La présente thèse s'intéresse à l'utilisation de méta-données textuelles pour l'analyse d'image. Nous cherchons à utiliser ces informations additionelles comme supervision faible pour l'apprentissage de modèles de reconnaissance visuelle. Nous avons observé un récent et grandissant intérêt pour les méthodes capables d'exploiter ce type de données car celles-ci peuvent potentiellement supprimer le besoin d'annotations manuelles, qui sont coûteuses en temps et en ressources. Nous concentrons nos efforts sur deux types de données visuelles associées à des informations textuelles. Tout d'abord, nous utilisons des images de dépêches qui sont accompagnées de légendes descriptives pour s'attaquer à plusieurs problèmes liés à la reconnaissance de visages. Parmi ces problèmes, la vérification de visages est la tâche consistant à décider si deux images représentent la même personne, et le nommage de visages cherche à associer les visages d'une base de données à leur noms corrects. Ensuite, nous explorons des modèles pour prédire automatiquement les labels pertinents pour des images, un problème connu sous le nom d'annotation automatique d'image. Ces modèles peuvent aussi être utilisés pour effectuer des recherches d'images à partir de mots-clés. Nous étudions enfin un scénario d'apprentissage multimodal semi-supervisé pour la catégorisation d'image. Dans ce cadre de travail, les labels sont supposés présents pour les données d'apprentissage, qu'elles soient manuellement annotées ou non, et absentes des données de test. Nos travaux se basent sur l'observation que la plupart de ces problèmes peuvent être résolus si des mesures de similarité parfaitement adaptées sont utilisées. Nous proposons donc de nouvelles approches qui combinent apprentissage de distance, modèles par plus proches voisins et méthodes par graphes pour apprendre, à partir de données visuelles et textuelles, des similarités visuelles spécifiques à chaque problème. Dans le cas des visages, nos similarités se concentrent sur l'identité des individus tandis que, pour les images, elles concernent des concepts sémantiques plus généraux. Expérimentalement, nos approches obtiennent des performances à l'état de l'art sur plusieurs bases de données complexes. Pour les deux types de données considérés, nous montrons clairement que l'apprentissage bénéficie de l'information textuelle supplémentaire résultant en l'amélioration de la performance des systèmes de reconnaissance visuelle.

Este trabalho tem dois objetivos principais. Primeiramente, analisamos a preservação de conexidade na extensão de espaços compactos por funções contínuas, técnica utilizada por Koszmider para obter $C(K)$ indecomponível com poucos operadores. Mostramos que para todo compacto metrizável $K$ existe um desconexo $L$ que é obtido a partir de $K$ por uma quantidade finita de extensões por funções contínuas. Em seguida, enfatizamos a construção de espaços de Banach da forma $C(K)$ com poucos operadores, com a propriedade de que $C(L)$ tem poucos operadores, para todo fechado $L \\subseteq K$. Assumindo o princípio diamante construímos uma família $(K_\\xi)_{\\xi < 2^{(2^\\omega)}}$ de espaços conexos e hereditariamente Koszmider tais que todo operador de $C(K_\\xi)$ em $C(K_\\eta)$ é fracamente compacto, para $\\xi$ diferente de $\\eta$. Em particular, $(C(K_\\xi))_{\\xi < 2^{(2^\\omega)}}$ é uma família de espaços de Banach indecomponíveis e dois a dois essencialmente incomparáveis, e cada espaço $K_\\xi$ responde positivamente ao problema de Efimov. Apresentamos também um método de construção via forcing de um espaço compacto e conexo $K$ hereditariamente fracamente Koszmider. / This work has two main objectives. First, we analyze the preservation of connectedness in the extension of compact spaces by continuous functions, a technique used by Koszmider to obtain an indecomposable Banach space $C(K)$ with few operators. We show that for any metrizable compactum $K$ there exists a disconnected $L$ which is obtained from $K$ by finitely many extensions by continuous functions. Next, we emphasize the construction of Banach spaces of the form $C(K)$ with the property that $C(L)$ has few operators, for every closed $L \\subseteq K$. Assuming the diamond principle we construct a family $(K_\\xi)_{\\xi < 2^{(2^\\omega)}}$ of connected and hereditarily Koszmider spaces such that every operator from $C(K_\\xi)$ into $C(K_\\eta)$ is weakly compact, for $\\xi$ different from $\\eta$. In particular, $(C(K_\\xi))_{\\xi < 2^{(2^\\omega)}}$ is a family of indecomposable and pairwise essentially incomparable Banach spaces, and each space $K_\\xi$ responds positively to the Efimov\'s problem. We also present a method of construction using forcing of a compact and connected hereditarily weakly Koszmider space $K$.

