L’objectif principal de cette thèse est de décliner et d’analyser l’algorithme kernel-LMS à noyau Gaussien dans trois cadres différents: celui des noyaux uniques et multiples, à valeurs réelles et à valeurs complexes, dans un contexte d’apprentissage distributé et coopératif dans les réseaux de capteurs. Plus précisement, ce travail s’intéresse à l’analyse du comportement en moyenne et en erreur quadratique de cas différents types d’algorithmes LMS à noyau. Les modèles analytiques de convergence obtenus sont validés par des simulations numérique. Tout d’abord, nous introduisons l’algorithme LMS, les espaces de Hilbert à noyau reproduisants, ainsi que les algorithmes de filtrage adaptatif à noyau existants. Puis, nous étudions analytiquement le comportement de l’algorithme LMS à noyau Gaussien dans le cas où les statistiques des éléments du dictionnaire ne répondent que partiellement aux statistiques des données d’entrée. Nous introduisons ensuite un algorithme LMS modifié à noyau basé sur une approche proximale. La stabilité de l’algorithme est également discutée. Ensuite, nous introduisons deux types d’algorithmes LMS à noyaux multiples. Nous nous concentrons en particulier sur l’analyse de convergence de l’un d’eux. Plus généralement, les caractéristiques des deux algorithmes LMS à noyaux multiples sont analysées théoriquement et confirmées par les simulations. L’algorithme LMS à noyau complexe augmenté est présenté et ses performances analysées. Enfin, nous proposons des stratégies de diffusion fonctionnelles dans les espaces de Hilbert à noyau reproduisant. La stabilité́ de cas de l’algorithme est étudiée. / The main objective of this thesis is to derive and analyze the Gaussian kernel least-mean-square (LMS) algorithm within three frameworks involving single and multiple kernels, real-valued and complex-valued, non-cooperative and cooperative distributed learning over networks. This work focuses on the stochastic behavior analysis of these kernel LMS algorithms in the mean and mean-square error sense. All the analyses are validated by numerical simulations. First, we review the basic LMS algorithm, reproducing kernel Hilbert space (RKHS), framework and state-of-the-art kernel adaptive filtering algorithms. Then, we study the convergence behavior of the Gaussian kernel LMS in the case where the statistics of the elements of the so-called dictionary only partially match the statistics of the input data. We introduced a modified kernel LMS algorithm based on forward-backward splitting to deal with $\ell_1$-norm regularization. The stability of the proposed algorithm is then discussed. After a review of two families of multikernel LMS algorithms, we focus on the convergence behavior of the multiple-input multikernel LMS algorithm. More generally, the characteristics of multikernel LMS algorithms are analyzed theoretically and confirmed by simulation results. Next, the augmented complex kernel LMS algorithm is introduced based on the framework of complex multikernel adaptive filtering. Then, we analyze the convergence behavior of algorithm in the mean-square error sense. Finally, in order to cope with the distributed estimation problems over networks, we derive functional diffusion strategies in RKHS. The stability of the algorithm in the mean sense is analyzed.

The purpose of this thesis is to implement Total Productive Maintenance (TPM) in Nigeria Plastics Manufacturing Company. Manufacturing companies round the world pay huge amount of money for purchasing new equipments to boost their production however nothing or little is done to achieve or obtain full output from the machine for which it is intended to do. Small losses in time or deviations from planned or calculated capability are taken as usual machine performance. But currently as a result of improved capability levels and demand of quality product at lower prices, purchasing latest machine/equipment is not a way out unless it is completely used. Total Productive Maintenance (TPM) is a method that involve everybody totally, from high management to all workers to implement a complete maintenance program for all machine/equipment during its life. This method ends up in most effectiveness of tools, equipment, virtuously improved workers, tidy up working area, neat and clean working environment. A structure is going to be developed with the potential of evaluating the impact of implementing total productive maintenance within. By evaluating the result or outcome of Total Productive Maintenance (TPM), manufacturing companies can create sensible/smart decisions to improve the potency and standard of the machine, equipment and also the product on XYZ Plastics Manufacturing Company in Nigerian.

