Submitted by Rosivalda Pereira (mrs.pereira@ufma.br) on 2017-08-14T18:15:08Z No. of bitstreams: 1 DanielCosta.pdf: 1087754 bytes, checksum: ada5f863f42efd8298fff788c37bded3 (MD5) / Made available in DSpace on 2017-08-14T18:15:08Z (GMT). No. of bitstreams: 1 DanielCosta.pdf: 1087754 bytes, checksum: ada5f863f42efd8298fff788c37bded3 (MD5) Previous issue date: 2008-02-25 / Female breast cancer is the major cause of death in western countries. Efforts in Computer Vision have been made in order to add improve the diagnostic accuracy by radiologists. In this work, we present a methodology that uses independent component analysis (ICA) along with support vector machine (SVM) and linear discriminant analysis (LDA) to distinguish between mass or non-mass and benign or malign tissues from mammograms. As a result, it was found that: LDA reaches 90,11% of accuracy to discriminante between mass or non-mass and 95,38% to discriminate between benign or malignant tissues in DDSM database and in mini-MIAS database we obtained 85% to discriminate between mass or non-mass and 92% of accuracy to discriminate between benign or malignant tissues; SVM reaches 99,55% of accuracy to discriminate between mass or non-mass and the same percentage to discriminate between benign or malignat tissues in DDSM database whereas, and in MIAS database it was obtained 98% to discriminate between mass or non-mass and 100% to discriminate between benign or malignant tissues. / Câncer de mama feminino é o câncer que mais causa morte nos países ocidentais. Esforços em processamento de imagens foram feitos para melhorar a precisão dos diagnósticos por radiologistas. Neste trabalho, nós apresentamos uma metodologia que usa análise de componentes independentes (ICA) junto com análise discriminante linear (LDA) e máquina de vetor de suporte (SVM) para distinguir as imagens entre nódulos ou não-nódulos e os tecidos em benignos ou malignos. Como resultado, obteve-se com LDA 90,11% de acurácia na discriminação entre nódulo ou não-nódulo e 95,38% na discriminação de tecidos benignos ou malignos na base de dados DDSM. Na base de dados mini- MIAS, obteve-se 85% e 92% na discriminação entre nódulos ou não-nódulos e tecidos benignos ou malignos respectivamente. Com SVM, alcançou-se uma taxa de até 99,55% na discriminação de nódulos ou não-nódulos e a mesma porcentagem na discriminação entre tecidos benignos ou malignos na base de dados DDSM enquanto que na base de dados mini-MIAS, obteve-se 98% e até 100% na discriminação de nódulos ou não-nódulos e tecidos benignos ou malignos, respectivamente.

Cette thèse développe des méthodes de diagonalisation conjointe de matrices et de tenseurs d’ordre trois, et son application à la séparation MIMO de sources de télécommunications numériques. Après un état, les motivations et objectifs de la thèse sont présentés. Les problèmes de la diagonalisation conjointe et de la séparation de sources sont définis et un lien entre ces deux domaines est établi. Par la suite, plusieurs algorithmes itératifs de type Jacobi reposant sur une paramétrisation LU sont développés. Pour chacun des algorithmes, on propose de déterminer les matrices permettant de diagonaliser l’ensemble considéré par l’optimisation d’un critère inverse. On envisage la minimisation du critère selon deux approches : la première, de manière directe, et la seconde, en supposant que les éléments de l’ensemble considéré sont quasiment diagonaux. En ce qui concerne l’estimation des différents paramètres du problème, deux stratégies sont mises en œuvre : l’une consistant à estimer tous les paramètres indépendamment et l’autre reposant sur l’estimation indépendante de couples de paramètres spécifiquement choisis. Ainsi, nous proposons trois algorithmes pour la diagonalisation conjointe de matrices complexes symétriques ou hermitiennes et deux algorithmes pour la diagonalisation conjointe d’ensembles de tenseurs symétriques ou non-symétriques ou admettant une décomposition INDSCAL. Nous montrons aussi le lien existant entre la diagonalisation conjointe de tenseurs d’ordre trois et la décomposition canonique polyadique d’un tenseur d’ordre quatre, puis nous comparons les algorithmes développés à différentes méthodes de la littérature. Le bon comportement des