Neurone sind anregbare Zellen, die mit Hilfe von elektrischen Signalen miteinander kommunizieren. Im allgemeinen werden eingehende Signale von den Nervenzellen in einer nichtlinearen Art und Weise verarbeitet. Wie diese Verarbeitung in einer umfassenden und exakten Art und Weise mathematisch beschrieben werden kann, ist bis heute nicht geklärt und ist Gegenstand aktueller Forschung. In dieser Arbeit untersuchen wir die nichtlineare Übertragung und Verarbeitung von Signalen durch stochastische Nervenzellen und wenden dabei zwei unterschiedliche Herangehensweisen an. Im ersten Teil der Arbeit befassen wir uns mit der Frage, auf welche Art und Weise ein Signal mit einer bekannten Zeitabhängigkeit die Rate der neuronalen Aktivität beeinflusst. Im zweiten Teil der Arbeit widmen wir uns der Rekonstruktion eingehender Signale aus der durch sie hervorgerufenen neuronalen Aktivität und beschäftigen uns mit der Abschätzung der übertragenen Informationsmenge. Die Ergebnisse dieser Arbeit demonstrieren, wie die etablierten linearen Theorien, die die Modellierung der neuronalen Aktivitätsrate bzw. die Rekonstruktion von Signalen beschreiben, um Beiträge höherer Ordnung erweitert werden können. Einen wichtigen Beitrag dieser Arbeit stellt allerdings auch die Darstellung der Signifikanz der nichtlinearen Theorien dar. Die nichtlinearen Beiträge erweisen sich nicht nur als schwache Korrekturen zu den etablierten linearen Theorien, sondern beschreiben neuartige Effekte, die durch die linearen Theorien nicht erfasst werden können. Zu diesen Effekten gehört zum Beispiel die Anregung von harmonischen Oszillationen der neuronalen Aktivitätsrate und die Kodierung von Signalen in der signalabhängigen Varianz einer Antwortvariablen. / Neurons are excitable cells which communicate with each other via electrical signals. In general, these signals are processed by the Neurons in a nonlinear fashion, the exact mathematical description of which is still an open problem in neuroscience. In this thesis, the broad topic of nonlinear signal processing is approached from two directions. The first part of the thesis is devoted to the question how input signals modulate the neural response. The second part of the thesis is concerned with the nonlinear reconstruction of input signals from the neural output and with the estimation of the amount of the transmitted information. The results of this thesis demonstrate how existing linear theories can be extended to capture nonlinear contributions of the signal to the neural response or to incorporate nonlinear correlations into the estimation of the transmitted information. More importantly, however, our analysis demonstrates that these extensions do not merely provide small corrections to the existing linear theories but can account for qualitatively novel effects which are completely missed by the linear theories. These effects include, for example, the excitation of harmonic oscillations in the neural firing rate or the estimation of information for systems with a signal-dependent output variance.

