Tesis sobre el tema "Analyse en composantes disjointes"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte los 50 mejores tesis para su investigación sobre el tema "Analyse en composantes disjointes".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Explore tesis sobre una amplia variedad de disciplinas y organice su bibliografía correctamente.
Zaghloul, Sara. "Application du DCA aux Radars de Surveillances Secondaires". Electronic Thesis or Diss., Reims, 2024. http://www.theses.fr/2024REIMS017.
Texto completoThe objective of this thesis was to develop a fast algorithm to separate a mixture of Secondary Surveillance Radar (SSR) signals. This mixture may include different modes, such as Mode A/C and Mode S, which complicate the separation due to their varied formats and different coding characteristics. During this thesis, three methods were developed using a relatively discrete criterion, the Disjoint Component Analysis (DCA), which aims to separate sources based on maximizing the disjointness between them.The first is a post-processing approach that uses linear algebra to solve the problems encountered when applying the real-valued version of DCA. However, the application of this method can pose several problems, including signal loss, residual mixing, and signal dependencies. Therefore, we concluded that it was necessary to develop a method that considers SSR signals in their original complex-valued format.The second method aims to demonstrate the effectiveness of the DCA criterion for SSR signals, using an exhaustive search approach while considering signals in their complex format. This method succeeds in separating signals with a high degree of accuracy but is computationally expensive.The third proposed method optimizes the search for the minimum using a gradient descent algorithm, which significantly improves computational efficiency while maintaining similar quality of results.These algorithms were tested in simulations and compared with various algorithms from the literature, to evaluate their performance as a function of different reception parameters. Finally, they were tested on real-world data
Barbedor, Pascal. "Analyse en composantes indépendantes par ondelettes". Phd thesis, Université Paris-Diderot - Paris VII, 2006. http://tel.archives-ouvertes.fr/tel-00119428.
Texto completoLe problème principal de l'ACI est d'estimer la matrice A, à partir de l'observation d'un échantillon i.i.d. de X, pour atteindre S qui constitue un système explicatif meilleur que X dans l'étude d'un phénomène particulier. Le problème se résout généralement par la minimisation d'un certain critère, issu d'une mesure de dépendance.
L'ACI ressemble à l'analyse en composantes principales (ACP) dans la formulation du problème. Dans le cas de l'ACP on cherche des composantes non corrélées, c'est-à-dire indépendantes par paire à l'ordre 2 ; dans le cas de l'ACI on cherche des composantes mutuellement indépendantes, ce qui est beaucoup plus contraignant; dans le cas général, il n'existe plus de solution algébrique simple. Les principaux problèmes d'identification de A sont évités par un certain nombre de conventions adoptées dans le modèle ACI classique.
L'approche qui est proposée dans cette thèse est du type non paramétrique. Sous des hypothèses de type Besov, on étudie plusieurs estimateurs d'un critère de dépendance exact donné par la norme L2 de la différence entre une densité et le produit de ses marges. Ce critère constitue une alternative à l'information mutuelle qui représentait jusqu'ici le critère exact de référence de la plupart des méthodes ACI.
On donne une majoration de l'erreur en moyenne quadratique de différents estimateurs du contraste L2. Cette majoration prend en compte le biais d'approximation entre le Besov et l'espace de projection qui, ici, est issu d'une analyse multirésolution (AMR) générée par le produit tensoriel d'ondelettes de Daubechies. Ce type de majoration avec prise en compte du biais d'approximation est en général absent des méthodes non paramétriques récentes en ACI (méthodes kernel, information mutuelle).
Le critère en norme L2 permet de se rapprocher de problèmes déjà connus dans la littérature statistique, estimation de l'intégrale de f au carré, tests d'homogénéité en norme L2, résultats de convergence d'estimateurs adoptant un seuillage par bloc.
On propose des estimateurs du contraste L2 qui atteignent la vitesse minimax optimale du problème de intégrale de f au carré. Ces estimateurs de type U-statistique ont des complexités numériques quadratique en n, ce qui peut poser un problème pour la minimisation du contraste à suivre, en vue d'obtenir l'estimation concrète de la matrice A. En revanche, ces estimateurs admettent une forme de seuillage par bloc où la connaissance de la régularité s de la densité multivariée sous-jacente est inutile pour obtenir une vitesse optimale.
On propose un estimateur de type plug-in dont la vitesse de convergence est sous-optimale mais qui est de complexité numérique linéaire en n. L'estimateur plug-in admet aussi une forme seuillée terme à terme, qui dégrade la vitesse de convergence mais permet d'obtenir un critère auto-adaptatif. Dans sa version linéaire, l'estimateur plug-in semble déjà quasiment auto-adaptatif dans les faits, c'est-à-dire que sous la contrainte 2^{jd} < n, où d est la dimension du problème et n le nombre d'observations, la majorité des résolutions j permettent d'estimer A après minimisation.
