La génomique comparative étudie les similarités et/ou les dissimilarités entre génomes et permet d'établir des relations entre les espèces afin notamment de construire des phylogénies. Elle permet également de mettre en évidence des régions conservées au sein des génomes et de trouver ainsi des ensembles de gènes impliqués dans des processus biologiques conservés au cours de l'évolution. Dans ce mémoire, nous nous intéressons au calcul de mesures entre deux génomes en présence de gènes dupliqués, et plus particulièrement aux mesures à base de points de cassure, d'adjacences, d'intervalles communs et d'intervalles conservés. Suivant une démarche informatique, nous proposons tout d'abord une étude avancée de la complexité algorithmique des problèmes rencontrés, en prouvant notamment pour la plupart d'entre eux soit leur NP-Complétude soit leur APX-Difficulté. Par la suite, nous exposons plusieurs méthodes de calcul de mesures entre deux génomes, à savoir (i) une approche exacte basée sur une transformation en un problème de contraintes à variables booléennes, (ii) une heuristique et (iii) une méthode hybride qui s'appuie sur la méthode exacte et l'heuristique proposées. Par une étude sur un jeu de données réel, nous montrons les qualités respectives de ces méthodes. Enfin, nous proposons un protocole de calcul des intervalles communs et mettons en évidence, par son utilisation et par un outil de visualisation, l'aspect fonctionnel de certains intervalles communs.

I report 1684 new proper motion systems in the southern sky (declinations -90 degrees to -47 degrees) with 0.50 arcsec/yr > mu >= 0.18 arcsec/yr. This effort is a continuation of the SuperCOSMOS-RECONS (SCR) proper motion search to lower proper motions than reported in Hambly et al. (2004); Henry et al. (2004); Subasavage et al. (2005a,b). Distance estimates are presented for the new systems, assuming that all stars are on the main sequence. I find that 34 systems are within 25 pc, including three systems --- SCR 0838-5855, SCR 1826-6542, and SCR 0630-7643AB --- anticipated to be within 10 pc. These mini-motion (MINIMO) discoveries constitute a more than ten-fold increase in new systems found in the same region of sky searched for systems with mu >= 0.50 arcsec/yr, suggesting a happy hunting ground for new nearby slower proper motion systems in the region just north (declinations -47 degrees to 0 degrees), much of which has not been rigorously searched during previous efforts.

Waiting for service was often happen in Service setting. The literatures about the consequence of the waiting were strong in certain respects, typically negative and erode the customers¡¦ overall service evaluation, and underdeveloped in other respects. In this article, I consider the influence of the waiting on the customers perceived values. I propose the effect of waiting experiences on the customers perceived values depends on both how far the time lags of distances between purchasing and consuming and the word-of ¡Vmouth information supply. For the purpose, I choose the oversea leisure travel products that appear to have a significant time lag of consuming. The sample data from 290 college and graduated students were analyzed using descriptive statistics, T-tests, One-Way ANOVA and Two-Way MANOVA. The results of the three studies are presented as follows: 1. When customers perceived higher the value of the service products, they were more patient about waiting duration and more purchasing intentions. 2. The time lags of distances between purchasing and consuming were longer, the customers perceived values would increase first and then drop late. 3. Word of mouth information supply will enhance the customers perceived value.

