In the human visual system, one of the most prominent functions of the extensive feedback from the higher brain areas within and outside of the visual cortex is attentional modulation. The feedback helps the brain to concentrate its resources on visual features that are relevant for recognition, i. e. it iteratively selects certain aspects of the visual scene for refined processing by the lower areas until the inference process in the higher areas converges to a single hypothesis about this scene. In order to minimize a number of required selection-refinement iterations, one has to find a short sequence of maximally informative portions of the visual input. Since the feedback is not static, the selection process is adapted to a scene that should be recognized. To find a scene-specific subset of informative features, the adaptive selection process on every iteration utilizes results of previous processing in order to reduce the remaining uncertainty about the visual scene. This phenomenon inspired us to develop a computational algorithm solving a visual classification task that would incorporate such principle, adaptive feature selection. It is especially interesting because usually feature selection methods are not adaptive as they define a unique set of informative features for a task and use them for classifying all objects. However, an adaptive algorithm selects features that are the most informative for the particular input. Thus, the selection process should be driven by statistics of the environment concerning the current task and the object to be classified. Applied to a classification task, our adaptive feature selection algorithm favors features that maximally reduce the current class uncertainty, which is iteratively updated with values of the previously selected features that are observed on the testing sample. In information-theoretical terms, the selection criterion is the mutual information of a class variable and a feature-candidate conditioned on the already selected features, which take values observed on the current testing sample. Then, the main question investigated in this thesis is whether the proposed adaptive way of selecting features is advantageous over the conventional feature selection and in which situations. Further, we studied whether the proposed adaptive information-theoretical selection scheme, which is a computationally complex algorithm, is utilized by humans while they perform a visual classification task. For this, we constructed a psychophysical experiment where people had to select image parts that as they think are relevant for classification of these images. We present the analysis of behavioral data where we investigate whether human strategies of task-dependent selective attention can be explained by a simple ranker based on the mutual information, a more complex feature selection algorithm based on the conventional static mutual information and the proposed here adaptive feature selector that mimics a mechanism of the iterative hypothesis refinement. Hereby, the main contribution of this work is the adaptive feature selection criterion based on the conditional mutual information. Also it is shown that such adaptive selection strategy is indeed used by people while performing visual classification.:1. Introduction 2. Conventional feature selection 3. Adaptive feature selection 4. Experimental investigations of ACMIFS 5. Information-theoretical strategies of selective attention 6. Discussion Appendix Bibliography

The main focus of this work lies in the simulation of the deformation of mechanical components which consist of nonlinear elastic, incompressible material and that are subject to large deformations. Starting from a nonlinear formulation one can derive a discrete problem by using linearisation techniques and an adaptive mixed finite element method. This turns out to be a saddle point problem that can be solved via a Bramble-Pasciak conjugate gradient method. With some modifications the simulation can be improved.:1. Introduction 2. Basics 3. Mixed variational formulation 4. Solution method 5. Error estimation 6. LBB conditions 7. Improvement suggestions

Title: Algebraic Error in Matrix Computations in the Context of Numerical Solution of Partial Differential Equations Author: Jan Papež Department: Department of Numerical Mathematics Supervisor: prof. Ing. Zdeněk Strakoš, DrSc., Department of Numerical Mathe- matics Abstract: Solution of algebraic problems is an inseparable and usually the most time-consuming part of numerical solution of PDEs. Algebraic computations are, in general, not exact, and in many cases it is even principally desirable not to perform them to a high accuracy. This has consequences that have to be taken into account in numerical analysis. This thesis investigates in this line some closely related issues. It focuses, in particular, on spatial distribution of the errors of different origin across the solution domain, backward error interpretation of the algebraic error in the context of function approximations, incorporation of algebraic errors to a posteriori error analysis, influence of algebraic errors to adaptivity, and construction of stopping criteria for (preconditioned) iterative algebraic solvers. Progress in these issues requires, in our opinion, understanding the interconnections between the phases of the overall solution process, such as discretization and algebraic computations. Keywords: Numerical solution of partial...

