L’optique adaptative classique est limitée par l'anisoplanétisme. Pour remédier à cette limitation, de nouveaux concepts, appelés optiques adaptatives grand champ, ont été développés. Ces systèmes analysent la turbulence atmosphérique dans le volume, ce qui accroît le champ de correction. Ces techniques requièrent une connaissance précise du profil de Cn2. Mon étude consiste à valider sur le ciel une nouvelle technique de mesure du profil de Cn2, appelée CO-SLIDAR, à partir des corrélations des mesures de pentes et de scintillation réalisées avec un analyseur Shack-Hartmann sur étoile binaire. Elle s’organise autour de deux grands axes. On réalise d’abord une simulation bout-en-bout de la reconstruction du profil de Cn2 dans un cas concret d’observation astronomique. On peut ainsi étudier l’impact des différentes sources d’erreur sur la reconstruction du profil de Cn2. Ceci nous permet d’améliorer la procédure d’estimation du profil de Cn2, en prenant en compte les bruits de détection. La deuxième partie de mon étude se consacre à la validation expérimentale. On dimensionne et caractérise en laboratoire un banc d’acquisition, le banc ProMeO. Ceci conduit à une bonne connaissance du fonctionnement du banc et nous permet de corriger certains effets instrumentaux. Le banc ProMeO est finalement couplé au télescope MeO de 1,5 m de diamètre. Les données acquises permettent une reconstruction du profil de Cn2, du sol jusqu’à 17 km, avec une résolution de 600 m. Les profils obtenus par la méthode CO-SLIDAR sont comparés avec succès à des profils issus de données météorologiques. L’ensemble de ces travaux constitue la première validation sur le ciel de la méthode CO-SLIDAR. / Classical adaptive optics is limited by anisoplanatism. New concepts, known as Wide Field Adaptive Optics systems, have been developed in order to go beyond this limitation. These systems analyse atmospheric turbulence within a volume, increasing the correction field. These techniques require a precise knowledge of the Cn2 profile. The purpose of my thesis is the on-sky validation of a new measurement method of the Cn2 profile, called CO-SLIDAR, using correlations of slopes and of scintillation, both measured with a Shack-Hartmann on a binary star. My study is organized as follows. First, we perform an end-to-end simulation of the reconstruction of the Cn2 profile in a practical astronomical case. We can thus examine the impact of the different error sources on the reconstruction of the Cn2 profile. This allows us to improve the reconstruction method, taking into account the detection noises. The second part is dedicated to the experimental validation. We design and characterize an acquisition bench, the ProMeO bench. This leads to a good knowledge of the bench's operation and we can compensate for some instrumental effects. The ProMeO bench is then coupled to the MeO 1.5 m telescope. The acquired data allow the estimation of the Cn2 profile, from the ground up to 17 km, with a resolution of 600 m. The CO-SLIDAR profiles are successfully compared with profiles estimated from meteorological data. This work is the first on-sky validation of the CO-SLIDAR method.

L'optique adaptative classique est limitée par l'anisoplanétisme. Pour remédier à cette limitation, de nouveaux concepts, appelés optiques adaptatives grand champ, ont été développés. Ces systèmes analysent la turbulence atmosphérique dans le volume, ce qui accroît le champ de correction. Ces techniques requièrent une connaissance précise du profil de Cn2. Mon étude consiste à valider sur le ciel une nouvelle technique de mesure du profil de Cn2, appelée CO-SLIDAR, à partir des corrélations des mesures de pentes et de scintillation réalisées avec un analyseur Shack-Hartmann sur étoile binaire. Elle s'organise autour de deux grands axes. On réalise d'abord une simulation bout-en-bout de la reconstruction du profil de Cn2 dans un cas concret d'observation astronomique. On peut ainsi étudier l'impact des différentes sources d'erreur sur la reconstruction du profil de Cn2. Ceci nous permet d'améliorer la procédure d'estimation du profil de Cn2, en prenant en compte les bruits de détection. La deuxième partie de mon étude se consacre à la validation expérimentale. On dimensionne et caractérise en laboratoire un banc d'acquisition, le banc ProMeO. Ceci conduit à une bonne connaissance du fonctionnement du banc et nous permet de corriger certains effets instrumentaux. Le banc ProMeO est finalement couplé au télescope MeO de 1,5 m de diamètre. Les données acquises permettent une reconstruction du profil de Cn2, du sol jusqu'à 17 km, avec une résolution de 600 m. Les profils obtenus par la méthode CO-SLIDAR sont comparés avec succès à des profils issus de données météorologiques. L'ensemble de ces travaux constitue la première validation sur le ciel de la méthode CO-SLIDAR.