algorithmes proposés est illustré au moyen de simulations numériques. Puis, ils sont validés dans le cadre de la séparation de sources de télécommunications numériques. / This thesis develops joint diagonalization of matrices and third-order tensors methods for MIMO source separation in the field of digital telecommunications. After a state of the art, the motivations and the objectives are presented. Then the joint diagonalisation and the blind source separation issues are defined and a link between both fields is established. Thereafter, five Jacobi-like iterative algorithms based on an LU parameterization are developed. For each of them, we propose to derive the diagonalization matrix by optimizing an inverse criterion. Two ways are investigated : minimizing the criterion in a direct way or assuming that the elements from the considered set are almost diagonal. Regarding the parameters derivation, two strategies are implemented : one consists in estimating each parameter independently, the other consists in the independent derivation of couple of well-chosen parameters. Hence, we propose three algorithms for the joint diagonalization of symmetric complex matrices or hermitian ones. The first one relies on searching for the roots of the criterion derivative, the second one relies on a minor eigenvector research and the last one relies on a gradient descent method enhanced by computation of the optimal adaptation step. In the framework of joint diagonalization of symmetric, INDSCAL or non symmetric third-order tensors, we have developed two algorithms. For each of them, the parameters derivation is done by computing the roots of the considered criterion derivative. We also show the link between the joint diagonalization of a third-order tensor set and the canonical polyadic decomposition of a fourth-order tensor. We confront both methods through numerical simulations. The good behavior of the proposed algorithms is illustrated by means of computing simulations. Finally, they are applied to the source separation of digital telecommunication signals.

In this thesis, improving the performance of the automatic recognition of the Doppler radar targets is studied. The radar used in this study is a ground-surveillance doppler radar. Target types are car, truck, bus, tank, helicopter, moving man and running man. The input of this thesis is the output of the real doppler radar signals which are normalized and preprocessed (TRP vectors: Target Recognition Pattern vectors) in the doctorate thesis by Erdogan (2002). TRP vectors are normalized and homogenized doppler radar target signals with respect to target speed, target aspect angle and target range. Some target classes have repetitions in time in their TRPs. By the use of these repetitions, improvement of the target type classification performance is studied. K-Nearest Neighbor (KNN) and Support Vector Machine (SVM) algorithms are used for doppler radar target classification and the results are evaluated. Before classification PCA (Principal Component Analysis), LDA (Linear Discriminant Analysis), NMF (Nonnegative Matrix Factorization) and ICA (Independent Component Analysis) are implemented and applied to normalized doppler radar signals for feature extraction and dimension reduction in an efficient way. These techniques transform the input vectors, which are the normalized doppler radar signals, to another space. The effects of the implementation of these feature extraction algoritms and the use of the repetitions in doppler radar target signals on the doppler radar target classification performance are studied.

<p>This thesis is concerned with the solution of the blind source separation problem (BSS). The BSS problem occurs frequently in various scientific and technical applications. In essence, it consists in separating meaningful underlying components out of a mixture of a multitude of superimposed signals.