Στην εργασία αυτή πραγματοποιείται μελέτη και εφαρμογή της μεθόδου ανάλυσης ανεξάρτητων συνιστωσών. Αφού παρουσιαστούν οι τεχνικές της μεθόδου και τα μαθηματικά μοντέλα που χρησιμοποιεί εξετάζεται η εξαγωγή ερμηνεύσιμων σημάτων που αντιστοιχούν στον καρδιακό ρυθμό και την αναπνοή. Ο κύριος αλγόριθμος που χρησιμοποιείται για το διαχωρισμό των σημάτων είναι ο FastICA. / -

Στην παρούσα διπλωματική εργασία γίνεται μελέτη της μεθόδου ανάλυσης ανεξάρτητων συνιστωσών στο ηλεκτροοεγκεφαλογράφημα. Αφού εξετάσουμε το κομμάτι της φυσιολογίας του εγκεφάλου θα δοθεί ένα μαθηματικό υπόβαθρο της ανάλυσης ανεξάρτητων συνιστωσών. Στη συνέχεια θα γίνει μια βιβλιογραφική έρευνα στη σύγκριση αλγορίθμων της ανάλυσης ανεξάρτητων συνιστωσών όταν εφαρμόζονται σε ηλεκτροεγκεφαλογραφήματα με στόχο την βέλτιστη εξαγωγή παρασίτων. Στο τέλος θα γίνει εφαρμογή της μεθόδου της ανάλυσης ανεξάρτητων συνιστωσών σε πραγματικά δεδομένα ηλεκτροεγκεφαλογραφήματος 64 καναλιών μέσω του περιβάλλοντος του Matlab. Στόχος της εφαρμογής αυτής είναι ο διαχωρισμός των ανεξάρτητων συνιστωσών μη-εγκεφαλικής προέλευσης και η αφαίρεση τους από τα αρχικά δεδομένα. / This thesis is a study of the independent componenet analysis in electroencephalogram. After looking at the piece of brain physiology we will give a mathematical framework of independent component analysis. Then it will become a literature search in the comparison algorithms of Independent Component Analysis in EEGs when applied with a view to optimal extraction of artifacts. At the end will be the method of independent component analysis to real EEG data 64 channels through the environment of Matlab. The aim of this application is the separation of independent components of non brain activity and removing them from the original data.

This report intends to verify the possibility that the FastICA algorithm could be applied to the GPS system to eliminate the impulsive noise from the receiver end. As the impulsive noise is so unpredictable in its pattern and of great energy level to swallow the signal we need, traditional signal selection methods exhibit no much use in dealing with this problem. Blind Source Separation seems to be a good way to solve this, but most of the other BSS algorithms beside FastICA showed more or less degrees of dependency on the pattern of the noise. In this thesis, the basic mathematic modelling of this advanced algorithm, along with the principles of the commonly used fast independent component analysis (fastICA) based on fixed-point algorithm are discussed. To verify that this method is useful under industrial use environment to remove the impulsive noises from digital BPSK modulated signals, an observation signal mixed with additive impulsive noise is generated and separated by fastICA method. And in the last part of the thesis, the fastICA algorithm is applied to the GPS receiver modeled in the SoftGNSS project and verified to be effective in industrial applications. The results have been analyzed. / 6