Este trabalho descreve experimentos realizados com Redes Neurais Artificiais e algoritmos de aprendizado simbólico. Também são investigados dois algoritmos de extração de conhecimento de Redes Neurais Artificiais. Esses experimentos são realizados com três bases de dados com o objetivo de comparar os desempenhos obtidos. As bases de dados utilizadas neste trabalho são: dados de falência de bancos brasileiros, dados do jogo da velha e dados de análise de crédito. São aplicadas sobre os dados três técnicas para melhoria de seus desempenhos. Essas técnicas são: partição pela menor classe, acréscimo de ruído nos exemplos da menor classe e seleção de atributos mais relevantes. Além da análise do desempenho obtido, também é feita uma análise da dificuldade de compreensão do conhecimento extraído por cada método em cada uma das bases de dados. / This work describes experiments carried out witch Artificial Neural Networks and symbolic learning algorithms. Two algorithms for knowledge extraction from Artificial Neural Networks are also investigates. This experiments are performed whit three data set with the objective of compare the performance obtained. The data set used in this work are: Brazilians banks bankruptcy data set, tic-tac-toe data set and credit analysis data set. Three techniques for data set performance improvements are investigates. These techniques are: partition for the smallest class, noise increment in the examples of the smallest class and selection of more important attributes. Besides the analysis of the performance obtained, an analysis of the understanding difficulty of the knowledge extracted by each method in each data bases is made.

Este trabalho descreve experimentos realizados com Redes Neurais Artificiais e algoritmos de aprendizado simbólico. Também são investigados dois algoritmos de extração de conhecimento de Redes Neurais Artificiais. Esses experimentos são realizados com três bases de dados com o objetivo de comparar os desempenhos obtidos. As bases de dados utilizadas neste trabalho são: dados de falência de bancos brasileiros, dados do jogo da velha e dados de análise de crédito. São aplicadas sobre os dados três técnicas para melhoria de seus desempenhos. Essas técnicas são: partição pela menor classe, acréscimo de ruído nos exemplos da menor classe e seleção de atributos mais relevantes. Além da análise do desempenho obtido, também é feita uma análise da dificuldade de compreensão do conhecimento extraído por cada método em cada uma das bases de dados. / This work describes experiments carried out witch Artificial Neural Networks and symbolic learning algorithms. Two algorithms for knowledge extraction from Artificial Neural Networks are also investigates. This experiments are performed whit three data set with the objective of compare the performance obtained. The data set used in this work are: Brazilians banks bankruptcy data set, tic-tac-toe data set and credit analysis data set. Three techniques for data set performance improvements are investigates. These techniques are: partition for the smallest class, noise increment in the examples of the smallest class and selection of more important attributes. Besides the analysis of the performance obtained, an analysis of the understanding difficulty of the knowledge extracted by each method in each data bases is made.

This thesis discusses methods to measure several atmospheric parameters related to turbulence. Techniques used by two different scintillometers based on weak turbulence theory are discussed along with a method to estimate the inner scale developed by Hill. The theory and minimization algorithm used to infer the atmospheric parameters are discussed. The main focus is on the analysis and collection of experimental data with a three-aperture receiver system. Intensity fluctuations from a CW laser source are collected over a 1km path with three different receiving apertures. The scintillation index is found for each receiving aperture and recently developed theory for all regimes of optical turbulence is used to infer three atmospheric parameters, Cn2, l0, and L0. The transverse wind speed is also calculated from the experimental data using a cross-correlation technique. Parallel to the three-aperture data collection is a commercial scintillometer unit which reports Cn2 and crosswind speed. There is also a weather station positioned at the receiver side which provides point measurements for temperature and wind speed. The Cn2 measurement obtained from the commercial scintillometer is used to infer l0, L0, and the scintillation index. Those values are then compared to the inferred atmospheric parameters from the experimental data. Hill's method is used as an estimate to l0 based upon path-averaged wind speed and is compared to the inferred l0 values. The optimal aperture sizes required for three-aperture data collection are presented. In closing, the technique for measuring crosswind speed is discussed along with the ideal aperture size and separation distance for data collection. Suggestions are offered for future experimentation.