A análise de textura é uma das mais básicas e famosas áreas de pesquisa em visão computacional. Ela é também de grande importância em muitas outras disciplinas, tais como ciências médicas e biológicas. Por exemplo, uma tarefa comum de análise de textura é a detecção de tecidos não saudáveis em imagens de Ressonância Magnética do pulmão. Nesta dissertação, nós propomos um método novo de caracterização de textura baseado nas medidas de complexidade tais como o expoente de Hurst, o expoente de Lyapunov e a complexidade de Lempel-Ziv. Estas medidas foram aplicadas sobre amostras de imagens no espaço de frequência. Três métodos de amostragem foram propostas, amostragem: radial, circular e por caminhadas determinísticas parcialmente auto- repulsivas (amostragem CDPA). Cada método de amostragem produz um vetor de características por medida de complexidade aplicada. Esse vetor contem um conjunto de descritores que descrevem a imagem processada. Portanto, cada imagem será representada por nove vetores de características (três medidas de complexidade e três métodos de amostragem), os quais serão comparados na tarefa de classificação de texturas. No final, concatenamos cada vetor de características conseguido calculando a complexidade de Lempel-Ziv em amostras radiais e circulares com os descritores obtidos através de técnicas de análise de textura tradicionais, tais como padrões binários locais (LBP), wavelets de Gabor (GW), matrizes de co-ocorrência en níveis de cinza (GLCM) e caminhadas determinísticas parcialmente auto-repulsivas em grafos (CDPAg). Este enfoque foi testado sobre três bancos de imagens: Brodatz, USPtex e UIUC, cada um com seus próprios desafios conhecidos. As taxas de acerto de todos os métodos tradicionais foram incrementadas com a concatenação de relativamente poucos descritores de Lempel-Ziv. Por exemplo, no caso do método LBP, o incremento foi de 84.25% a 89.09% com a concatenação de somente cinco descritores. De fato, simplesmente concatenando cinco descritores são suficientes para ver um incremento na taxa de acerto de todos os métodos tradicionais estudados. Por outro lado, a concatenação de un número excessivo de descritores de Lempel-Ziv (por exemplo mais de 40) geralmente não leva a melhora. Neste sentido, vendo os resultados semelhantes obtidos nos três bancos de imagens analisados, podemos concluir que o método proposto pode ser usado para incrementar as taxas de acerto em outras tarefas que envolvam classificação de texturas. Finalmente, com a amostragem CDPA também se obtém resultados significativos, que podem ser melhorados em trabalhos futuros. / Texture analysis is one of the basic and most popular computer vision research areas. It is also of importance in many other disciplines, such as medical sciences and biology. For example, non-healthy tissue detection in lung Magnetic Resonance images is a common texture analysis task. We proposed a novel method for texture characterization based on complexity measures such as Lyapunov exponent, Hurst exponent and Lempel-Ziv complexity. This measurements were applied over samples taken from images in the frequency domain. Three types of sampling methods were proposed: radial sampling, circular sampling and sampling by using partially self-avoiding deterministic walks (CDPA sampling). Each sampling method produce a feature vector which contains a set of descriptors that characterize the processed image. Then, each image will be represented by nine feature vectors which are means to be compared in texture classification tasks (three complexity measures over samples from three sampling methods). In the end, we combine each Lempel-Ziv feature vector from the circular and radial sampling with descriptors obtained through traditional image analysis techniques, such as Local Binary Patterns (LBP), Gabor Wavelets (GW), Gray Level Co-occurrence Matrix (GLCM) and Self-avoiding Deterministic Walks in graphs (CDPAg). This approach were tested in three datasets: Brodatz, USPtex and UIUC, each one with its own well-known challenges. All traditional methods success rates were increased by adding relatively few Lempel-Ziv descriptors. For example in the LBP case the increment went from 84.25% to 89.09% with the addition of only five descriptors. In fact, just adding five Lempel-Ziv descriptors are enough to see an increment in the success rate of every traditional method. However, adding too many Lempel-Ziv descriptors (for example more than 40) generally doesnt produce better results. In this sense, seeing the similar results we obtain in all three databases, we conclude that this approach may be used to increment the success rate in a lot of others texture classification tasks. Finally, the CDPA sampling also obtain very promising results that we can improve further on future works.