Pour obtenir ces résultats on a été amené à développer une technique combinatoire spécifique permettant de majorer le moment d'ordre r d'une U-statistique ou d'une V-statistique. Les résultats classiques sur les U-statistiques ne sont en effet pas directement utilisables et pas facilement adaptables dans le contexte d'étude de la thèse. La méthode développée est utilisable dans d'autres contextes.
La méthode par ondelettes s'appuie sur le paradigme usuel estimation d'un critère de dépendance, puis minimisation. On étudie donc dans la thèse les éléments permettant de faciliter la minimisation. On donne notamment des formulations du gradient et du hessien de l'estimateur du contraste qui se prêtent à un changement de résolution par simple filtrage et qui se calculent selon une complexité équivalente à celle de l'évaluation de l'estimateur lui même.
Des simulations proposées dans la thèse confirment l'applicabilité de la méthode et donnent des résultats excellents. Tous les éléments nécessaires à l'implémentation de la méthode, et le code commenté des parties clefs de la programmation (notamment des algorithmes d-dimensionnels) figurent également dans le document.
Le, Borgne Hervé. "Analyse de scènes naturelles par Composantes Indépendantes". Phd thesis, Grenoble INPG, 2004. http://tel.archives-ouvertes.fr/tel-00005925.
Texto completoJaupi, Luan. "Methodes robustes en analyse en composantes principales". Paris, CNAM, 1992. http://www.theses.fr/1992CNAM0151.
Texto completoPatanchon, Guillaume. "Analyse multi-composantes d'observations du fond diffus cosmologique". Phd thesis, Université Pierre et Marie Curie - Paris VI, 2003. http://tel.archives-ouvertes.fr/tel-00004512.
Texto completoBiton, Anne. "Analyse en composantes indépendantes du transcriptome de cancers". Thesis, Paris 11, 2011. http://www.theses.fr/2011PA11T028.
Texto completoPractice of gene expression data analysis shows that it is advantageous to analyze biologicalprocesses in terms of modules rather than simply consider gene one by one. In this project, we conducted anunsupervised analysis of the gene expression data of several cohorts of urothelial tumors, applying theIndependent Component Analysis method. Several studies demonstrated the outperformance of ICA overPCA and clustering-based methods in obtaining a more realistic decomposition of the expression data intoconsistent patterns of coexpressed genes associated with the studied phenotypes[1, 2, 3, 4].Urothelial tumors arise and evolve through two distinct pathways which radically differ on their probabilityof progression to muscle invasion. Except the mutation of FGFR3 in the less aggressive group, theunderlying molecular processes have not been completely identified. The first and main objective of the PhDthesis was dedicated to the biological interpretation of the different independent components to help toconfirm and extend the list of biological processes known to be involved in bladder cancer.Each independent component (IC) is characterized by a list of gene projections on the one hand and weightedcontributions of tumor samples on the other hand. By combining biological expertise and comparison of theassociated list of genes to known pathways, and jointly studying the association of the components tomolecular and clinical annotations, we have been able to differentiate components that were caused bytechnical factors, such as surgical sampling, from those having consistent biological interpretationin terms of tumor biology. Moreover, among the biologically meaningful signals, this analysis allowed us todifferentiate the signals from stroma of the tumor, like immune response mediated by B- and T-lymphocytes,from the signals produced by the tumors themselves, like signals related to proliferation, or differentiation.The clustering of the tumor samples according to their contributions on some ICs can be closely associated toanatomo-clinical annotations, and highlighted new potential subtypes of tumors which suggest existence ofadditional pathways of bladder cancer progression. Similarly, the study of the contributions of preestablishedgroups of tumors based on clinical or molecular criteria showed different levels of stromacontamination between FGFR3 non-mutated and mutated tumors. The reproducibility of the components wasinvestigated using correlation graphs. The major part of the interpreted ICs was validated on threeindependent bladder cancer datasets, and several of them were also identified in an urothelial cancer celllines data set.A second study about retinoblastoma gave us the occasion to show that we can take advantage ofICA in various contexts. Retinoblastoma is initiated by the loss of both alleles of the RB1 tumor suppressorgene. Although necessary for initiation, other genomic events, that remain to be identified, are needed for theprogression of the disease [5]. We observed, as it was previously described [6], an association between age ofthe patients and levels of genomic aberrations, the younger patients having fewer alterations than the olderpatients, which generally present gain of 1q and loss of 16q. We showed that this tendency of the tumors tobe clustered into two groups of age can also be observed on the expression data by applying ICA whose oneof the component was highly correlated to the age of the patients. These results suggest the existence of twopathways of progression in retinoblastoma.The analysis of high throughput data provides many lists of genes. To interpret them, a possibility isto study the latest related publications grouped by pre-defined group of topics (function, cellular location...).To that aim, in a third study, we introduced a web-based Java application tool named GeneValorization whichgives a clear and handful overview of the bibliography corresponding to one particular gene list [7]
Narozny, Michel. "Analyse en composantes indépendantes et compression de données". Paris 11, 2005. http://www.theses.fr/2005PA112268.