This thesis proposes and implements a system for movement analysis of vehicles on crossroads. It detects and tracks the movement of vehicles in the video, gained from the stationary video camera, which has the view of some crossroad. The trajectories are stored and their number and directions are analysed. The detection was made using cascade classifier. A dataset of 10500 positive and 10500 negative samples has been created to train the classifier. Vehicles are tracked using KCF method. For trajectory clustering, needed by analysis, the Mean Shift method is used. Testing showed, that the overall success of vehicle movement analysis is 92.77%.

Cancer cells display characteristic traits acquired in a step-wise manner during carcinogenesis. Some of these traits are autonomous growth, induction of angiogenesis, invasion and metastasis. In this thesis, the focus is on one of the latest stages of tumor progression, tumor invasion. Tumor invasion emerges from the combined effect of tumor cell-cell and cell-microenvironment interactions, which can be studied with the help of mathematical analysis. Cellular automata (CA) can be viewed as simple models of self-organizing complex systems in which collective behavior can emerge out of an ensemble of many interacting "simple" components. In particular, we focus on an important class of CA, the so-called lattice-gas cellular automata (LGCA). In contrast to traditional CA, LGCA provide a straightforward and intuitive implementation of particle transport and interactions. Additionally, the structure of LGCA facilitates the mathematical analysis of their behavior. Here, the principal tools of mathematical analysis of LGCA are the mean-field approximation and the corresponding Lattice Boltzmann equation. The main objective of this thesis is to investigate important aspects of tumor invasion, under the microscope of mathematical modeling and analysis: Impact of the tumor environment: We introduce a LGCA as a microscopic model of tumor cell migration together with a mathematical description of different tumor environments. We study the impact of the various tumor environments (such as extracellular matrix) on tumor cell migration by estimating the tumor cell dispersion speed for a given environment. Effect of tumor cell proliferation and migration: We study the effect of tumor cell proliferation and migration on the tumor’s invasive behavior by developing a simplified LGCA model of tumor growth. In particular, we derive the corresponding macroscopic dynamics and we calculate the tumor’s invasion speed in terms of tumor cell proliferation and migration rates. Moreover, we calculate the width of the invasive zone, where the majority of mitotic activity is concentrated, and it is found to be proportional to the invasion speed. Mechanisms of tumor invasion emergence: We investigate the mechanisms for the emergence of tumor invasion in the course of cancer progression. We conclude that the response of a microscopic intracellular mechanism (migration/proliferation dichotomy) to oxygen shortage, i.e. hypoxia, maybe responsible for the transition from a benign (proliferative) to a malignant (invasive) tumor. Computing in vivo tumor invasion: Finally, we propose an evolutionary algorithm that estimates the parameters of a tumor growth LGCA model based on time-series of patient medical data (in particular Magnetic Resonance and Diffusion Tensor Imaging data). These parameters may allow to reproduce clinically relevant tumor growth scenarios for a specific patient, providing a prediction of the tumor growth at a later time stage. / Krebszellen zeigen charakteristische Merkmale, die sie in einem schrittweisen Vorgang während der Karzinogenese erworben haben. Einige dieser Merkmale sind autonomes Wachstum, die Induktion von Angiogenese, Invasion und Metastasis. Der Schwerpunkt dieser Arbeit liegt auf der Tumorinvasion, einer der letzten Phasen der Tumorprogression. Die Tumorinvasion ensteht aus der kombinierten Wirkung von den Wechselwirkungen Tumorzelle-Zelle und Zelle-Mikroumgebung, die mit die Hilfe von mathematischer Analyse untersucht werden können. Zelluläre Automaten (CA) können als einfache Modelle von selbst-organisierenden komplexen Systemen betrachtet werden, in denen kollektives Verhalten aus einer Kombination von vielen interagierenden "einfachen" Komponenten entstehen kann. Insbesondere konzentrieren wir uns auf eine wichtige CA-Klasse, die sogenannten Zelluläre Gitter-Gas Automaten (LGCA). Im Gegensatz zu traditionellen CA bieten LGCA eine einfache und intuitive Umsetzung der Teilchen und Wechselwirkungen. Zusätzlich erleichtert die Struktur der LGCA die mathematische Analyse ihres Verhaltens. Die wichtigsten Werkzeuge der mathematischen Analyse der LGCA sind hier die Mean-field Approximation und die entsprechende Lattice - Boltzmann - Gleichung. Das wichtigste Ziel dieser Arbeit ist es, wichtige Aspekte der Tumorinvasion unter dem Mikroskop der mathematischen Modellierung und Analyse zu erforschen: Auswirkungen der Tumorumgebung: Wir stellen einen LGCA als mikroskopisches Modell der Tumorzellen-Migration in Verbindung mit einer mathematischen Beschreibung der verschiedenen Tumorumgebungen vor. Wir untersuchen die Auswirkungen der verschiedenen Tumorumgebungen (z. B. extrazellulären Matrix) auf die Migration von Tumorzellen dürch Schätzung der Tumorzellen-Dispersionsgeschwindigkeit in einem gegebenen Umfeld. Wirkung von Tumor-Zellenproliferation und Migration: Wir untersuchen die Wirkung von Tumorzellenproliferation und Migration auf das invasive Verhalten der Tumorzellen durch die Entwicklung eines vereinfachten LGCA Tumorwachstumsmodells. Wir leiten die entsprechende makroskopische Dynamik und berechnen die Tumorinvasionsgeschwindigkeit im Hinblick auf die Tumorzellenproliferation- und Migrationswerte. Darüber hinaus berechnen wir die Breite der invasiven Zone, wo die Mehrheit der mitotischer Aktivität konzentriert ist, und es wird festgestellt, dass diese proportional zu den Invasionsgeschwindigkeit ist. Mechanismen der Tumorinvasion Entstehung: Wir untersuchen Mechanismen, die für die Entstehung von Tumorinvasion im Verlauf des Krebs zuständig sind. Wir kommen zu dem Schluss, dass die Reaktion eines mikroskopischen intrazellulären Mechanismus (Migration/Proliferation Dichotomie) zu Sauerstoffmangel, d.h. Hypoxie, möglicheweise für den Übergang von einem gutartigen (proliferative) zu einer bösartigen (invasive) Tumor verantwortlich ist. Berechnung der in-vivo Tumorinvasion: Schließlich schlagen wir einen evolutionären Algorithmus vor, der die Parameter eines LGCA Modells von Tumorwachstum auf der Grundlage von medizinischen Daten des Patienten für mehrere Zeitpunkte (insbesondere die Magnet-Resonanz-und Diffusion Tensor Imaging Daten) ermöglicht. Diese Parameter erlauben Szenarien für einen klinisch relevanten Tumorwachstum für einen bestimmten Patienten zu reproduzieren, die eine Vorhersage des Tumorwachstums zu einem späteren Zeitpunkt möglich machen.

Avec l'hétérogénéité des réseaux et la demande croissante de services multimédias, la Qualité d'Expérience (QoE) devient un facteur déterminant de réussite ou d’échec de ces services. Dans cette thèse, nous avons d’abord analysé l'impact des paramètres de qualité sur l'engagement de l'utilisateur, afin de comprendre l’effet des paramètres vidéo (temps de démarrage, qualité intrinsèque, taux de blocage) et la popularité du service sur l'engagement de l'utilisateur. Nos résultats ont montré que le taux de blocage et la popularité du service sont des paramètres critiques qui impactent fortement la satisfaction et l'engagement de l’utilisateur, tandis que, le temps de démarrage a moins d’impact. Ensuite, les approches subjectives de QoE comme le Mean Opinion Score (MOS) ont été traité dans cette thèse, où les utilisateurs sont appelés à donner une note de satisfaction en fonction des informations de contextes. Une analyse statistique des résultats obtenus à partir de tests utilisateurs a montré que le type de terminal et de contenu sont des paramètres qui impactent fortement le MOS. En plus, des modèles mathématiques d’estimation de QoE en fonction des informations de contextes ont été proposés. Enfin, deux techniques d’optimisations ont été proposées : Le MDASH (MOS Dynamic Adaptive Streaming over HTTP) pour l’optimisation vidéo de flux partageant les mêmes ressources, de manière à garantir une certaine QoE pour tous les utilisateurs. Et une optimisation de livraison, basée sur des fonctions d’utilités dans laquelle, une fonction globale d'utilité est calculée en fonction de différentes contraintes (qui dépendent fortement des stratégies des acteurs de la chaîne de livraison) / With the network heterogeneity and increasing demand of multimedia services, Quality of Experience (QoE) becomes a crucial determinant of the success or failure of these services. In this thesis, we first propose to analyze the impact of quality metrics on user engagement, in order to understand the effects of video metrics (video startup time, average bitrate, buffering ratio) and content popularity on user engagement. Our results show that video buffering and content popularity are critical parameters which strongly impacts the end-user’s satisfaction and user engagement, while the video startup time appears as less significant. On other hand, we consider subjective approaches such as the Mean Opinion Score (MOS) for evaluating QoE, in which users are required to give their assessment according to contextual information. A detailed statistical analysis of our study shows the existence of non-trivial parameters impacting MOS (the type of device and the content type). We propose mathematical models to develop functional relationships between the QoE and the context information which in turn permits us to estimate the QoE. A video content optimization technique called MDASH (for MOS Dynamic Adaptive Streaming over HTTP) is proposed, which improves the perceived QoE for different video sessions sharing the same local network, while taking QoE fairness among users as a leitmotiv. We also propose a utility-based approach for video delivery optimization, in which a global utility function is computed based on different constraints (e.g. target strategies coming from the actors of the delivery chain)