Identify some digestive features in people through Electrogastrogram (EGG) is important because this is a cheap, non-invasive and less bother way than traditional endoscopy procedure. This work evaluates the learning behavior of Artificial Neural Networks (ANN) and Hidden Markov Model (HMM) on components extracted by Independent Component Analysis (ICA) algorithms. In this research, an experiment was made with statistical analysis that shows the relationship between neutral, negative or positive images and digestive reactions. Training some classifiers with an EGG signal database, where the emotional states of individuals are known during processing, would it be possible to carry out the other way? Meaning, just from the EGG signal, estimate the emotional state of individuals. The initial challenge is to treat the EGG signal, which is mixed with the signals from other organs such as heart and lung. For this, the FastICA and Tensorial Methods algorithms were used, in order to produce a set of independent components, where one can identify the stomach component. Then, the EGG signal classification is performed through ANN and HMM models. The results have shown that extracting only the stomach signal component before the experiment can reduce the learning error rate in classifiers. / Identificar características digestivas de pessoas através da Eletrogastrografia (EGG) é importante pois esta costuma ser uma opção barata, não-invasiva e incomoda menos que o tradicional procedimento de Endoscopia. Este trabalho avalia o comportamento do aprendizado das Redes Neurais Artificiais (RNA) e do Modelo Oculto de Markov (HMM) diante de componentes extraídas por algoritmos de Análise de Componentes Independentes (ICA). Nesta pesquisa é realizado um experimento com análise estatística cujo objetivo apresenta a relação entre a visualização de imagens neutras, negativas ou positivas e as reações digestivas. Treinando alguns classificadores com uma base de dados de sinais EGG, onde se conhece os estados emocionais dos indivíduos durante a sua obtenção, seria possível realizar o caminho inverso? Em outras palavras, apenas a partir dos sinais EGG, pode-se estimar o estado emocional de indivíduos? O desafio inicial é tratar o sinal EGG que encontra-se misturado aos sinais de outros órgãos como coração e pulmão. Para isto foi utilizado o algoritmo FastICA e os métodos tensoriais, com o intuito de produzir um conjunto de componentes independentes onde se possa identificar a componente do estômago. Em seguida, a classifição do sinal EGG é realizada por meio dos modelos de RNA e HMM. Os resultados mostraram que classificar apenas as componentes com mais presença da frequência do sinal do estômago pode reduzir a taxa de erro do aprendizado dos classificadores no experimento realizado.

La première partie de cette thèse est consacrée aux mélanges convolutifs (sur-)déterminés. Nous cherchons à étendre l'algorithme FastICA aux mélanges convolutifs en proposant un algorithme à point fixe opérant dans le domaine temporel. Nous introduisons un processus de blanchiment spatio-temporel non-causal, qui, en initialisant les paramètres d'extraction d'une façon particulière, permet d'utiliser des itérations d'optimisation de type point fixe. L'estimation des contributions dans les observations est réalisée grâce à un critère quadratique optimisé par un filtre de Wiener non-causal.<br /> Dans la deuxième partie, consacrée aux mélanges sous-déterminés instantanés et convolutifs, nous cherchons aussi à étendre l'algorithme FastICA en nous basant sur le concept de séparation différentielle. Nous proposons des procédures de blanchiment différentiel des observations qui permettent l'emploi d'itérations à point fixe pour séparer entre elles des sources dites utiles. En convolutif, les contributions sont estimées au moyen d'un filtre de Wiener différentiel non-causal. <br /> La troisième partie, consacrée aux mélanges instantanés de sources contenant un grand nombre d'échantillons, met d'abord en évidence une limitation de l'algorithme FastICA, qui, à chaque itération d'optimisation, calcule des statistiques sur l'ensemble des échantillons disponibles. Le critère du kurtosis étant polynômial relativement aux coefficients d'extraction, nous proposons une procédure d'identification de ce polynôme qui permet d'effectuer l'optimisation dans un espace de calcul moins gourmand. Ce principe est ensuite appliqué aux mélanges instantanés sous-déterminés.

Το θέμα της κατασκευής ενός συστήματος αναγνώρισης προτύπων για τα ηχητικά σήματα ενός ανθρώπου που κοιμάται είναι ένα από τα ανοιχτά ζητήματα της Βιοιατρικής. Στην παρούσα διπλωματική εξετάζουμε την εξαγωγή ερμηνεύσιμων σημάτων που αντιστοιχούν στον καρδιακό ρυθμό, την αναπνοή και το ροχαλητό. Χρησιμοποιούμε μεθόδους Ανάλυσης σε Ανεξάρτητες Συνιστώσες και μεθόδους Τυφλού Διαχωρισμού που εκμεταλεύονται Στατιστικές Δεύτερης Τάξης. Συμπεραίνουμε ότι οι δεύτερες είναι οι πλέον κατάλληλες όταν συνοδεύονται από ένα στάδιο προεπεξεργασίας που αφορά ανάλυση σε ζώνες συχνοτήτων. / The design of a non-intrusive Pattern Recognition System to estimate the sleep sounds is an open problem of Bioengineering. We use recordings from body-sensors to estimate the heart beat, the breathing and the snoring. In this thesis we examine the effectiveness of Independent Component Analysis for this Blind Source Separation Problem and we compare it with methods that perform Source Separation using Second Order Statistics. We take into account the temporal structure of the sources as well as the presence of noise. Our system is greatly improved by a preprocessing stage of targeted subband decomposition which uses a priori knowledge about the sources. We propose an efficient solution to this problem which is confirmed by medical data.