Texto completoIn this thesis we are interested in the performances of independent component analysis (ICA) when it is used for data compression. First we show that the ICA transformations yield poor performances compared to the Karhunen-Loeve transform (KIT) for the coding of some continuous-tone images and a musical signal, but can outperform the KTL on some synthetic signals. In medium-to-high (resp. Low) bit rate coding, the bit-rate measured is the empirical first (resp. Second, fourth and ninth) order entropy. The mean square error between the original signal and that reconstructed is used for the evaluation of the distortion. Then we show that for non Gaussian signals the problem of finding the optimal linear transform in transform coding is equivalent to finding the solution of a modified ICA problem. Two new algorithms, GCGsup and ICAorth, are then proposed to compute the optimal linear transform and the optimal orthogonal transform, respectively. In our simulations, we show that GCGsup and ICAorth can outperform the KLT or some continuous-tone images and some synthetic signals. Finally, we are also interested in a multicomponent images coding scheme which employs a wavelet transform for reducing the spatial redundancy and the transformations returned by GCGsup et ICAorth for reducing the spectral redundancy. In this case, further work has to be done in order to find some images whose compression using the new transforms is significantly better than that obtained with the TKL
Ibazizen, Mohamed. "Contribution à l'étude d'une analyse en composantes principales". Grenoble 2 : ANRT, 1986. http://catalogue.bnf.fr/ark:/12148/cb375984301.
Texto completoHamdoun, Omar. "Détection et ré-identification de piétons par points d'intérêt entre caméras disjointes". Phd thesis, École Nationale Supérieure des Mines de Paris, 2010. http://pastel.archives-ouvertes.fr/pastel-00566417.
Texto completoFaouzi, Nour-Eddin el-Faouzi. "Extensions non-linéaires de l'analyse en composantes principales". Montpellier 2, 1992. http://www.theses.fr/1992MON20118.
Texto completoChenguiti, Khalid. "Analyse en composantes principales fonctionnelle cas des prises alimentaires". Mémoire, Université de Sherbrooke, 2006. http://savoirs.usherbrooke.ca/handle/11143/4722.
Texto completoBesse, Philippe. "Approximation spline et optimalité en analyse en composantes principales". Toulouse 3, 1989. http://www.theses.fr/1989TOU30190.
Texto completoIbazizen, Mohamed. "Contribution a l'etude d'une analyse en composantes principales robuste". Toulouse 3, 1986. http://www.theses.fr/1986TOU30101.
Texto completoHarkat, Mohamed-Faouzi. "Détection et localisation de défauts par analyse en composantes principales". Vandoeuvre-les-Nancy, INPL, 2003. http://docnum.univ-lorraine.fr/public/INPL_T_2003_HARKAT_M_F.pdf.
Texto completoThe aim of this thesis is to study the fault detection and isolation using principal components analysis (PCA). In the first chapter the fundamental principles of linear principal componerit analysis are presented. PCA is used to model normal process behaviour. In the second chapter the problem of fault detection and isolation based on linear PCA is tackled. On the basis of the analysis of the classical detection indices, a new fault detection index based on the last principal components is developed. For fault isolation, the classical methods, using for instance the reconstruction principle or the contribution calculation, are adapted for the proposed fault detection index. The third chapter is focused on the nonlinear PCA. An extension of the PCA for nonlinear systems, combining principal curves algorithm and RBF networks, is proposed. For the determination of the number of principal components to be kept in the NLPCA model, we propose an extension of the unreconstructed variance criteria in the non-linear case. . Finally, an application, carried out in collaboration with air quality monitoring network in Lorraine AIRLOR, is presented in the fourth chapter. This application concerns the sensqr fault detection and isolation of this network by using the fault detection and isolation procedure developed in the linear case
HARKAT, Mohamed-Faouzi. "Détection et localisation de défauts par analyse en composantes principales". Phd thesis, Institut National Polytechnique de Lorraine - INPL, 2003. http://tel.archives-ouvertes.fr/tel-00005283.
Texto completoJutten, Christian. "Calcul neuromimétique et traitement du signal : analyse en composantes indépendantes". Grenoble INPG, 1987. http://www.theses.fr/1987INPG0068.