</p> <P> In the recent research literature there are two related approaches to the BSS problem: The first is known as Independent Component Analysis (ICA), where the goal is to transform the data such that the components become as independent as possible. The second is based on the notion of diagonality of certain characteristic matrices derived from the data. Here the goal is to transform the matrices such that they become as diagonal as possible. In this thesis we study the latter method of approximate joint diagonalization (AJD) to achieve a solution of the BSS problem. After an introduction to the general setting, the thesis provides an overview on particular choices for the set of target matrices that can be used for BSS by joint diagonalization.</p> <P> As the main contribution of the thesis, new algorithms for approximate joint diagonalization of several matrices with non-orthogonal transformations are developed.</p> <P> These newly developed algorithms will be tested on synthetic benchmark datasets and compared to other previous diagonalization algorithms.</p> <P> Applications of the BSS methods to biomedical signal processing are discussed and exemplified with real-life data sets of multi-channel biomagnetic recordings.</p> / <p>Diese Arbeit befasst sich mit der Lösung des Problems der blinden Signalquellentrennung (BSS). Das BSS Problem tritt häufig in vielen wissenschaftlichen und technischen Anwendungen auf. Im Kern besteht das Problem darin, aus einem Gemisch von überlagerten Signalen die zugrundeliegenden Quellsignale zu extrahieren.</p> <P> In wissenschaftlichen Publikationen zu diesem Thema werden hauptsächlich zwei Lösungsansätze verfolgt:</p> <P> Ein Ansatz ist die sogenannte "Analyse der unabhängigen Komponenten", die zum Ziel hat, eine lineare Transformation <B>V</B> der Daten <B>X</B> zu finden, sodass die Komponenten U_n der transformierten Daten <B>U</B> = <B> V X</B> (die sogenannten "independent components") so unabhängig wie möglich sind. Ein anderer Ansatz beruht auf einer simultanen Diagonalisierung mehrerer spezieller Matrizen, die aus den Daten gebildet werden. Diese Möglichkeit der Lösung des Problems der blinden Signalquellentrennung bildet den Schwerpunkt dieser Arbeit.</p> <P> Als Hauptbeitrag der vorliegenden Arbeit präsentieren wir neue Algorithmen zur simultanen Diagonalisierung mehrerer Matrizen mit Hilfe einer nicht-orthogonalen Transformation.</p> <P> Die neu entwickelten Algorithmen werden anhand von numerischen Simulationen getestet und mit bereits bestehenden Diagonalisierungsalgorithmen verglichen. Es zeigt sich, dass unser neues Verfahren sehr effizient und leistungsfähig ist. Schließlich werden Anwendungen der BSS Methoden auf Probleme der biomedizinischen Signalverarbeitung erläutert und anhand von realistischen biomagnetischen Messdaten wird die Nützlichkeit in der explorativen Datenanalyse unter Beweis gestellt.</p>

The last twenty years presented increased interest for the study of cerebral processes caused by external events (stimuli). One of the most significant endogenous components of Evoked Potentials is the P600 component. The P600 component may be defined as the most positive peak in the time window between 500 and 800 msec after an eliciting stimulus. This component is thought to reflect the response selection stage of information processing. P600 component is usually less pronounced compared to other components, such as the N100 or the P300. Frequently the P600 component appears as a not-easily discernible secondary peak overlying the ascending negative-going slope of the P300 waveform. In our study we used ERP data from various groups of patients and healthy controls. Patients were recruited from the outpatient university clinic of Eginition Hospital of the University of Athens. The controls were recruited from hospital staff and local volunteer groups. The aim of the study is the implementation of classification systems for these groups, using P600 features. This is usually not achieved well using as features the ERPs amplitude and latency. So for that reason, in our study, we want to extract new features using advanced techniques for processing the original ERPs, such as the Independent Component Analysis (ICA) method. However as a precursor of ICA, is considered the Principal Component Analysis (PCA) method, which we used for comparison reasons to ICA. In the application of ICA we achieve the decomposition of the recorded signals in ICs, supposing temporally independent components and propose ICs selection techniques in order to recompose the P600 component. The next stage was the