Lorsque le processus de génération des données ne respecte pas certains des postulats fondant l'analyse statistique du modèle classique de régression linéaire, les tests de permutations offrent une alternative non paramétrique fiable de construction de tests d'hypothèse libres. La première application de cette méthode d'inférence statistique au modèle de régression linéaire simple renvoie à Fisher (1935) et Pitman (1937a,b,1938). Cette méthode de ré-échantillonnage est fondée sur des postulats moins forts que la méthode paramétrique classique et facilement vérifiables en pratique : l'échangeabilité des observations sous l'hypothèse nulle. Si l'utilisation des tests de permutation fait consensus en régression linéaire simple et pour tester l'adéquation d'un modèle en régression multiple, le problème se complique lorsqu'on souhaite mettre à l'épreuve une hypothèse de nullité d'un coefficient de régression partielle. L'étude des conditions d'échangeabilité n'est plus simple dans ce cas. Il n'est alors plus possible de construire des tests exacts plusieurs propositions de tests sont en concurrence. <br />L'objectif principal de notre travail est la comparaison des tests de permutation adaptés aux hypothèses de nullité d'un coefficient de régression partielle dans un modèle linéaire à p variables explicatives, conditionnellement à l'observation d'un échantillon. Quatre méthodes sont comparées, d'une part en recourant à des simulations effectuées dans le cas d'une régression double, puis théoriquement, afin de déterminer les propriétés de biais, de couverture et de puissance de ces tests. Les résultats obtenus sont ensuite étendus au cas de la régression linéaire multiple.<br />Un dernier chapitre complète cette étude en traitant le problème de test de la dépendance partielle entre tableaux de distances interpoints. Nous avons comparé les adaptations des quatre méthodes de test de permutation à ce contexte marqué par la dépendance existant entre éléments d'une matrice de distance et nous avons obtenu dans ce cas des résultats tout à fait différents de ceux qui caractérisent.

This dissertation presents the results of a study to (1) determine the census of the nearby southern M dwarf stellar population via three types of distances and (2) determine the multiplicity rate of nearby M dwarfs using two different search methodologies. The first part of this work reports three types of distance calculations (photographic, photometric, and trigonometric) for 1748 southern M dwarfs. Distances were estimated for 500 red dwarfs using photographic plate BRI magnitudes from SuperCOSMOS, while estimates were made for 667 stars using CCD VRI magnitudes. Both BRI and VRI were combined with 2MASS infrared JHK magnitudes. Distances for an additional 581 red dwarfs were derived from trigonometric parallaxes, 124 of which were measured for the first time during this work. For the second portion of this thesis, an all-sky sample of 1122 M dwarfs, known via trigonometric parallaxes to lie within 25 pc of the Sun, was surveyed for stellar companions at separations of 2" to 600". I-band images using primarily the CTIO/SMARTS 0.9m and the Lowell 42in telescopes were obtained in order to search these systems for companions at separations of 2" to 180". A complementary reconnaissance of wider companions to 600" was also done via blinking SuperCOSMOS BRI images. We find a stellar multiplicity fraction of 27.4 $\pm$ 1.3% for M dwarfs. Using this new gauge of M dwarf multiplicity near the end of the stellar main sequence, we calculate a multiplicity fraction of 30.1% for stellar systems of all types, implying that most systems are single. We find a peak in the separation distribution of the companions at 26 AU, i.e., distances on the scale of our Solar System, with a weak trend of smaller projected separations for lower mass primaries. A hint that M dwarf multiplicity may be a function of age/composition was revealed, with faster moving (and generally older) systems being multiple slightly less often. We calculate that at least 16% of M dwarf mass is made up of the stellar companions of multiple systems. Finally, we show that the mass function for M dwarfs increases to the end of the main sequence.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