Texto completoLiu, Xiaoqing. "Analyse d'images couleur en composantes indépendantes par réseau de neurones". Grenoble INPG, 1991. http://www.theses.fr/1991INPG0120.
Texto completoJutten, Christian. "Calcul neuromimétique et traitement du signal analyse en composantes indépendantes /". Grenoble 2 : ANRT, 1987. http://catalogue.bnf.fr/ark:/12148/cb376063328.
Texto completoMakosso, Kallyth Sun. "Analyse en composantes principales de variables symboliques de type histogramme". Paris 9, 2010. https://bu.dauphine.psl.eu/fileviewer/index.php?doc=2010PA090045.
Texto completoThe objective of this thesis is to extend Principal Component Analysis PCA to symbolic histogram variable. Three firsts are methods I, II and III. Method I extends Nagabhushan and Kumar (2007) method which presents some similarity with the Multiple Factorial Analysis. Method II bases himself on the matrix scalar product and tries to come at the end of an inconvenience of Method I concerning the problem of the definition of a compromise system of axes. Method III uses an operator we call 2. However, the usage of this operator requires the appeal to a normalizing transformation of variables. It allows taking into account the compositional nature of the relative frequencies. However, it is necessary to specify that in the methods based on operators use strong hypothesis about bins of histogram. The methods based on the transformation of histograms in intervals (Method IV) get overcome this constraint about bins of histogram. Then we propose methods V which are a synthesis of tools proposed in method III and those proposed in method IV. Different variants of method V are less boring in terms of cost than method method IV, but method V estimates the dispersion of the individuals less well than method IV. We overcome this disadvantage by offering a version improved by method IV that we call method VI. The possible limits of method VI stem from the fact the usage of quantiles can lead to an overestimation of the variability of the individuals. By the end, we propose method VII which gives another approach to represent the variability of the individuals. In Method VII, we transform histograms into intervals via the rule of Tchebychev. Then, we determine the length of every interval and we project in additional lengths of the intervals on the principal axes of the averages of variables
Mnassri, Baligh. "Analyse de données multivariées et surveillance des processus industriels par analyse en composantes principales". Phd thesis, Aix-Marseille Université, 2012. http://tel.archives-ouvertes.fr/tel-00749282.
Texto completoDans l'objectif d'un choix optimal du modèle ACP, une étude comparative de quelques critères connus dans la littérature nous a permis de conclure que le problème rencontré est souvent lié à une ignorance des variables indépendantes et quasi-indépendantes. Dans ce cadre, nous avons réalisé deux démonstrations mettant en évidence les limitations de deux critères en particulier la variance non reconstruite (VNR). En s'appuyant sur le principe d'une telle variance, nous avons proposé trois nouveaux critères. Parmi eux, deux ont été considérés comme étant empiriques car seule l'expérience permettra de prouver leur efficacité. Le troisième critère noté VNRVI représente un remède à la limitation du critère VNR. Une étude de sa consistance théorique a permis d'établir les conditions garantissant l'optimalité de son choix. Les résultats de simulation ont validé une telle théorie en prouvant ainsi que le critère VNRVI étant plus efficace que ceux étudiés dans cette thèse.
Dans le cadre d'un diagnostic de défauts par ACP, l'approche de reconstruction des indices de détection ainsi que celle des contributions ont été utilisées. A travers une étude de généralisation, nous avons étendu le concept d'isolabilité de défauts par reconstruction à tout indice quadratique. Une telle généralisation nous a permis d'élaborer une analyse théorique d'isolabilité de défauts par reconstruction de la distance combinée versus celles des indices SPE et T2 de Hotelling en mettant en avant l'avantage de l'utilisation d'une telle distance. D'autre part, nous avons proposé une nouvelle méthode de contribution par décomposition partielle de l'indice SPE. Cette approche garantit un diagnostic correct de défauts simples ayant de grandes amplitudes. Nous avons également étendu une méthode de contribution classiquement connue par la RBC au cas multidimensionnel. Ainsi, la nouvelle forme garantit un diagnostic correct de défauts multiples de grandes amplitudes. En considérant la complexité de défauts, nous avons exploité la nouvelle approche de contribution RBC afin de proposer une nouvelle qui s'appelle RBCr. Cette dernière s'appuie sur un seuil de tolérance pour l'isolation de défauts. Une analyse de diagnosticabilité basée sur la RBCr montre que celle-ci garantit l'identification des défauts détectables. Ces derniers sont garantis isolables si leurs amplitudes satisfont les mêmes conditions d'isolabilité établies pour l'approche de reconstruction des indices.