Magister Scientiae - MSc / This study examines the behaviour of the South African financial markets with regards to the Geometric Brownian motion process. It uses the daily, weekly, and monthly stock returns time series of some major securities trading in the South African financial market, more specifically the US dollar/Euro, JSE ALSI Total Returns Index, South African All Bond Index, Anglo American Corporation, Standard Bank, Sasol, US dollar Gold Price , Brent spot oil price, and South African white maize near future. The assumptions underlying the Geometric Brownian motion in finance, namely the stationarity, the normality and the independence of stock returns, are tested using both graphical (histograms and normal plots) and statistical test (Kolmogorov-Simirnov test, Box-Ljung statistic and Augmented Dickey-Fuller test) methods to check whether or not the Brownian motion as a model for South African financial markets holds. The Hurst exponent or independence index is also applied to support the results from the previous test. Theoretically, the independent or Geometric Brownian motion time series should be characterised by the Hurst exponent of ½. A value of a Hurst exponent different from that would indicate the presence of long memory or fractional Brownian motion in a time series. The study shows that at least one assumption is violated when the Geometric Brownian motion process is examined assumption by assumption. It also reveals the presence of both long memory and random walk or Geometric Brownian motion in the South African financial markets returns when the Hurst index analysis is used and finds that the Currency market is the most efficient of the South African financial markets. The study concludes that although some assumptions underlying the rocess are violated, the Brownian motion as a model in South African financial markets can not be rejected. It can be accepted in some instances if some parameters such as the Hurst exponent are added. / South Africa

The thesis deals with the design and optimization of artificial neural networks (specifically nonlinear autoregressive networks) and their subsequent usage in predictive application of stock market time series.

This thesis deals with a complex analysis and prediction of foreign exchange markets. It uses advanced artificial intelligence methods, namely neural networks and chaos theory. It introduces unconventional approaches and methods of each of these areas, compares them and uses on a real problem. The core of this thesis is a comparison of several prediction models based on completely different principles and underlying theories. The outcome is then a selection of the most appropriate prediction model called NAR + H. The model is evaluated according to several criteria, the pros and cons are discussed and approximate expected profitability and risk are calculated. All analytical, prediction and partial algorithms are implemented in Matlab development environment and form a unified library of all used functions and scripts. It also may be considered as a secondary main outcome of the thesis.

The aim of this thesis is to examine an empirical relationship between multifrac- tality of financial time series and its returns. We approach the multifractality of a given time series as a measure of its complexity. Multifractal financial time series exhibit repeating self-similar patterns. Multifractality could be a good predictor of stock returns or a factor which can be used in asset pricing. We expected that capturing the complexity of a given time series by a model, a positive or a negative risk premia for investing into "more multifractal assets" could be found. Daily prices of 31 stock indices and daily returns of 10-years US government bonds were downloaded. All the data were recorded between 2012 and 2021. After estimation the multifractal spectra, applying MF-DFA method, of all stock indices, we ordered all stock indices from the lowest to the most multifractal. Then, we constructed a "multifractal portfolio" holding a long position in the 7 most multifractal and holding a short position in the 7 least multifractal stock indices. Fama-MacBeth regression with market risk premia and multifractal variable as independent variables was applied. Multi- fractality in all examined financial time series was found. We also found a very low negative risk premia for holding "a multifractal...

This Ph.D. dissertation processes and analyzes signals from the neonatal intensive care units (NICUs) for the study of maturity, systemic infection (sepsis) and the influence of immunization in the premature newborn. A special attention is payed to the electroencephalography and the breathing signal. The former is often contaminated by several sources of noise, thus methods based on the signals decomposition and optimal noise cancellation, adapted to the characteristics of the immature EEG, were proposed and evaluated objectively on real and simulated signals. By means of the EEG and delta burst analysis, detected automatically by a proposed classifier, infant's maturation and the effects of vaccination are studied. Concerning the second signal, breathing, non-linear and fractal methods are adapted to evaluate maturity and sepsis. A robustness study of estimation methods is also conducted, showing that the Hurst exponent, estimated on respiratory variability signals, is a good detector of infection.