Das stetig wachsende Datenaufkommen - die weltweite Datenmenge verdoppelt sich alle zwei Jahre - ist eine wesentliche Herausforderung für den Menschen in allen Bereichen des beruflichen und privaten Alltags. Um trotzdem relevante Informationen zu identifizieren und auch zu verstehen, nehmen Techniken und Anwendungen zur InfoVis einen immer größeren Stellenwert ein. Leider hat sich die Vision der "InfoVis for and by the masses" aufgrund des notwendigen Daten-, Visualisierungs- und Programmierwissens noch nicht durchgesetzt. Zudem sind heutige InfoVis-Softwareanbieter mit dem Problem konfrontiert, verschiedenste Kontexte, wie Nutzergruppen oder Hard- und Softwareplattformen, unterstützen zu müssen. Ein möglicher Lösungsansatz für dieses Problem ist das Paradigma der kompositen Webanwendungen. Auf deren Basis können Daten und UI-Widgets je nach Anwendungsfall teils automatisch kombiniert werden. Dies erhöht die Wiederverwendbarkeit und spart Zeit sowie Entwicklungskosten. Unter Zuhilfenahme von (semantischen) Modellen ist es zudem möglich, eine komposite RIA an die vorliegende Situation zu adaptieren. Um dem Endanwender Zugang zu den kompositen RIA zu verschaffen, mangelt es jedoch an einem Integrationsprozess, der den speziellen Anforderungen der InfoVis gerecht wird. Diese Dissertation stellt deshalb neue Konzepte für einen ganzheitlichen Semantik-gestützten InfoVis-Prozess vor, der bspw. die Endnutzer-gerechte Filterung großer Datensätze, die kontextsensitive Auswahl von InfoVis-Komponenten, die Nutzerunterstützung bei der Exploration und Interpretation der Daten sowie die Gewinnung und Wiederverwendung von Visualisierungswissen adressiert. Zur Unterstützung des InfoVis-Prozesses werden weiterhin Konzepte für eine formale Wissensbasis mit Domänenwissen vorgeschlagen. Die modulare, mit W3C-Standards prototypisch realisierte Visualisierungsontologie definiert u.a. Konzepte und Relationen zu Daten, graphischen Vokabular, menschlicher Aktivität sowie veränderliches Faktenwissen. Ein weiterer, wesentlicher Beitrag der Arbeit liegt in der Architekturkonzeption für modellbasierte, komposite RIA für die InfoVis-Domäne, womit ein neues Anwendungsfeld des Software-Paradigmas erschlossen wird. Damit steht nun erstmals für eine komposite, webbasierte InfoVis-Lösung ein ganzheitliches Architekturkonzept zur Verfügung, das die Ausführbarkeit der Anwendungen in der heute existierenden, heterogenen Landschaft der (mobilen) Endgeräte gewährleisten kann. Durch die Implementierung entscheidender Architekturkonzepte sowie einer beispielhaften InfoVis-Anwendung für semantische Daten wurde die Tragfähigkeit der geschaffenen Konzepte nachgewiesen. Anhand einer Vielzahl von formativen sowie einer summativen Nutzerstudien konnte validiert werden, dass sich aus den neuen Konzepten Vorteile für den Endanwender bei der Erstellung einer InfoVis ergeben.

Die Zugriffslokalität referenzierter Daten ist eine wichtige Eigenschaft verteilter Anwendungen. Lokales Zwischenspeichern abgefragter entfernter Daten (Caching) wird vielfach bei der Entwicklung solcher Anwendungen eingesetzt, um diese Eigenschaft auszunutzen. Anschliessende Zugriffe auf diese Daten können so beschleunigt werden, indem sie aus dem lokalen Zwischenspeicher bedient werden. Gegenwärtige Middleware-Architekturen bieten dem Anwendungsprogrammierer jedoch kaum Unterstützung für diesen nicht-funktionalen Aspekt. Die vorliegende Arbeit versucht deshalb, Caching als separaten, konfigurierbaren Middleware-Dienst auszulagern. Durch die Einbindung in den Softwareentwicklungsprozess wird die frühzeitige Modellierung und spätere Wiederverwendung caching-spezifischer Metadaten gewährleistet. Zur Laufzeit kann sich das entwickelte System außerdem bezüglich der Cachebarkeit von Daten adaptiv an geändertes Nutzungsverhalten anpassen. / Locality of reference is an important property of distributed applications. Caching is typically employed during the development of such applications to exploit this property by locally storing queried data: Subsequent accesses can be accelerated by serving their results immediately form the local store. Current middleware architectures however hardly support this non-functional aspect. The thesis at hand thus tries outsource caching as a separate, configurable middleware service. Integration into the software development lifecycle provides for early capturing, modeling, and later reuse of cachingrelated metadata. At runtime, the implemented system can adapt to caching access characteristics with respect to data cacheability properties, thus healing misconfigurations and optimizing itself to an appropriate configuration. Speculative prefetching of data probably queried in the immediate future complements the presented approach.