Viron, Cyril. "Analyse en composantes indépendantes pour la caractérisation d'images hyperspectrales en télédétection". Mémoire, École de technologie supérieure, 2006. http://espace.etsmtl.ca/486/1/VIRON_Cyril.pdf.
Texto completoMerigot, Bastien. "Analyse multi-composantes de la diversité spécifique : applications aux peuplements marins". Aix-Marseille 2, 2008. http://theses.univ-amu.fr.lama.univ-amu.fr/2008AIX22039.pdf.
Texto completoFor many years, biodiversity has been of growing interest, in both the lay and the scientific community, with regard to its assessment and preservation. In this context, the present study proposes a methodological framework which explicitly takes into account the multi-component character of diversity with the aim of improving diversity assessment and management. This approach allows to choose only complementary diversity indices that can serve to describe the various facets of the species diversity studied. Our results are important for monitoring the diversity of assemblages as it can serve as a basis for drawing up a preliminary shortlist of indices and thus facilitate the description of these systems. Finally, this approach could allow better assessment of the relative suitability of the various descriptors studied as indicators for monitoring settlements and ecosystems
Viron, Cyril. "Analyse en composantes indépendantes pour la caractérisation d'images hyperspectrales en télédétection /". Thèse, Montréal : École de technologie supérieure, 2006. http://proquest.umi.com/pqdweb?did=1253509051&sid=1&Fmt=2&clientId=46962&RQT=309&VName=PQD.
Texto completo"Mémoire présenté à l'École de technologie supérieure comme exigence partielle à l'obtention de la maîtrise en génie de la production automatisée". CaQMUQET Bibliogr.: f. [222]-228. Également disponible en version électronique. CaQMUQET
Décarie, Yann. "Analyse en composantes principales et analyse discriminante fonctionnelles appliquées à des données de prises alimentaires animales". Mémoire, Université de Sherbrooke, 2011. http://savoirs.usherbrooke.ca/handle/11143/4899.
Texto completoAlbera, Laurent. "Identification autodidacte de mélanges potentiellement sous-déterminés". Nice, 2003. http://www.theses.fr/2003NICE4066.
Texto completoAbbas, Arezki. "Intégration de composantes passives dans un algorithme de synthèse de controleurs". Sherbrooke : Université de Sherbrooke, 2002.
Buscar texto completoBetoule, Marc. "Analyse des données du fond diffus cosmologique : simulation et séparation de composantes". Phd thesis, Observatoire de Paris, 2009. http://tel.archives-ouvertes.fr/tel-00462157.
Texto completoAnani, Kwami Dodzivi. "Diagnostic de systèmes non linéaires par analyse en composantes principales à noyau". Thesis, Université de Lorraine, 2019. http://www.theses.fr/2019LORR0026/document.
Texto completoIn this thesis, the diagnosis of a nonlinear system was performed using data analysis. Initially developed to analyze linear system, Principal Component Analysis (PCA) is coupled with kernel methods for detection, isolation and estimation of faults' magnitude for nonlinear systems. Kernel PCA consists in projecting data using a nonlinear mapping function into a higher dimensional space called feature space where the linear PCA is applied. Due to the fact that the projections are done using kernels, the detection can be performed in the feature space. However, estimating the magnitude of the fault requires the resolution of a nonlinear optimization problem. The variables' contributions make it possible to isolate and estimate these magnitudes. The variable with the largest contribution may be considered as faulty. In our work, we proposed new methods for the isolation and estimation phases for which previous work has some limitations. The new proposed method in this thesis is based on contributions under constraints. The effectiveness of the developed methods is illustrated on the simulated continuous stirred tank reactor (CSTR)
Betoule, Marc. "Analyse des données du fond diffus cosmologique : simulation et séparation de composantes". Phd thesis, Observatoire de Paris (1667-....), 2009. https://theses.hal.science/tel-00462157v2.
Texto completoThe next generation of experiments dedicated to measuring temperature and polarization anisotropies of the microwave background radiation (CMB), inaugurated with the launch of the Planck satellite, will enable the detection and study of increasingly subtle effects. However, the superposition of astrophysical foreground emissions hinder the analysis of the cosmological signal and will contribute as the main source of uncertainty in the forthcoming measurements. An improved modeling of foreground emissions and the development of statistical methods to extract the cosmological information from this contamination are thus crucial steps in the scientific analysis of incoming datasets. In this work we describe the development of the Planck Sky Model, a tool for modeling and simulating the sky emission. We then make use of these simulations to develop and evaluate statistical treatments of foreground emission. We explore the efficiency of wavelet analysis on the sphere (needlets) in the domain of spectral estimation on incomplete data with inhomogeneous contamination, and design a method for treating small scales contamination induced by point sources in the Planck and WMAP data. We also study the impact of foregrounds on our ability to detect primordial gravitational waves (predicted by inflation) and offer forecasts of the performance of future dedicated experiments
Anani, Kwami Dodzivi. "Diagnostic de systèmes non linéaires par analyse en composantes principales à noyau". Electronic Thesis or Diss., Université de Lorraine, 2019. http://www.theses.fr/2019LORR0026.