use of a classification method based on the features extracted using the original data, data extracted through PCA processing and ICA-processed data. First we applied Kolmogorov-Smirnov test to check the normality of the distribution of the features, then we used the Logistic Regression method for classification and finally we have done two implementations of classification using Probabilistic Neural Networks. The first implementation was done with the creation of 15 features from the P600 peak amplitudes from the subjects’ data and the second implementation was done with the creation of four meta-features from the subjects’ P600 amplitude data. The results show that the application of ICA, combined with the logistic regression classification technique, provides notable improvement, compared to the classification performance based on the original ERPs. The main merit of the application is that classification is based on single parameters, i.e. amplitude of the P600 component, or its latency or its termination latency, which are directly related to the brain mechanisms related to ERP generation and pathological processes. / Τα τελευταία 20 χρόνια παρουσιάζεται αυξημένο ενδιαφέρον για την μελέτη εγκεφαλικών επεξεργασιών που προκλήθηκαν από εξωτερικά γεγονότα (ερέθισμα). Ένα από τα πιο σημαντικά ενδογενή συστατικά των Προκλητών Δυναμικών είναι το συστατικό P600. Το συστατικό P600 μπορεί να οριστεί σαν η πιο θετική αιχμή στο χρονικό διάστημα μεταξύ 500 και 800 msec μετά από ένα εκλυτικό ερέθισμα. Το συστατικό αυτό θεωρείται ότι απεικονίζει το στάδιο επιλογής απόκρισης της επεξεργασίας πληροφορίας. Το συστατικό P600 είναι συνήθως λιγότερο έντονο συγκρίνοντας το με άλλα συστατικά, όπως το N100 ή το P300. Συχνά το συστατικό P600 εμφανίζεται ως μια δυσδιάκριτη δεύτερη αιχμή, επικαλύπτοντας την ανοδική αρνητική κλίση της κυματομορφής του P300. Στη μελέτη μας χρησιμοποιήσαμε δεδομένα ΠΔ από ποικίλες ομάδες ασθενών και υγιών μαρτύρων. Οι ασθενείς συλλέχθησαν από τη πανεπιστημιακή κλινική του Αιγηνήτειου Νοσοκομείου του Πανεπιστημίου Αθηνών. Οι υγιείς συλλέχθησαν από το προσωπικό του νοσοκομείου και ομάδες εθελοντών. Ο σκοπός της μελέτης είναι η εφαρμογή συστημάτων ταξινόμησης για αυτές τις ομάδες, χρησιμοποιώντας χαρακτηριστικά του P600. Αυτό συνήθως δεν επιτυγχάνεται καλά χρησιμοποιώντας σαν χαρακτηριστικά το πλάτος και τον λανθάνοντα χρόνο των ΠΔ. Για αυτό το λόγο, στην μελέτη μας, θέλουμε να εξάγουμε νέα χαρακτηριστικά χρησιμοποιώντας προηγμένες τεχνικές για επεξεργασία των αρχικών ΠΔ, όπως τη μέθοδο Ανάλυσης Ανεξαρτήτων Συνιστωσών (ICA). Εντούτοις ως πρόδρομο της ICA, θεωρείται η μέθοδος Ανάλυσης Κύριων Συνιστωσών (PCA), την οποία χρησιμοποιήσαμε για συγκριτικούς λόγους με την ICA. Στην εφαρμογή της ICA προχωρήσαμε στην αποσύνθεση των καταγραφόμενων σημάτων σε Ανεξάρτητες Συνιστώσες και διερευνήσαμε τρεις τεχνικές επιλογής ανεξαρτήτων συνιστωσών μέσω των οποίων επανασυνθέσαμε το συστατικό P600. Το επόμενο βήμα ήταν η χρήση μεθόδου ταξινόμησης βασισμένης στα χαρακτηριστικά που εξάχθηκαν χρησιμοποιώντας τα αρχικά δεδομένα, τα δεδομένα με επεξεργασία PCA και τα δεδομένα με επεξεργασία ICA. Πρώτα εφαρμόσαμε το τεστ Kolmogorov-Smirnov για τον έλεγχο της κανονικότητας της κατανομής των χαρακτηριστικών, μετά χρησιμοποιήσαμε τη μέθοδο Λογαριθμικής Παλινδρόμησης (Logistic Regression) για ταξινόμηση και τελικά πραγματοποιήσαμε δύο εφαρμογές ταξινόμησης χρησιμοποιώντας Πιθανοκρατικά Νευρωνικά Δίκτυα (Probabilistic Neural Networks). Η πρώτη εφαρμογή έγινε με την δημιουργία 15 χαρακτηριστικών από τα πλάτη των αιχμών του P600 από τα δεδομένα των ομάδων και η δεύτερη εφαρμογή έγινε με την δημιουργία τεσσάρων μετά-χαρακτηριστικών από τα δεδομένα των πλατών των ομάδων. Τα αποτελέσματα δείχνουν ότι η εφαρμογή της ICA, συνδυασμένη με την τεχνική ταξινόμησης λογαριθμικής παλινδρόμησης, παρέχει αξιοσημείωτη βελτίωση, συγκριτικά με την απόδοση ταξινόμησης βάση των αρχικών ΠΔ. Η κύρια αξία της εφαρμογής είναι ότι η ταξινόμηση πετυχαίνει ποσοστά μεγαλύτερα του 80% βασιζόμενη σε μία μόνο κάθε φορά παράμετρο, π.χ. το πλάτος του συστατικού P600, ή τον λανθάνοντα χρόνο του ή τον λανθάνοντα χρόνο τερματισμού του, οι οποίες σχετίζονται άμεσα με τους μηχανισμούς του εγκεφάλου σχετικούς με την παραγωγή ΠΔ και τις παθολογικές διαδικασίες.