Dans le cadre des recherches de la qualité des ressources en eau, l’étude du processus de transfert de masse du sol vers la nappe phréatique constitue un élément primordial pour la compréhension de la pollution de cette dernière. En effet, les éléments polluants solubles à la surface (produits liés aux activités humaines tels engrais, pesticides...) peuvent transiter vers la nappe à travers le milieu poreux qu’est le sol. Ce scénario de transfert de pollution repose sur deux phénomènes : la pluie qui génère la masse d’eau à la surface et la dispersion de celle-ci à travers le milieu poreux. La dispersion de masse dans un milieu poreux naturel comme le sol forme un sujet de recherche vaste et difficile aussi bien au plan expérimental que théorique. Sa modélisation constitue une préoccupation du laboratoire EMMAH, en particulier dans le cadre du projet Sol Virtuel dans lequel un modèle de transfert (modèle PASTIS) a été développé. Le couplage de ce modèle de transfert avec en entrée un modèle décrivant la dynamique aléatoire de la pluie est un des objectifs de la présente thèse. Ce travail de thèse aborde cet objectif en s’appuyant d’une part sur des résultats d’observations expérimentaux et d’autre part sur de la modélisation inspirée par l’analyse des données d’observation. La première partie du travail est consacrée à l’élaboration d’un modèle stochastique de pluie. Le choix et la nature du modèle sont basés sur les caractéristiques obtenus à partir de l’analyse de données de hauteur de pluie recueillies sur 40 ans (1968-2008) sur le Centre de Recherche de l’INRA d’Avignon. Pour cela, la représentation cumulée des précipitations sera assimilée à une marche aléatoire dans laquelle les sauts et les temps d’attente entre les sauts sont respectivement les amplitudes et les durées aléatoires entre deux occurrences d’événements de pluie. Ainsi, la loi de probabilité des sauts (loi log-normale) et celle des temps d’attente entre les sauts (loi alpha-stable) sont obtenus en analysant les lois de probabilité des amplitudes et des occurrences des événements de pluie. Nous montrons alors que ce modèle de marche aléatoire tend vers un mouvement brownien géométrique subordonné en temps (quand les pas d’espace et de temps de la marche tendent simultanément vers zéro tout en gardant un rapport constant) dont la loi de densité de probabilité est régie par une équation de Fokker Planck fractionnaire (FFPE). Deux approches sont ensuite utilisées pour la mise en œuvre du modèle. La première approche est de type stochastique et repose sur le lien existant entre le processus stochastique issu de l’équation différentielle d’Itô et la FFPE. La deuxième approche utilise une résolution numérique directe par discrétisation de la FFPE. Conformément à l’objectif principal de la thèse, la seconde partie du travail est consacrée à l’analyse de la contribution de la pluie aux fluctuations de la nappe phréatique. Cette analyse est faite sur la base de deux relevés simultanées d’observations de hauteurs de pluie et de la nappe phréatique sur 14 mois (février 2005-mars 2006). Une étude statistique des liens entre les signaux de pluie et de fluctuations de la nappe est menée comme suit : Les données de variations de hauteur de nappe sont analysées et traitées pour isoler les fluctuations cohérentes avec les événements de pluie. Par ailleurs, afin de tenir compte de la dispersion de masse dans le sol, le transport de la masse d’eau pluviale dans le sol sera modélisé par un code de calcul de transfert (modèle PASTIS) auquel nous appliquons en entrée les données de hauteurs de pluie mesurées. Les résultats du modèle permettent entre autre d’estimer l’état hydrique du sol à une profondeur donnée (ici fixée à 1.6m). Une étude de la corrélation entre cet état hydrique et les fluctuations de la nappe sera ensuite effectuée en complément à celle décrite ci-dessus pour illustrer la possibilité de modéliser l’impact de la pluie sur les fluctuations de la nappe / Within the research quality of water resources, the study of the process of mass transfer from soil to groundwater is a key element for understanding the pollution of the latter. Indeed, soluble contaminants to the surface (related to human activities such fertilizers, pesticides products ...) can transit to the web through the porous medium that is the ground. This scenario transfer pollution based on two phenomena: the rain that generates the body of water to the dispersion and the surface thereof through the porous medium. The dispersion of mass in a natural porous medium such as soil forms a subject of extensive research and difficult both experimental and theoretical grounds. Its modeling is a concern EMMAH laboratory, particularly in the context of Virtual Sol project in which a transfer model (PASTIS model) was developed. The coupling of this transfer model with input a model describing the dynamics of random rain is one of the objectives of this thesis. This thesis addresses this goal by relying in part on the results of experimental observations and also on modeling inspired by the analysis of observational data. The first part of the work is devoted to the development of a stochastic model of rain. The choice and nature of the model are based on the features obtained from the analysis of data collected rainfall over 40 years (1968-2008) on the Research Centre INRA Avignon. For this, the cumulative rainfall representation will be treated as a random walk in which the jumps and waiting times between jumps are the amplitudes and durations between two random occurrences of rain events. Thus, the probability jumps (log-normal distribution) and that of waiting between jumps (Law alpha-stable) time is obtained by analyzing the laws of probability amplitudes and occurrences of rain events. We show that the random walk model tends towards a subordinate in time geometric Brownian motion (when space step and time step walking simultaneously tend to zero while maintaining a constant ratio), the law of probability density is governed by a Fokker Planck fractional (FFPE). Two approaches are then used to implement the model. The first approach is based on stochastic type and the relationship between the stochastic process derived from the differential equation of Itô and FFPE. The second approach uses a direct numerical solution by discretization of the FFPE. Accordance with the main objective of the thesis, the second part of the work is devoted to the analysis of the contribution of rain to fluctuations in groundwater. We approach this analysis on the basis of two simultaneous records of observations of rainfall amounts and groundwater over 14 months (February 2005-March 2006). A statistical study of the relationship between the signals of rain and fluctuating water will be conducted. Data sheet height variations are analyzed and processed to isolate coherent fluctuations with rain events. In addition, to take account of the mass dispersion in the soil, the mass transport of storm water in the soil layer is modeled by a calculation code transfer (PASTIS model) which we apply input data measured heights of rain. The model results allow between another estimate soil water status at a given depth (here set at 1.6m). A study of the correlation between the water status and fluctuating water will then be performed in addition to that described above to illustrate the ability to model the impact of rain on the water table fluctuations