Texto completoIn this thesis, the diagnosis of a nonlinear system was performed using data analysis. Initially developed to analyze linear system, Principal Component Analysis (PCA) is coupled with kernel methods for detection, isolation and estimation of faults' magnitude for nonlinear systems. Kernel PCA consists in projecting data using a nonlinear mapping function into a higher dimensional space called feature space where the linear PCA is applied. Due to the fact that the projections are done using kernels, the detection can be performed in the feature space. However, estimating the magnitude of the fault requires the resolution of a nonlinear optimization problem. The variables' contributions make it possible to isolate and estimate these magnitudes. The variable with the largest contribution may be considered as faulty. In our work, we proposed new methods for the isolation and estimation phases for which previous work has some limitations. The new proposed method in this thesis is based on contributions under constraints. The effectiveness of the developed methods is illustrated on the simulated continuous stirred tank reactor (CSTR)
Bry, Xavier. "Une méthodologie exploratoire pour l'analyse et la synthèse d'un modèle explicatif : l'Analyse en Composantes Thématiques". Paris 9, 2004. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=2004PA090055.
Texto completoDelville, Joël. "La décomposition orthogonale aux valeurs propres et l'analyse de l'organisation tridimensionnelle des écoulements turbulents cisaillés libres". Poitiers CEAT, 1995. http://www.theses.fr/1995POIT2270.
Texto completoDescales, Bernard. "Modélisation des propriétés de coupes d'hydrocarbures par spectroscopie proche-infrarouge". Aix-Marseille 3, 1989. http://www.theses.fr/1989AIX30060.
Texto completoZwald, Laurent. "PERFORMANCES STATISTIQUES D'ALGORITHMES D'APPRENTISSAGE : ``KERNEL PROJECTION MACHINE'' ET ANALYSE EN COMPOSANTES PRINCIPALES A NOYAU". Phd thesis, Université Paris Sud - Paris XI, 2005. http://tel.archives-ouvertes.fr/tel-00012011.
Texto completodes contributions à la communauté du machine learning en utilisant des
techniques de statistiques modernes basées sur des avancées dans l'étude
des processus empiriques. Dans une première partie, les propriétés statistiques de
l'analyse en composantes principales à noyau (KPCA) sont explorées. Le
comportement de l'erreur de reconstruction est étudié avec un point de vue
non-asymptotique et des inégalités de concentration des valeurs propres de la matrice de
Gram sont données. Tous ces résultats impliquent des vitesses de
convergence rapides. Des propriétés
non-asymptotiques concernant les espaces propres de la KPCA eux-mêmes sont également
proposées. Dans une deuxième partie, un nouvel
algorithme de classification a été
conçu : la Kernel Projection Machine (KPM).
Tout en s'inspirant des Support Vector Machines (SVM), il met en lumière que la sélection d'un espace vectoriel par une méthode de
réduction de la dimension telle que la KPCA régularise
convenablement. Le choix de l'espace vectoriel utilisé par la KPM est guidé par des études statistiques de sélection de modéle par minimisation pénalisée de la perte empirique. Ce
principe de régularisation est étroitement relié à la projection fini-dimensionnelle étudiée dans les travaux statistiques de
Birgé et Massart. Les performances de la KPM et de la SVM sont ensuite comparées sur différents jeux de données. Chaque thème abordé dans cette thèse soulève de nouvelles questions d'ordre théorique et pratique.
Zwald, Laurent. "Performances statistiques d'algorithmes d'apprentissage : "Kernel projection machine" et analyse en composantes principales à noyau". Paris 11, 2005. https://tel.archives-ouvertes.fr/tel-00012011.
Texto completoThis thesis takes place within the framework of statistical learning. It brings contributions to the machine learning community using modern statistical techniques based on progress in the study of empirical processes. The first part investigates the statistical properties of Kernel Principal Component Analysis (KPCA). The behavior of the reconstruction error is studied with a non-asymptotique point of view and concentration inequalities of the eigenvalues of the kernel matrix are provided. All these results correspond to fast convergence rates. Non-asymptotic results concerning the eigenspaces of KPCA themselves are also provided. A new algorithm of classification has been designed in the second part: the Kernel Projection Machine (KPM). It is inspired by the Support Vector Machines (SVM). Besides, it highlights that the selection of a vector space by a dimensionality reduction method such as KPCA regularizes suitably. The choice of the vector space involved in the KPM is guided by statistical studies of model selection using the penalized minimization of the empirical loss. This regularization procedure is intimately connected with the finite dimensional projections studied in the statistical work of Birge and Massart. The performances of KPM and SVM are then compared on some data sets. Each topic tackled in this thesis raises new questions
Sahmer, Karin. "Propriétés et extensions de la classification de variables autour de composantes latentes : application en évaluation sensorielle". Phd thesis, Université Rennes 2, 2006. http://tel.archives-ouvertes.fr/tel-00129227.