L'analyse en composantes indépendantes (ACI) est une méthode d'analyse statistique qui consiste à exprimer les données observées (mélanges de sources) en une transformation linéaire de variables latentes (sources) supposées non gaussiennes et mutuellement indépendantes. Dans certaines applications, on suppose que les mélanges de sources peuvent être groupés de façon à ce que ceux appartenant au même groupe soient fonction des mêmes sources. Ceci implique que les coefficients de chacune des colonnes de la matrice de mélange peuvent être regroupés selon ces mêmes groupes et que tous les coefficients de certains de ces groupes soient nuls. En d'autres mots, on suppose que la matrice de mélange est éparse par groupe. Cette hypothèse facilite l'interprétation et améliore la précision du modèle d'ACI. Dans cette optique, nous proposons de résoudre le problème d'ACI avec une matrice de mélange éparse par groupe à l'aide d'une méthode basée sur le LASSO par groupe adaptatif, lequel pénalise la norme 1 des groupes de coefficients avec des poids adaptatifs. Dans ce mémoire, nous soulignons l'utilité de notre méthode lors d'applications en imagerie cérébrale, plus précisément en imagerie par résonance magnétique. Lors de simulations, nous illustrons par un exemple l'efficacité de notre méthode à réduire vers zéro les groupes de coefficients non-significatifs au sein de la matrice de mélange. Nous montrons aussi que la précision de la méthode proposée est supérieure à celle de l'estimateur du maximum de la vraisemblance pénalisée par le LASSO adaptatif dans le cas où la matrice de mélange est éparse par groupe. / Independent component analysis (ICA) is a method of statistical analysis where the main goal is to express the observed data (mixtures) in a linear transformation of latent variables (sources) believed to be non-Gaussian and mutually independent. In some applications, the mixtures can be grouped so that the mixtures belonging to the same group are function of the same sources. This implies that the coefficients of each column of the mixing matrix can be grouped according to these same groups and that all the coefficients of some of these groups are zero. In other words, we suppose that the mixing matrix is sparse per group. This assumption facilitates the interpretation and improves the accuracy of the ICA model. In this context, we propose to solve the problem of ICA with a sparse group mixing matrix by a method based on the adaptive group LASSO. The latter penalizes the 1-norm of the groups of coefficients with adaptive weights. In this thesis, we point out the utility of our method in applications in brain imaging, specifically in magnetic resonance imaging. Through simulations, we illustrate with an example the effectiveness of our method to reduce to zero the non-significant groups of coefficients within the mixing matrix. We also show that the accuracy of the proposed method is greater than the one of the maximum likelihood estimator with an adaptive LASSO penalization in the case where the mixing matrix is sparse per group.

In immersive audio environments with distributed microphones, Independent Component Analysis (ICA) can be applied to uncover signals from a mixture of other signals and noise, such as in a cocktail party recording. ICA algorithms have been developed for instantaneous source mixtures and convolutional source mixtures. While ICA for instantaneous mixtures works when no delays exist between the signals in each mixture, distributed microphone recordings typically result various delays of the signals over the recorded channels. The convolutive ICA algorithm should account for delays; however, it requires many parameters to be set and often has stability issues. This thesis introduces the Channel Aligned FastICA (CAICA), which requires knowledge of the source distance to each microphone, but does not require knowledge of noise sources. Furthermore, the CAICA is combined with Time Frequency Masking (TFM), yielding even better SOI extraction even in low SNR environments. Simulations were conducted for ranking experiments tested the performance of three algorithms: Weighted Beamforming (WB), CAICA, CAICA with TFM. The Closest Microphone (CM) recording is used as a reference for all three. Statistical analyses on the results demonstrated superior performance for the CAICA with TFM. The algorithms were applied to experimental recordings to support the conclusions of the simulations. These techniques can be deployed in mobile platforms, used in surveillance for capturing human speech and potentially adapted to biomedical fields.