La méthode d'intégration produit a été proposée pour résoudre des équations linéaires de Fredholm de deuxième espèce singulières dont la solution exacte est régulière, au moins continue. Dans ce travail on adapte cette méthode à des équations dont la solution est juste intégrable. On étudie également son extension au cas non linéaire posé dans l'espace des fonctions intégrables. Ensuite, on propose une autre manière de mettre en oeuvre la méthode d'intégration produit : on commence par linéariser l'équation par une méthode de type Newton puis on discrétise les itérations de Newton par la méthode d'intégration produit / The product integration method has been proposed for solving singular linear Fredholm equations of the second kind whose exact solution is smooth, at least continuous. In this work, we adapt this method to the case where the solution is only integrable. We also study the nonlinear case in the space of integrable functions. Then, we propose a new version of the method in the nonlinear framework : we first linearize the eqaution by a Newton type method and then discretize the Newton iterations by the product integration method

In this thesis, the main objects of study are probability measures on the isomorphism classes of countable, connected rooted graphs. An important class of such measures is formed by unimodular measures, which satisfy a certain equation, sometimes referred to as the intrinsic mass transport principle. The so-called law of a finite graph is an example of a unimodular measure. We say that a measure is sustained by a countable graph if the set of rooted connected components of the graph has full measure. We demonstrate several new results involving sustained unimodular measures, and provide thorough arguments for known ones. In particular, we give a criterion for unimodularity on connected graphs, deduce that connected graphs sustain at most one unimodular measure, and prove that unimodular measures sustained by disconnected graphs are convex combinations. Furthermore, we discuss weak limits of laws of finite graphs, and construct counterexamples to seemingly reasonable conjectures.

Este trabalho tem dois objetivos principais. Primeiramente, analisamos a preservação de conexidade na extensão de espaços compactos por funções contínuas, técnica utilizada por Koszmider para obter $C(K)$ indecomponível com poucos operadores. Mostramos que para todo compacto metrizável $K$ existe um desconexo $L$ que é obtido a partir de $K$ por uma quantidade finita de extensões por funções contínuas. Em seguida, enfatizamos a construção de espaços de Banach da forma $C(K)$ com poucos operadores, com a propriedade de que $C(L)$ tem poucos operadores, para todo fechado $L \\subseteq K$. Assumindo o princípio diamante construímos uma família $(K_\\xi)_{\\xi < 2^{(2^\\omega)}}$ de espaços conexos e hereditariamente Koszmider tais que todo operador de $C(K_\\xi)$ em $C(K_\\eta)$ é fracamente compacto, para $\\xi$ diferente de $\\eta$. Em particular, $(C(K_\\xi))_{\\xi < 2^{(2^\\omega)}}$ é uma família de espaços de Banach indecomponíveis e dois a dois essencialmente incomparáveis, e cada espaço $K_\\xi$ responde positivamente ao problema de Efimov. Apresentamos também um método de construção via forcing de um espaço compacto e conexo $K$ hereditariamente fracamente Koszmider. / This work has two main objectives. First, we analyze the preservation of connectedness in the extension of compact spaces by continuous functions, a technique used by Koszmider to obtain an indecomposable Banach space $C(K)$ with few operators. We show that for any metrizable compactum $K$ there exists a disconnected $L$ which is obtained from $K$ by finitely many extensions by continuous functions. Next, we emphasize the construction of Banach spaces of the form $C(K)$ with the property that $C(L)$ has few operators, for every closed $L \\subseteq K$. Assuming the diamond principle we construct a family $(K_\\xi)_{\\xi < 2^{(2^\\omega)}}$ of connected and hereditarily Koszmider spaces such that every operator from $C(K_\\xi)$ into $C(K_\\eta)$ is weakly compact, for $\\xi$ different from $\\eta$. In particular, $(C(K_\\xi))_{\\xi < 2^{(2^\\omega)}}$ is a family of indecomposable and pairwise essentially incomparable Banach spaces, and each space $K_\\xi$ responds positively to the Efimov\'s problem. We also present a method of construction using forcing of a compact and connected hereditarily weakly Koszmider space $K$.