The languages spoken in developing countries are diverse and most are currently under-resourced from an automatic speech recognition (ASR) perspective. In South Africa alone, 10 of the 11 official languages belong to this category. Given the potential for future applications of speech-based information systems such as spoken dialog system (SDSs) in these countries, the design of minimal ASR audio corpora is an important research area. Specifically, current ASR systems utilise acoustic models to represent acoustic variability, and effective ASR corpus design aims to optimise the amount of relevant variation within training data while minimising the size of the corpus. Therefore an investigation of the effect that different amounts and types of training data have on these models is needed. With this dissertation specific consideration is given to the data sufficiency principals that apply to the training of acoustic models. The investigation of this task lead to the following main achievements: 1) We define a new stability measurement protocol that provides the capability to view the variability of ASR training data. 2) This protocol allows for the investigation of the effect that various acoustic model complexities and ASR normalisation techniques have on ASR training data requirements. Specific trends with regard to the data requirements for different phone categories and how these are affected by various modelling strategies are observed. 3) Based on this analysis acoustic distances between phones are estimated across language borders, paving the way for further research in cross-language data sharing. Finally the knowledge obtained from these experiments is applied to perform a data sufficiency analysis of a new speech recognition corpus of South African languages: The Lwazi ASR corpus. The findings correlate well with initial phone recognition results and yield insight into the sufficient number of speakers required for the development of minimal telephone ASR corpora. / Thesis (M. Ing. (Computer and Electronical Engineering))--North-West University, Potchefstroom Campus, 2009.

The languages spoken in developing countries are diverse and most are currently under-resourced from an automatic speech recognition (ASR) perspective. In South Africa alone, 10 of the 11 official languages belong to this category. Given the potential for future applications of speech-based information systems such as spoken dialog system (SDSs) in these countries, the design of minimal ASR audio corpora is an important research area. Specifically, current ASR systems utilise acoustic models to represent acoustic variability, and effective ASR corpus design aims to optimise the amount of relevant variation within training data while minimising the size of the corpus. Therefore an investigation of the effect that different amounts and types of training data have on these models is needed. With this dissertation specific consideration is given to the data sufficiency principals that apply to the training of acoustic models. The investigation of this task lead to the following main achievements: 1) We define a new stability measurement protocol that provides the capability to view the variability of ASR training data. 2) This protocol allows for the investigation of the effect that various acoustic model complexities and ASR normalisation techniques have on ASR training data requirements. Specific trends with regard to the data requirements for different phone categories and how these are affected by various modelling strategies are observed. 3) Based on this analysis acoustic distances between phones are estimated across language borders, paving the way for further research in cross-language data sharing. Finally the knowledge obtained from these experiments is applied to perform a data sufficiency analysis of a new speech recognition corpus of South African languages: The Lwazi ASR corpus. The findings correlate well with initial phone recognition results and yield insight into the sufficient number of speakers required for the development of minimal telephone ASR corpora. / Thesis (M. Ing. (Computer and Electronical Engineering))--North-West University, Potchefstroom Campus, 2009.

[Formulae and special characters can only be approximated here. Please see the pdf version of the abstract for an accurate reproduction.] This thesis is a new proposal for analysing spatial point patterns in spatial statistics using the outputs of popular techniques of (classical, non-spatial, multivariate) cluster analysis. The outputs of a chosen hierarchical algorithm, named fusion distances, are applied to investigate important spatial characteristics of a given point pattern. The fusion distances may be regarded as a missing link between the fields of spatial statistics and multivariate cluster analysis. Up to now, these two fields have remained rather separate because of fundamental differences in approach. It is shown that fusion distances are very good at discriminating different types of spatial point patterns. A detailed study on the power of the Monte Carlo test under the null hypothesis of Complete Spatial Randomness (the benchmark of spatial statistics) against chosen alternative models is also conducted. For instance, the test (based on the fusion distance) is very powerful for some arbitrary values of the parameters of the alternative. A new general approach is developed for analysing a given point pattern using several graphical techniques for exploratory data analysis and inference. The new strategy is applied to univariate and multivariate point patterns. A new extension of a popular strategy in spatial statistics, named the analysis of the local configuration, is also developed. This new extension uses the fusion distances, and analyses a localised neighbourhood of a given point of the point pattern. New spatial summary function and statistics, named the fusion distance function H(t), area statistic A, statistic S, and spatial Rg index, are introduced, and proven to be useful tools for identifying relevant features of spatial point patterns. In conclusion, the new methodology using the outputs of hierarchical clustering algorithms can be considered as an essential complement to the existing approaches in spatial statistics literature.

No description available.