La sécurité des personnes et des équipements est un élément capital dans le domaine des transports routiers et ferroviaires. Depuis quelques années, les Passages à Niveau (PN) ont fait l’objet de davantage d'attention afin d'accroître la sécurité des usagers sur cette portion route/rail considérée comme dangereuse. Nous proposons dans cette thèse un système de vision stéréoscopique pour la détection automatique des situations dangereuses. Un tel système permet la détection et la localisation d'obstacles sur ou autour du PN. Le système de vision proposé est composé de deux caméras supervisant la zone de croisement. Nous avons développé des algorithmes permettant à la fois la détection d'objets, tels que des piétons ou des véhicules, et la localisation 3D de ces derniers. L'algorithme de détection d'obstacles se base sur l'Analyse en Composantes Indépendantes et la propagation de croyance spatio-temporelle. L'algorithme de localisation tridimensionnelle exploite les avantages des méthodes locales et globales, et est composé de trois étapes : la première consiste à estimer une carte de disparité à partir d'une fonction de vraisemblance basée sur les méthodes locales. La deuxième étape permet d'identifier les pixels bien mis en correspondance ayant des mesures de confiances élevées. Ce sous-ensemble de pixels est le point de départ de la troisième étape qui consiste à ré-estimer les disparités du reste des pixels par propagation de croyance sélective. Le mouvement est introduit comme une contrainte dans l'algorithme de localisation 3D permettant l'amélioration de la précision de localisation et l'accélération du temps de traitement. / Within the past years, railways undertakings became interested in the assessment of Level Crossings (LC) safety. We propose in this thesis an Automatic Video-Surveillance system (AVS) at LC for an automatic detection of specific events. The system allows automatically detecting and 3D localizing the presence of one or more obstacles which are motionless at the level crossing. Our research aims at developing an AVS using the passive stereo vision principles. The proposed imaging system uses two cameras to detect and localize any kind of object lying on a railway level crossing. The cameras are placed so that the dangerous zones are well (fully) monitored. The system supervises and estimates automatically the critical situations by detecting objects in the hazardous zone defined as the crossing zone of a railway line by a road or path. The AVS system is used to monitor dynamic scenes where interactions take place among objects of interest (people or vehicles). After a classical image grabbing and digitizing step, the processing is composed of the two following modules: moving and stationary objects detection and 3-D localization. The developed stereo matching algorithm stems from an inference principle based on belief propagation and energy minimization. It takes into account the advantages of local methods for reducing the complexity of the inference step achieved by the belief propagation technique which leads to an improvement in the quality of results. The motion detection module is considered as a constraint which allows improving and speeding up the 3D localization algorithm.

La séparation aveugle de sources aveugle (BSS) est une technique d’estimation des différents signaux observés au travers de leurs mélanges à l’aide de plusieurs capteurs, lorsque le mélange et les signaux sont inconnus. Bien qu’il ait été démontré mathématiquement que pour des mélanges linéaires, sous des conditions faibles, des sources mutuellement indépendantes peuvent être estimées, il n’existe dans de résultats théoriques généraux dans le cas de mélanges non-linéaires. La littérature sur ce sujet est limitée à des résultats concernant des mélanges non linéaires spécifiques.Dans la présente étude, le problème est abordé en utilisant une nouvelle approche utilisant l’information temporelle des signaux. L’idée originale conduisant à ce résultat, est d’étudier le problème de mélanges linéaires, mais variant dans le temps, déduit du problème non linéaire initial par dérivation. Il est démontré que les contre-exemples déjà présentés, démontrant l’inefficacité de l’analyse par composants indépendants (ACI) pour les mélanges non-linéaires, perdent leur validité, considérant l’indépendance au sens des processus stochastiques, au lieu de l’indépendance au sens des variables aléatoires. Sur la base de cette approche, de bons résultats théoriques et des développements algorithmiques sont fournis. Bien que ces réalisations ne soient pas considérées comme une preuve mathématique de la séparabilité des mélanges non-linéaires, il est démontré que, compte tenu de quelques hypothèses satisfaites dans la plupart des applications pratiques, elles sont séparables.De plus, les BSS non-linéaires pour deux ensembles utiles de signaux sources sont également traités, lorsque les sources sont (1) spatialement parcimonieuses, ou (2) des processus Gaussiens. Des méthodes BSS particulières sont proposées pour ces deux cas, dont chacun a été largement étudié dans la littérature qui correspond à des propriétés réalistes pour de nombreuses applications pratiques.Dans le cas de processus Gaussiens, il est démontré que toutes les applications non-linéaires ne peuvent pas préserver la gaussianité de l’entrée, cependant, si on restreint l’étude aux fonctions polynomiales, la seule fonction préservant le caractère gaussiens des processus (signaux) est la fonction