La symétrie est une propriété géométrique importante en perception visuelle qui traduit notre perception des correspondances entre les différents objets ou formes présents dans une scène. Elle est utilisée comme élément caractéristique dans de nombreuses applications de la vision par ordinateur (comme par exemple la détection, la segmentation ou la reconnaissance d'objets) mais également comme une caractéristique formelle en sciences de l'art (ou en analyse esthétique). D’importants progrès ont été réalisés ces dernières décennies pour la détection de la symétrie dans les images mais il reste encore de nombreux verrous à lever. Dans cette thèse, nous nous intéressons à la détection des symétries de réflexion, dans des images réelles, à l'échelle globale. Nos principales contributions concernent les étapes d'extraction de caractéristiques et de représentation globale des axes de symétrie. Nous proposons d'abord une nouvelle méthode d'extraction de segments de contours à l'aide de bancs de filtres de Gabor logarithmiques et une mesure de symétrie intersegments basée sur des caractéristiques locales de forme, de texture et de couleur. Cette méthode a remporté la première place à la dernière compétition internationale de symétrie pour la détection mono- et multi-axes. Notre deuxième contribution concerne une nouvelle méthode de représentation des axes de symétrie dans un espace linéaire-directionnel. Les propriétés de symétrie sont représentées sous la forme d'une densité de probabilité qui peut être estimée, de manière non-paramétrique, par une méthode à noyauxbasée sur la distribution de Von Mises-Fisher. Nous montrons que la détection des axes dominants peut ensuite être réalisée à partir d'un algorithme de type "mean-shift” associé à une distance adaptée. Nous introduisons également une nouvelle base d'images pour la détection de symétrie mono-axe dans des photographies professionnelles issue de la base à grande échelle AVA (Aestetic Visual Analysis). Nos différentes contributions obtiennent des résultats meilleurs que les algorithmes de l'état de l'art, évalués sur toutes les bases disponibles publiquement, spécialement dans le cas multi-axes. Nous concluons que les propriétés de symétrie peuvent être utilisées comme des caractéristiques visuelles de niveau sémantique intermédiaire pour l'analyse et la compréhension de photographies. / Symmetry is a fundamental principle of the visual perception to feel the equally distributed weights within foreground objects inside an image. It is used as a significant visual feature through various computer vision applications (i.e. object detection and segmentation), plus as an important composition measure in art domain (i.e. aesthetic analysis). The development of symmetry detection has been improved rapidly since last century. In this thesis, we mainly aim to propose new approaches to detect reflection symmetry inside real-world images in a global scale. In particular, our main contributions concern feature extraction and globalrepresentation of symmetry axes. First, we propose a novel approach that detects global salient edges inside an image using Log-Gabor filter banks, and defines symmetry oriented similarity through textural and color around these edges. This method wins a recent symmetry competition worldwide in single and multiple cases.Second, we introduce a weighted kernel density estimator to represent linear and directional symmetrical candidates in a continuous way, then propose a joint Gaussian-vonMises distance inside the mean-shift algorithm, to select the relevant symmetry axis candidates along side with their symmetrical densities. In addition, we introduce a new challenging dataset of single symmetry axes inside artistic photographies extracted from the large-scale Aesthetic Visual Analysis (AVA) dataset. The proposed contributions obtain superior results against state-of-art algorithms among all public datasets, especially multiple cases in a global scale. We conclude that the spatial and context information of each candidate axis inside an image can be used as a local or global symmetry measure for further image analysis and scene understanding purposes.