Ce mémoire de thèse porte sur le traitement et l'analyse des signaux issus des unités de soins intensifs néonatales (USIN) pour l'étude de la maturité, de l'infection généralisée et de l'influence de l'immunisation chez le nouveau-né prématuré. Une attention particulière est portée sur l'électroencéphalographie et le signal de respiration. Pour le premier, ce signal est souvent bruité en USIN et des méthodes de décomposition du signal et d'annulation optimale du bruit, adaptées aux particularités des EEG immatures, ont été proposées et évaluées objectivement sur signaux réels et simulés. L'analyse de l'EEG et des bouffées delta, repérées automatiquement par un classificateur proposé, ont permis d'étudier la maturation et les effets de la vaccination. Pour la seconde modalité, la respiration, des méthodes non-linéaires et fractales sont retenues et adaptées pour évaluer la maturité et l'infection généralisée. Une étude de robustesse des méthodes d'estimation est menée et on montre que l'exposant de Hurst, estimé sur des signaux de variabilité respiratoire, est un bon détecteur de l'infection. / This Ph.D. dissertation processes and analyzes signals from the neonatal intensive care units (NICUs) for the study of maturity, systemic infection (sepsis) and the influence of immunization in the premature newborn. A special attention is payed to the electroencephalography and the breathing signal. The former is often contaminated by several sources of noise, thus methods based on the signals decomposition and optimal noise cancellation, adapted to the characteristics of the immature EEG, were proposed and evaluated objectively on real and simulated signals. By means of the EEG and delta burst analysis, detected automatically by a proposed classifier, infant's maturation and the effects of vaccination are studied. Concerning the second signal, breathing, non-linear and fractal methods are adapted to evaluate maturity and sepsis. A robustness study of estimation methods is also conducted, showing that the Hurst exponent, estimated on respiratory variability signals, is a good detector of infection.

The thesis deals with design, implementation and optimization of a model based on artificial intelligence and neural networks, which is able to predict future time series prices on a stock market. Main goal is to create an object oriented application for successful future trend prediction of financial derivatives with the use of cooperating methods such as Hurst exponent evaluation and automated market simulation.