Submitted by Jesiel Ferreira Gomes (jesielgomes@ufcg.edu.br) on 2018-04-20T20:53:07Z No. of bitstreams: 1 SHIRLENE KELLY SANTOS CARMO – TESE (PPGEQ) 2015.pdf: 3441674 bytes, checksum: 2a66c0c04d01e56f10189d8b206ebc1c (MD5) / Made available in DSpace on 2018-04-20T20:53:07Z (GMT). No. of bitstreams: 1 SHIRLENE KELLY SANTOS CARMO – TESE (PPGEQ) 2015.pdf: 3441674 bytes, checksum: 2a66c0c04d01e56f10189d8b206ebc1c (MD5) Previous issue date: 2015-02-06 / Capes / Grande parte das indústrias apresenta complexidade no que diz respeito ao seu modo de operação. A fim de reduzir os problemas relacionados ao forte acoplamento existente nesses processos, a busca pela incorporação de dispositivos de inteligência artificial vem apresentando uma tendência crescente nos últimos anos. Devido à complexidade de operação e controle em processos multivariáveis, o diagnóstico e monitoramento de falhas nos processos tornaram-se cada vez mais difícil, com isso a aplicação destes dispositivos tem alcançado resultados satisfatórios em relação aos procedimentos executados com operadores humanos. A análise de componentes independentes (ICA) é uma técnica de separação de sinais que se baseia no uso de estatísticas de ordem superior para estimar cada uma das fontes desconhecidas por meio da observação de diversas misturas geradas a partir destas fontes. Embora sejam encontrados trabalhos recentes sobre a utilização do ICA em processos industriais, apenas dois trabalhos até o presente momento, foram aplicados em processos envolvendo colunas de destilação. O presente trabalho tem como objetivo propor uma estratégia de controle a uma coluna de destilação de alta pureza. A estratégia é baseada na técnica de separação de sinais ICA, tornando as malhas de controle desacopladas e facilitando assim o desempenho do controle. O desempenho do sistema de controle utilizando a técnica apresentou excelentes resultados em relação a uma estrutura convencional sem desacoplamento. As estruturas de controle foram implementadas em ambiente Aspen Plus DynamicsTM e Simulink/ Matlab®. O processo foi estruturado em ambiente Aspen Plus Dynamics™ e os controladores foram implementados no Simulink. / Much of the industry presents complexity with regard to its mode of operation. In order to reduce the problems related to existing strong engagement in these processes, the search for the incorporation of artificial intelligence devices has shown an increasing trend in recent years. Due to the complexity of operation and control in multivariate processes, the diagnosis and fault monitoring in the processes have become increasingly difficult, thus the application of these devices has achieved satisfactory results in relation to procedures performed with human operators. The independent component analysis (ICA) is a signal separation technique that is based on the use of higher order statistics to estimate each of the unknown source by observing various mixtures generated from these sources. Although found recent work on the use of the ICA in industrial processes, only two studies to date, have been applied in cases involving distillation columns. This paper aims to propose a control strategy to a high purity distillation column. The strategy is based on the ICA signal separation technique, making decoupled control loops, thus facilitating control performance. The performance of the control system using the technique showed excellent results compared to a conventional structure without decoupling. The control structures have been implemented in Aspen Plus Dynamics™ and Simulink / Matlab® environment. The process was structured environment Aspen Plus Dynamics™ and the controls were implemented in Simulink.