Over the last decades, there has been considerable attention in the accurate mathematical modeling and numerical simulations of free surface wave propagation in near-shore environments. A physical correct description of the large scale phenomena, which take place in the shallow water region, must account for strong nonlinear and dispersive effects, along with the interaction with complex topographies. First, a study on the behavior in nonlinear regime of different Boussinesq-type models is proposed, showing the advantage of using fully-nonlinear models with respect to weakly-nonlinear and weakly dispersive models (commonly employed). Secondly, a new flexible strategy for solving the fully-nonlinear and weakly-dispersive Green-Naghdi equations is presented, which allows to enhance an existing shallow water code by simply adding an algebraic term to the momentum balance and is particularly adapted for the use of hybrid techniques for wave breaking. Moreover, the first discretization of the Green-Naghdi equations on unstructured meshes is proposed via hybrid finite volume/ finite element schemes. Finally, the models and the methods developed in the thesis are deployed to study the physical problem of bore formation in convergent alluvial estuary, providing the first characterization of natural estuaries in terms of bore inception. / Ces dernières décennies, une attention particulière a été portée sur la modélisation mathématique et la simulation numérique de la propagation de vagues en environnements côtiers. Une description physiquement correcte des phénomènes à grande échelle, qui apparaissent dans les régions d'eau peu profonde, doit prendre en compte de forts effets non-linéaires et dispersifs, ainsi que l'interaction avec des bathymétries complexes. Dans un premier temps, une étude du comportement en régime non linéaire de différents modèles de type Boussinesq est proposée, démontrant l'avantage d'utiliser des modèles fortement non-linéaires par rapport à des modèles faiblement non-linéaires et faiblement dispersifs (couramment utilisés). Ensuite, une nouvelle approche flexible pour résoudre les équations fortement non-linéaires et faiblement dispersives de Green-Naghdi est présentée. Cette stratégie permet d'améliorer un code "shallow water" existant par le simple ajout d'un terme algébrique dans l'équation du moment et est particulièrement adapté à l'utilisation de techniques hybrides pour le déferlement des vagues. De plus, la première discrétisation des équations de Green-Naghdi sur maillage non structuré est proposée via des schémas hybrides Volume Fini/Élément Fini. Finalement, les modèles et méthodes développés dans la thèse sont appliqués à l'étude du problème physique de la formation du mascaret dans des estuaires convergents et alluviaux. Cela a amené à la première caractérisation d'estuaire naturel en terme d'apparition de mascaret.

In this thesis, the main objects of study are probability measures on the isomorphism classes of countable, connected rooted graphs. An important class of such measures is formed by unimodular measures, which satisfy a certain equation, sometimes referred to as the intrinsic mass transport principle. The so-called law of a finite graph is an example of a unimodular measure. We say that a measure is sustained by a countable graph if the set of rooted connected components of the graph has full measure. We demonstrate several new results involving sustained unimodular measures, and provide thorough arguments for known ones. In particular, we give a criterion for unimodularity on connected graphs, deduce that connected graphs sustain at most one unimodular measure, and prove that unimodular measures sustained by disconnected graphs are convex combinations. Furthermore, we discuss weak limits of laws of finite graphs, and construct counterexamples to seemingly reasonable conjectures.