linéaire. Cette idée est utilisée pour proposer un algorithme de linéarisation qui, en cascade par une méthode BSS linéaire classique, sépare les mélanges polynomiaux de processus Gaussiens.En ce qui concerne les sources parcimonieuses, on montre qu’elles constituent des variétés distinctes dans l’espaces des observations et peuvent être séparées une fois que les variétés sont apprises. À cette fin, plusieurs problèmes d’apprentissage multiple ont été généralement étudiés, dont les résultats ne se limitent pas au cadre proposé du SRS et peuvent être utilisés dans d’autres domaines nécessitant un problème similaire. / Blind Source Separation (BSS) is a technique for estimating individual source components from their mixtures at multiple sensors, where the mixing model is unknown. Although it has been mathematically shown that for linear mixtures, under mild conditions, mutually independent sources can be reconstructed up to accepted ambiguities, there is not such theoretical basis for general nonlinear models. This is why there are relatively few results in the literature in this regard in the recent decades, which are focused on specific structured nonlinearities.In the present study, the problem is tackled using a novel approach utilizing temporal information of the signals. The original idea followed in this purpose is to study a linear time-varying source separation problem deduced from the initial nonlinear problem by derivations. It is shown that already-proposed counter-examples showing inefficiency of Independent Component Analysis (ICA) for nonlinear mixtures, loose their validity, considering independence in the sense of stochastic processes instead of simple random variables. Based on this approach, both nice theoretical results and algorithmic developments are provided. Even though these achievements are not claimed to be a mathematical proof for the separability of nonlinear mixtures, it is shown that given a few assumptions, which are satisfied in most practical applications, they are separable.Moreover, nonlinear BSS for two useful sets of source signals is also addressed: (1) spatially sparse sources and (2) Gaussian processes. Distinct BSS methods are proposed for these two cases, each of which has been widely studied in the literature and has been shown to be quite beneficial in modeling many practical applications.Concerning Gaussian processes, it is demonstrated that not all nonlinear mappings can preserve Gaussianity of the input. For example being restricted to polynomial functions, the only Gaussianity-preserving function is linear. This idea is utilized for proposing a linearizing algorithm which, cascaded by a conventional linear BSS method, separates polynomial mixturesof Gaussian processes.Concerning spatially sparse sources, it is shown that spatially sparsesources make manifolds in the observations space, and can be separated once the manifolds are clustered and learned. For this purpose, multiple manifold learning problem has been generally studied, whose results are not limited to the proposed BSS framework and can be employed in other topics requiring a similar issue.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Monitoring of process control systems is extremely important for industries to ensure the quality of the product and the safety of the process. Predictive controllers, also known by MPC (Model Predictive Control), usually has a well performance initially. However, after a period, many factors contribute to the deterioration of its performance. This highlights the importance of monitoring the MPC control systems. In this work, tools based on multivariate statistical methods are discussed and applied to the problem of monitoring and Performance Assessment of predictive controllers. The methods presented here are: PCA (Principal Component Analysis) and ICA (Independent Component Analysis). Both are techniques that use data collected directly from the process. The first is widely used in Performance Assessment of predictive controllers. The second is a more recent technique that has arisen, mainly in order to be used in fault detection systems. The analyzes are made when applied in simulated processes characteristic of the petrochemical industry operating under MPC control. / O monitoramento de sistemas de controle de processos é extremamente importante no que diz respeito às indústrias, para garantir a qualidade do que é produzido e a segurança do processo. Os controladores preditivos, também conhecidos pela sigla em inglês MPC (Model Predictive Control), costumam ter um bom desempenho inicialmente. Entretanto, após um certo período, muitos fatores contribuem para a deterioração de seu desempenho. Isto evidencia a importância do monitoramento dos sistemas de controle MPC. Neste trabalho aborda-se ferramentas, baseada em métodos estatísticos multivariados, aplicados ao problema de monitoramento e avaliação de desempenho de controladores preditivos. Os métodos aqui apresentados são: o PCA (Análise por componentes principais) e o ICA (Análise por componentes independentes). Ambas são técnicas que utilizam dados coletados diretamente do processo. O primeiro é largamente utilizado na avaliação de desempenho de controladores preditivos. Já o segundo, é uma técnica mais recente que surgiu, principalmente, com o intuito de ser utilizado em sistemas de detecção de falhas. As análises são feitas quando aplicadas em processos simulados característicos da indústria petroquímica operando sob controle MPC.