Abstract Resting-state (RS) measurements during functional magnetic resonance imaging (fMRI) have become an established approach for studying spontaneous brain activity. RS-fMRI results are often obtained using explorative approaches like spatial independent component analysis (sICA). These approaches and their software implementations are rarely evaluated extensively or specifically concerning RS-fMRI. Trust is placed in the software that they will work according to the published method descriptions. Many methods and parameters are used despite the lack of test data, and the validity of the underlying models remains an open question. A substantially greater number of evaluations would be needed to ensure the quality of exploratory RS-fMRI analyses. This thesis investigates the applicability of sICA methodology and software in the RS-fMRI context. The experiences were used to formulate general guidelines to facilitate future method evaluations. Additionally, a novel multiple comparison correction (MCC) method, Maxmad, was devised for adjusting evaluation results statistically. With regard to software considerations, the source code of FSL Melodic, popular sICA software, was analyzed against its published method descriptions. Unreported and unevaluated details were found, which implies that one should not automatically assume a correspondence between the literature and the software implementations. The method implementations should rather be subjected to independent reviews. An experimental contribution of this thesis is that the credibility of the emerging sliding window sICAs has been improved by the validation of sICA related preprocessing procedures. In addition to that, the estimation accuracy regarding the results in existing RS-fMRI sICA literature was also shown not to suffer even though repeatability tools like Icasso have not been used in their computation. Furthermore, the evidence against conventional sICA model suggests the consideration of different approaches to analysis of RS-fMRI. The guidelines developed for facilitation of evaluations include adoption of 1) open software development (improved error detection), 2) modular software designs (easier evaluations), 3) data specific evaluations (increased validity), and 4) extensive coverage of parameter space (improved credibility). The proposed Maxmad MCC addresses a statistical problem arising from broad evaluations. Large scale cooperation efforts are proposed concerning evaluations in order to improve the credibility of exploratory RS-fMRI methods. / Tiivistelmä Aivoista toiminnallisella magneettikuvantamisella (engl. functional magnetic resonance imaging, fMRI) lepotilassa tehdyt mittaukset ovat saaneet vakiintuneen aseman spontaanin aivotoiminnan tutkimuksessa. Lepotilan fMRI:n tulokset saadaan usein käyttämällä exploratiivisia menetelmiä, kuten spatiaalista itsenäisten komponenttien analyysia (engl. spatial independent component analysis, sICA). Näitä menetelmiä ja niiden ohjelmistototeutuksia evaluoidaan harvoin kattavasti tai erityisesti lepotilan fMRI:n kannalta. Ohjelmistojen luotetaan toimivan menetelmäkuvausten mukaisesti. Monia menetelmiä ja parametreja käytetään testidatan puuttumisesta huolimatta, ja myös menetelmien taustalla olevien mallien pätevyys on edelleen epäselvä asia. Eksploratiivisten lepotilan fMRI-datan analyysien laadun varmistamiseksi tarvittaisiin huomattavasti nykyistä suurempi määrä evaluaatioita. Tämä väitöskirja tutki sICA-menetelmien ja -ohjelmistojen soveltuvuutta lepotilan fMRI-tutkimuksiin. Kokemuksien perusteella luotiin yleisiä ohjenuoria helpottamaan tulevaisuuden menetelmäevaluaatioita. Lisäksi väitöskirjassa kehitettiin uusi monivertailukorjausmenetelmä, Maxmad, evaluaatiotulosten tilastolliseen korjaukseen. Tunnetun sICA-ohjelmiston, FSL Melodicin, lähdekoodi analysoitiin suhteessa julkaistuihin menetelmäkuvauksiin. Analyysissa ilmeni aiemmin raportoimattomia ja evaluoimattomia menetelmäyksityiskohtia, mikä tarkoittaa, ettei kirjallisuudessa olevien menetelmäkuvausten ja niiden ohjelmistototeutusten välille pitäisi automaattisesti olettaa vastaavuutta. Menetelmätoteutukset pitäisi katselmoida riippumattomasti. Väitöskirjan kokeellisena panoksena parannettiin liukuvassa ikkunassa suoritettavan sICA:n uskottavuutta varmistamalla sICA:n esikäsittelyjen oikeellisuus. Lisäksi väitöskirjassa näytettiin, että aiempien sICA-tulosten tarkkuus ei ole kärsinyt, vaikka niiden estimoinnissa ei ole käytetty toistettavuustyökaluja, kuten Icasso-ohjelmistoa. Väitöskirjan tulokset kyseenalaistavat myös perinteisen sICA-mallin, minkä vuoksi tulisi harkita siitä poikkeavia lähtökohtia lepotilan fMRI-datan analyysiin. Evaluaatioiden helpottamiseksi kehitetyt ohjeet sisältävät seuraavat periaatteet: 1) avoin ohjelmistokehitys (parantunut virheiden havaitseminen), 2) modulaarinen ohjelmistosuunnittelu (nykyistä helpommin toteutettavat evaluaatiot), 3) datatyyppikohtaiset evaluaatiot (parantunut validiteetti) ja 4) parametriavaruuden laaja kattavuus evaluaatioissa (parantunut uskottavuus). Ehdotettu Maxmad-monivertailukorjaus tarjoaa ratkaisuvaihtoehdon laajojen evaluaatioiden tilastollisiin haasteisiin. Jotta lepotilan fMRI:ssä käytettävien exploratiivisten menetelmien uskottavuus paranisi, väitöskirjassa ehdotetaan laaja-alaista yhteistyötä menetelmien evaluoimiseksi.