Texto completoTchana, Esther Petnga. "Application de l'analyse multidimensionnelle à la classification qualitative du café robusta en fonction de la maturité". Montpellier 2, 1986. http://www.theses.fr/1986MON20196.
Texto completoSockeel, Stéphane. "Détection de réseaux et étude de la dynamique des connectivités fonctionnelles cérébrales en EEG couplée avec l'IRMf". Paris 6, 2012. http://www.theses.fr/2012PA066598.
Texto completoThe main goal of this work is answering this two questions: could electroencephalography (EEG) detect large-scale functional networks? What contribution could bring EEG in the study of dynamics of connectivity? We have developed an original method specifically for EEG to detect large-scale functional networks in the brain. We first assume that BOLD signal in fMRI is correlated with the power of EEG signal and then, we applied a spatial Independent Components Analysis (sICA). Our method can be divided in six steps : preprocessing of EEG data, sources localization, power computation in five frequency bands, sICA for each subject, hierarchical clustering (group analysis), stepwise regression with fMRI templates. The main advantages in regards of the other classic methods (seed-based methods, atlas, …) are: a fully data-driven method, use of individual sICA and a group analysis, multifrequency analysis. We validated our method with a comparison with a fMRI study on healthy subjects during resting states and visual tasks. The networks detected with EEG and fMRI overlap. Then, we have studied the dynamics of connectivity inside the extracted EEG networks using three connectivity measures. The temporal precision of EEG can evaluate connection and information flow between this regions and give us acces to a frequency scale similar to that of brain's rythmes. This study has provided some interesting and promising results, for example a memory trace after a visual task in the visual network
Sahmer, Karin Carbon Michel Kunert Joachim. "Propriétés et extensions de la classification de variables autour de composantes latentes application en évaluation sensorielle /". Rennes : Université Rennes 2, 2007. http://tel.archives-ouvertes.fr/tel-00129227/fr.
Texto completoJausions-Picaud, Claire. "Analyse en composantes curvilignes et representation de donnees multidimensionnelles : application au routage adaptatif de messages". Grenoble INPG, 1999. http://www.theses.fr/1999INPG0080.
Texto completoCHAGUE, VINCENT. "Analyse des composantes du management technologique et de ses incidences sur les performances des pme". Limoges, 1995. http://www.theses.fr/1995LIMO0446.
Texto completoMakany, Roger Armand. "Techniques de validation en statistiques : application à l'analyse en composantes principales et à la régression". Paris 11, 1985. http://www.theses.fr/1985PA112222.
Texto completoIn this work, the results validation is being studied according to the stability proceeding from these results. The present study is based upon the stability of latent roots and subspaces in principal component analysis. As regards regression, the author has focused on the stability of the regression coefficients. Beyond the stability criteria put forward, there has been included a display of the software designed for data processing
Touati, Sami. "Estimation de statistiques d'ordre supérieur et de composantes cycliques de signaux cyclostationnaires à l'aide d'ondelettes". Paris 11, 2007. http://www.theses.fr/2007PA112341.
Texto completoGarbez, Morgan. "Construction de l'architecture et des composantes visuelles d'un buisson ligneux d'ornement : le rosier". Thesis, Rennes, Agrocampus Ouest, 2016. http://www.theses.fr/2016NSARB287/document.
Texto completoShrubs form a key plant model to meet social and environmentalconcerns. Usually transposed on the tree model, their architecturaldevelopment is still ill-known and understudied to address visualquality. To identify and anticipate such expectations, the visual qualitymanagement of ornamental plants through a multidisciplinarymethodology is proposed. It includes architecture of the plants withits phenotypic plasticity and their visual appearance perception.On a rose bush: Rosa hybrida L. ‘Radrazz’, this work shows howarchitectural analysis with its modeling tools, sensory evaluationand image analysis can form a coherent scientifi c framework toface up to such a purpose, and be transposed for other taxa. Onvirtual rose bushes, and real ones exposed to a light gradient, thevisual appearance can be characterized objectively by means ofsensory tests using rotating plant video at different stages.Thevideo stand enables a better mental representation of the plant 3Dby the subjects, leading to a more complete and reliable descriptionof the plant visual appearance; then to predict this descriptionthrough statistically integrated image analysis of multiple plantfacets. Some relevant architectural variables, with numerousequivalents, potentially interesting to study the architecturaldevelopment of bushes during their life cycle, enabled to predicteven explain how visual components were built for a cultivar. Fora better market responsiveness, this work lays the foundationfor drafting interactive decision and innovation support tools forb
Toque, Carole. "Pour l'identification de modèles factoriels de séries temporelles : application aux ARMA stationnaires /". Paris : École nationale supérieure des télécommunications, 2006. http://catalogue.bnf.fr/ark:/12148/cb40949643z.
