielattribut bestimmt. Ein anschließender statistischer Test erlaubt Aussagen über die Signifikanz der beobachteten Unterschiede. Die Teststärke hängt von der Verteilung ab, nach der die Instanzen ausgewählt wurden. Wir bestimmen die Verteilung, die die Teststärke maximiert und damit die Wahrscheinlichkeit minimiert, sich für das schlechtere Modell zu entscheiden. Des Weiteren geben wir eine Möglichkeit an, den entwickelten Ansatz für den Vergleich von mehreren Modellen zu verwenden. Wir zeigen empirisch, dass die aktive Evaluierungsmethode im Vergleich zur zufälligen Auswahl von Testinstanzen in vielen Anwendungen eine höhere Teststärke aufweist. Im letzten Teil der Arbeit werden das Konzept der aktiven Evaluierung und das des aktiven Modellvergleichs auf Rankingprobleme angewendet. Wir leiten die optimalen Verteilungen für das Schätzen der Qualitätsmaße Discounted Cumulative Gain und Expected Reciprocal Rank her. Eine empirische Studie zur Evaluierung von Suchmaschinen zeigt, dass die neu entwickelten Verfahren signifikant genauere Schätzungen der Rankingqualität liefern als die untersuchten Referenzverfahren.

Daktaro disertacijos tyrimo objektai yra rezervuotos sistemos ir degradaciniai modeliai. Norint užtikrinti svarbių sistemos elementų aukštą patikimumą, naudojami jų rezerviniai elementai, kurie gali būti įjungiami sugedus šiems pagrindiniams elementams. Rezerviniai elementai gali funkcionuoti skirtinguose režimuose: „karštame“, „šaltame“ arba „šiltame“. Disertacijoje yra nagrinėjamos sistemos su „šiltai“ rezervuotais elementais. Darbe suformuluojama rezervinio elemento „sklandaus įjungimo“ hipotezė ir konstruojami statistiniai kriterijai šiai hipotezei tikrinti. Nagrinėjami neparametrinio ir parametrinio taškinio bei intervalinio vertinimo uždaviniai. Disertacijoje nagrinėjami pakankamai bendri degradacijos modeliai, kurie aprašo elementų gedimų intensyvumą kaip funkciją kiek naudojamų apkrovų, tiek ir degradacijos lygio, kuri savo ruožtu modeliuojama naudojant stochastinius procesus. / In doctoral thesis redundant systems and degradation models are considered. To ensure high reliability of important elements of the system, the stand-by units can be used. These units are commuted and operate instead of the main failed unit. The stand-by units can function in the different conditions: “hot”, “cold” or “warm” reserving. In the thesis systems with “warm” stand-by units are analyzed. Hypotheses of smooth commuting are formulated and goodness-of-fit tests for these hypotheses are constructed. Nonparametric and parametric point and interval estimation procedures are given. Modeling and statistical estimation of reliability of systems from failure time and degradation data are considered.