Detta arbete har handlat om att få förståelse vad barn eftersöker för designfaktorer i en mobilapplikation, som ska vara ekonomiskt lärande och hjälpa barn att planera vecko-och månadspengen. Studien skulle resultera i ett designkoncept baserat på barnens egna behövande och önskade faktorer för en ekonomiskt lärande mobilapplikation. Anledningen till att utföra en sådan studie uppkom då svenska banker uppmärksammat att barn har det idag svårare att förstå pengars värde och planera budgeten. Samtidigt som kontanthanteringen har minskat används heller inte fickpengar i samma grad, vilket har tidigare bevisats vara en bra praktik att lära barn med. Således saknas det ett naturligt sätt för dagens barn att handskas med budgetering och lära sig pengarsvärde. Studiens syfte var därför att synliggöra dessa transaktioner genom att främja vecko- och månadspengen och barnens finansiella lärdomar i ett designarbete. Därmed blev frågeställningen: “Vilka designfaktorer prioriterar barn i en ekonomsikt lärande mobilapplikation”, vilket var tänkt kunna besvaras med hjälp av en skolklass med barn i elva till tolv årsåldern. Projektets syfte var att ha en hög användarinvolvering för att kunna få djupare insikter för barnens behov och även låta dem vara en del av designarbetet för att åstadkomma ett resultat de gillar. Teori som spelifiering användes och även en del normativa teorier som bygger sin teori på att försöka förändra eller främja ett visst beteende hos användarna, vilket i denna studie har valts att kallas för teorier inom social design. För datainsamlingsmetoder genomfördes tre primära insamlingar med en femteklass, där barnen fick delas in i tre grupper och genomföra varsin metod. Metoderna ämnade antingen att framkalla empiri om barnens ekonomiska och sociala tillvaro, samt låta dem generera egna designförslag och koncept för vecko-och månadsplanering. För att sedan analysera datan anpassades Tromp och Heckerts teoretiska modell som uppvisar ett sexfaldig process: från att definiera ett socialt fenomen, hitta problemområden inom den, använda strategier för att motarbeta och slutligen att framkalla ett designförslag. I denna modell integrerades studiens teorier för att kunna nyttja de strategier vid dataanalysen. Slutsatsen blev att barnen vill lära sig med spel, kunna personifiera innehållet, kunna känna igen sin skolidentitet i tjänstens kontext, illustrera information med olika diagram och där de visats ha en fascination för högteknologiska prylar, som också tros ge dem social status. Barnen visade sig vilja vara ansvarsfulla och kunna planera sin egna budget, men vill kunna såväl belönas, antingen med godis eller pengar, samt även med roande och underhållande upplevelser. Därmed resulterade detta i en prototyp för en mobilapplikation som är format utefter barnens önskade funktioner. / This work is about gaining insight for what kinds of design elements children search in a mobile application, aimed for learning economics and help them plan their budget. This study would result with a design concept based on children’s own needed and wished elements for an economic learning mobile application. The reason to start a project like this is because Swedish banks have noticed that children today have it difficult to understand money’s worth and plan their budget. While the card-transactions have been reduced on the market have also the use of pocket money been minimized, even though it have been proved to be a good way to teach children money’s worth. Thus, there is not a natural way of today’s kids to deal with their budget and learn money’s worth. The study’s purpose was to visualize these transactions by promoting children’s weekly and monthly allowance and their financial lessons in a design work. This made the question: “What design elements prioritize children in an economic learning mobile application”, which was supposed to be answered by a school class with children at age eleven to twelve. The projects purpose was to have a high user involvement for gaining deeper insight for children’s needs and also let them be apart of the design work to achieve a result they would like.  