Non-exhaust PM10 is an issue in the urban environment linked to health issues. Emissions of non-exhaust PM10 is relatable to pavement properties. Also of importance is resuspension of road dust stored from surfaces. This depends on the traffic and metrological conditions. Given this, the purpose of the thesis was to give an overview limited to Sweden and the Nordic countries regarding non-exhaust PM10 emissions and road dust. The overview includes how particles are related to human health. Also included is the principle of how particles are emitted from road surface and tyre interaction, both directly and through resuspension of road dust. This thesis also includes an overview of how the use of studded tyres impact on asphalt surfacings and how the properties of the materials used impact on the abrasion wear. This is then linked to the emissions of non-exhaust particles. Further described is how measurements can be done of ambient particles and road dust, followed on two major models for road abrasion wear and non-exhaust PM prediction. Also included is how road operation, e.g. traction sanding and dust binding, influence the particle emissions together with other options to reduce the emissions through, e.g. limiting the use of studded tyres. One special issue discussed in this thesis is the lack of holistic view regarding the environmental problems in the urban environment with focus on particle emissions and road noise emissions, both from the road surface and tyre interaction. Currently the most problematic issue is prioritized and the resulting solution to that specific problem might increase other problems. This thesis shows that much knowledge is available regarding non-exhaust PM10 emissions and road dust, but also that several knowledge gaps exists. Several suggestions on further studies is given together with a brief overview on the continued work forward from this thesis. / Icke-avgasemissioner av PM10 är ett problem i urbana miljöer länkat till flera hälsoaspekter. Dessa emissioner kan relateras till beläggningars egenskaper. Även resuspension av vägdamm från ytor är av betydelse och beror på både trafiken och meterologin. Baserat på detta är syftet med denna avhandling att ge en översikt kring icke-avgas PM10 emissioner och vägdamm, begränsat till Sverige och de övriga nordiska länderna. Denna översikt inkluderar hur partiklar relaterar till människans hälsa. Annat som inkluderas är hur partiklarna emitteras från vägyta-däckinteraktionen, både direkt och genom resuspension av vägdamm. Avhandlingen inkluderar även en översikt kring hur användandet av dubbdäck inverkar på vägbeläggningar och hur dess egenskaper inverkar på nötningsslitage. Detta länkas därefter till partikelemissioner. Vidare beskrivs även hur mätningar kan genomföras av partiklar samt vägdamm vilket följs upp av beskrivningar kring två större modeller kring prediktion av nötningsslitage och prediktion av icke-avgasemissioner. Även hur driftåtgärder inverkar på emissionerna tillsammans med alternativ för att minska emissionerna tas upp. Ett särskilt problem som tas upp i avhandlingen är bristen på helhetssyn beträffande miljöproblem i den urbana miljön med fokus på partikel- och bulleremissioner från vägyta-däckinteraktionen. För närvarande brukar det värsta problemet prioriteras och lösningen till det detta problem kan i sin tur medföra att andra miljöproblem istället förvärras. Denna avhandling visar på att mycket kunskap existerar kring icke-avgasemissioner av PM10 och kring vägdamm, men även att flertalet kunskapsluckor existerar. Flertalet förslag på vidare studier ges tillsammans med en överblick kring det fortsatta arbetet. / <p>QC 20180202</p>

In a single capture-recapture problem, two new Wilson methods for interval estimation of population size are derived. Classical Chapman interval, Wilson and Wilson-cc intervals are examined and compared in terms of their expected interval width and exact coverage properties in two models. The new approach performs better than the Chapman in each model. Bayesian analysis also gives a different way to estimate population size.