Texto completoSicard, Emeline. "Choix de composantes optimales pour l'analyse spatiale et la modélisation : application aux pluies mensuelles du Nordeste brésilien". Montpellier 1, 2004. http://www.theses.fr/2004MON13509.
Texto completoImam, Wae͏̈l. "Une extension spline additive de l'analyse en composantes principales sur variables instrumentables". Montpellier 2, 1998. http://www.theses.fr/1998MON20009.
Texto completoHomayouni, Saéid. "Caractérisation de scènes urbaines par analyse des images hyperspectrales". Paris, ENST, 2005. http://www.theses.fr/2005ENST0055.
Texto completoUrban area characterization is a delicate task, since these kinds of environments are complex features from various aspects. The geographic aspect of an urban environment may be the most important aspect which could be studied by Remote Sensing techniques. In particular, Hyperspectral Remote Sensing provides valuable information which a priori could effectively help us in information extraction tasks for urban area modelling. In fact, the need for precise, updated and detailed information is necessary within a lot of applications. HIS has been employed for various applications of detection and mapping of materials in natural and urban environment. For HSI analysis, two strategies may be considered; firstly, a supervised strategy and secondly, an unsupervised strategy. For urban materials mapping, we have applied the Spectral Matching techniques, as the supervised methods. In order to improve the results of these techniques, we proposed a fusion technique in decision level. Then, an unsupervised technique based on Independent Components Analysis (ICA), as a solution of mixing problem, is proposed. ICA has been used for spectral separation and classification. The Fuzzy C-Means clustering technique then has been applied in order to obtain a fuzzy classification map. These techniques are applied on hyperspectral images data acquired by CASI sensor over the city of Toulouse in France. These image sets contain the 32 and 48 band images with 2 and 4 meter of spatial resolution, respectively. We have compared the results with the ground truths data by evaluation of the classification accuracy using the matrix of confusion
De, Blois Sylvie. "Le paysage comme espace conceptuel et fonctionnel en écologie, analyse des composantes végétales de paysages agroforestiers". Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2001. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp05/NQ61382.pdf.
Texto completoCabral, Emmanuel Nicolas. "Etude spectrale des processus stationnaires multidimensionnels et analyse en composantes principales dans le domaine des fréquence". Toulouse 3, 2010. http://thesesups.ups-tlse.fr/802/.
Texto completoMultidimensional statistical methods knew in these last decades very important developments both for theoretical (particularly in operatorial statistics) and application fields, allowing functional data treatments. The frequency domain's field (and its important tool the Fourier transform) allows particularly to permit the treatment of random functions and particular stochastic processes (stationary for example). In the study of random phenomenons which can be modeled by a p-dimensional process (Xt)t epsilon T, where each X belongs to Cp, it should be interesting, in one hand, to control the spectral tools associated with (Xt)t epsilon T and, in other hand, to obtain, in view of clarification, a q-dimensional process (with q < p) which best summarizes as possible (Xt)t epsilon T in a certain sense. There exists methods which allow to summarize a p-dimensional stationary continuous random function, defined on T = Z, R, [-pi,pi[ or, more generally, on a locally compact Abélian group. They lean on a criterion strictly closed to stationarity and their implementation is the Principal Component Analysis (PCA) of random measure associated with (Xt)t epsilon T. This PCA, also called, PCA in the frequency domain of (Xt)t epsilon T, consists on the PCA of each spectral component of (Xt)t epsilon T. We can present our contributions in the following four points: At first, it is well known that any stationary process can be associated with various spectral tools such as the random measure, the spectral measure with projector values and a unitary operator. We carry out a synthetic work about tensor and convolution products of random and spectral measures. This investment allows to answer, in practice, to interpolation problems, to identification of a spatial process, or to inverse Fourier transform problem. Secondly, given a periodic stationary process, we are interested in the specificity of its spectral tools. We define a notion of quasi-periodicity and quantify the closeness between the associated random measures of a quasi-periodic process and its periodic version. We study then the proximity between the corresponding PCA's in the frequency domain. .