Theory like gamification was used and a couple normative theories based their theory to promote a certain behavior of the users, which in this study have been choosed to call them all for theories of social design. There were three primary methods for data collection methods with a class from fifth-grade, were the children was divided in three groups to complete each method. The methods were aimed to either induce data for the children’s economic and social statement and also let them generate their own design concept for weekly and monthly allowance. Tromp and Heckerts theoretical model was used to analyze the data, were their model was costume to match this specific study. The model is a six fold process: from defining a social phenomenon, find issues with the area, then use strategies to counteract the problems and with developing design proposals. The studies theories was integrated in this model so that it would be possible to make use of the strategies during the analyses. The conclusion was that children want to learning with games, to personalize the content, to recognize their school identity in the service, to illustrate the information with different charts and that they have a big fascination with high-tech gadgets, that also give them social status. The children showed that they want to be responsible for their own actions and budget, but they want also to be reward when they do something good. The reward should then be money or candy and they wanted to have amusing experience during education. Therefore this project was result in a prototype for / <p>Hoppas ni finner rapporten intressant och kontanta oss vid eventuella frågor!</p>

In dieser Arbeit wurden die Morpho-Physiologie und das Verhalten zweier Arten Afrikanischer schwach elektrischer Fische, Marcusenius victoriae und Petrocephalus degeni, im Labor und in einem ihrer natürlichen Habitate im Lwamunda Sumpf in Uganda untersucht. Die zwei Hauptziele dieser Arbeit waren (i) tageszeitabhängige Verhaltensrhythmen (Aktivität, Habitatnutzung) im Labor und im Freiland zu untersuchen und (ii) die Ausprägung und Plastizität der morpho-physiologischen Merkmale von P. degeni zu untersuchen, die ihnen erlauben bei natürlich vorkommender, geringer Sauerstoffverfügbarkeit (Hypoxie) zu überleben. Tageszeitabhängige Verhaltensrhythmen beider Arten wurden im Labor über 42 Stunden und im natürlichen Habitat dieser Fische für sechs Tage erfasst. In den Laborversuchen verbrachten beide Arten tagsüber annähernd 100% der Zeit in einem bereitgestellten Versteck und schwammen nachts heraus um aktiv ihre Umwelt zu erkunden. Im Habitat wurden die meisten Fische in strukturell komplexen Habitaten unter schwimmenden Pflanzen detektiert. Nachts schwammen die Fische aktiv in die offenen und ungeschützten Bereiche der Lagune, vermutlich um nach Futter zu suchen und zu interagieren. Die Begleitende in-situ Messung der Sauerstoffverfügbarkeit zeigte, dass beide Arten präsent und vermutlich sogar am aktivsten waren während Phasen extremer nächtlicher Hypoxie. Zur Untersuchung der respiratorischen Merkmale von P. degeni wurden Respirometrieversuche mit hypoxie-akklimatisierten Tieren durchgeführt, Hämoglobin- und Laktatkonzentration im Blut gemessen, und morphologische Parameter an den ersten beiden Kiemenbögen erfasst. Die Fische zeigten niedrige Sauerstoffverbrauchsraten, welche sie bis zu einem sehr niedrigem äußeren Sauerstoffpartialdruck aufrechterhielten. Zusätzlich zeigten sie hohe Hämoglobin- und Laktatkonzentrationen im Blut. Bis zu 75 Tage Normoxie-Akklimatisierung führte zu reduzierter Hämoglobinkonzentration und kürzeren Kiemenfilamenten. / In this thesis, I investigated the morpho-physiology and behavior of two species of African mormyrid weakly electric fish, Marcusenius victoriae and Petrocephalus degeni, in the laboratory and in one of their natural habitats, the Lwamunda Swamp in Uganda. The two main objectives of this work were to (i) observe behavioral rhythms and habitat use patterns of both species under natural and laboratory conditions, and (ii) assess expression and plasticity of morpho-physiological traits that might enable P. degeni to survive naturally occurring low oxygen conditions (hypoxia). Behavioral rhythms were recorded in the laboratory over 42 hours and in the habitat on six sampling days. In the laboratory, both species spent close to 100% of the time in their shelter during the day and actively explore their environment at night. In the swamp lagoon, fish were most often encountered in structurally complex habitats under floating vegetation and ventured into open and unsheltered areas of the lagoon at night, presumably to forage and interact. Concomitant in-situ oxygen measurements revealed that these fish were present, and presumably most active during periods of extreme nocturnal hypoxia in their swamp habitat. To investigate respiratory traits of swamp-dwelling P. degeni, I conducted respirometry experiments and measured blood lactate and hemoglobin and gill morphometrics on the first two gill arches. Fish showed low routine oxygen consumption rates, which they maintained until a very low ambient oxygen partial pressure was reached. Additionally, they had high concentrations of hemoglobin and lactate in their blood. Up to 75 days of normoxia exposure reduced blood hemoglobin and gill filament length.