Mean variance optimization has shortcomings making the strategy far from optimal from an investor’s perspective. The purpose of the study is to conduct an empirical investigation as to how modern methods of portfolio optimization address the shortcomings associated with mean variance optimization. Equal risk contribution, the Most diversified portfolioand a modification of the Minimum variance portfolio are considered as alternatives to the mean variance model. Portfolio optimization models introduced are explained in detail and solved using the optimization algorithms Cyclical coordinate descent and Alternating direction method of multipliers. Through implementation and backtesting using a diverse set of indices representing various asset classes, the study shows that the mean variance model suffers from high turnover and sensitivity to input parameters in comparison to the modern alternatives. The sophisticated asset allocation models equal risk contribution and the most diversified portfolio do not rely on expected return as an input parameter, which is seen as an advantage, and are not affected to the same extent by the shortcomings associated with mean variance optimization. The paper concludes by discussing the findings critically and suggesting ideas for further research. / Maximering av avkastning i samband med minimering av varians, på engelska kallat Mean variance optimization, är inte optimalt ur en investerares synpunkt. Syftet med denna uppsats är att genomföra en empirisk studie av hur moderna metoder för portföljallokering adresserar de problem som är förknippade med Mean variance optimization. Mer specifikt undersöks allokeringsstrategierna Equal risk contribution, Most diversified portfolio samt en variant av Minimum variance som ersättare till Mean variance optimization. Allokeringsmetoderna beskrivs detaljerat och löses med optimeringsalgoritmerna Cyclical coordinate descent och Alternating direction method of multipliers. Genom implementering och historisk simulering med ett antal index som representerar olika tillgångsslag visar studien att Mean variance optimization innebär hög portföljomsättning och har en större känslighet för ingångsparametrar i jämförelse med de moderna alternativen. De sofistikerade allokeringsmodellerna Equal risk contribution och Most diversified portfolio bygger inte på ingångsparametern förväntad avkastning, vilket ses som en fördel, och drabbas inte i samma utsträckning av problemen associerade med Mean variance optimization. Studien avslutas med att diskutera resultatet kritiskt och ge förslag på vidare studier som bygger på den teori och det resultat som har presenterats.

This thesis evaluates the mean-variance analysis framework by comparing the performance of an optimized portfolio consisting of stocks from the Dow Jones Industrial Average to the performance of the Dow Jones Industrial Average index itself. The results show that the optimized portfolio performs better than the corresponding index when evaluated on the period between 2015 and 2019. However, the variance of the returns are high and therefore it is difficult to determine if mean-variance analysis performs better than its corresponding index in the general case. Furthermore, it is shown that individual stocks can still influence the movement of an optimized portfolio significantly, even though the model is supposed to diversify firm-specific risk. Thus, the authors recommend modifying the model by restricting the amount that is allowed to be invested in a single stock, if one wishes to apply mean-variance analysis in reality. To be able to draw further conclusions, more practical research within the subject needs to be done. / Denna uppsats utvärderar ramverket ”mean-variance analysis” genom att jämföra prestandan av en optimerad portfölj bestående av aktier från Dow Jones Industrial Average med prestandan av indexet Dow Jones Industrial Average självt. Resultaten visar att att den optimerade portföljen presterar bättre än motsvarande index när de utvärderas på perioden 2015 till 2019. Dock är variansen av avkastningen hög och det är därför svårt att bedöma om mean-variance analysis generellt sett presterar bättre än sitt motsvarande index. Vidare visas det att individuella aktier fortfarande kan påverka den optimerade portföljens rörelser, fastän modellen antas diversifiera företagsspecifik risk. På grund av detta rekommenderar författarna att modifiera modellen genom att begränsa mängden som kan investeras i en individuell aktie, om man önskar att tillämpa mean-variance analysis i verkligheten. För att kunna dra vidare slutsatser så krävs mer praktisk forskning inom området.