To accommodate the growing demand on higher speeds, low latencies and low energy consumption, the interconnections within and between data centers are supposed to be implemented as optical fiber and waveguide interconnects in future. Optical fiber interconnects provide several advantages over their electrical counterparts as they enable higher bandwidth densities and lower losses at high frequencies over distances longer than few centimeters. However, nowadays optical fiber interconnects are usually not very energy-efficient. The systems in optical networks are mostly optimized for running at their peak performance to transmit the information with the highest available error-free data rate. But the work load of a processor system and hence of an optical link is not constant and varies over time due to the demand of the running applications and users. Therefore, optical interconnects consume the same high power at all times even if lower performance is required. In this paper a new method for the tuning of optical interconnects for on-board and board-to-board optical communication is described. In this way the performance of the transceiver systems of the link is adapted to the present transmission workload and link requirements. If for example lower data rates are required, the bandwidth and therefore the power consumption of the systems can be reduced. This tuning is enabled by the integrated circuitry of the optical link. Different methods for such an adaptive tuning are described and several practical examples are reviewed. By using adaptive bandwidth reduction in the circuits, more than 50 % of the consumed power can be saved. These savings can result in tremendous reductions of the carbon footprint and of the operating costs produced by data centers.

This project describes the role of ritual in the basic entrainment processes of Canadian soldiers. Building on the ecological systems theories of Urie Bronfenbrenner and Roy Rappaport, this project construes human adaptation to occur within multiple interdependent planes of ordered biological, sociostructural, psychosocial and symbolic (even transcendent) meanings and interactions within integrated social ecologies or “living systems.” Rappaport’s theory supports the argument that invariant, embodied actions and impulses not encoded by ritual performers establish social order, values, motivations, competencies, dispositions and representational or symbolic meanings—understood within this project as worldview—circulating within and regulating integrated human ecologies. Ordered sequences of invariant actions and impulses have also come to be conveyed within human phylogenic and ontogenetic developmental processes. This project specifically explores the hypothesis that embodied ritual dynamics pervade the basic entrainment rite of Canadian soldiers. The analysis draws on the ritual theory of Rappaport and the psychosocial developmental theory of Erik Erikson to describe the manner in which innate social regulating impulses and liturgically ordered ritual processes are exploited, in conjunction with predictable human psychosocial developmental imperatives, to build foundational martial dispositions, a spontaneous impulse to radical solidarity and a robust, homogeneous and multivocalic worldview in Canadian soldiers. Such a worldview is adaptive to all aspects of service within the Canadian Armed Forces. The rudimentary martial worldview inscribed upon recruit soldiers and officer candidates forms the foundational background to all subsequent martial meaning and adaptation in so far as it is collectively maintained throughout the military career. This argument maintains that a ritual analysis of adaptive meaning and solidarity among soldiers has profound implications for the structure and direction of future research investigating the persistent and well documented rates of distress, maladaptation and health pathology among serving members of the Canadian Armed Forces.

The Special Session on “Pathways to Servers of the Future” outlines a new research program set up at Technische Universität Dresden addressing the increasing energy demand of global internet usage and the resulting ecological impact of it. The program pursues a novel holistic approach that considers hardware as well as software adaptivity to significantly increase energy efficiency, while suitably addressing application demands. The session presents the research challenges and industry perspective.

Este trabalho apresenta a formulação de um sistema híbrido de apoio à decisão que, através de técnicas adaptativas, permite que múltiplos dispositivos sejam utilizados de forma colaborativa para encontrar uma solução para um problema de tomada de decisão. É proposta uma estratégia particular para o trabalho colaborativo que restringe o acesso aos dispositivos mais lentos com base na dificuldade encontrada pelos dispositivos mais rápidos para solucionar um problema específico. As soluções encontradas por cada dispositivo são propagadas aos demais, permitindo que cada um deles agregue estas novas soluções com o auxílio de técnicas adaptativas. É feito um estudo sobre aprendizagem de máquina mediante incertezas para verificar e minimizar os impactos negativos que uma nova solução, possivelmente errônea, possa ter. O sistema híbrido proposto é apresentado numa aplicação particular, utilizando testes padronizados para compará-lo com os dispositivos individuais que o compõem e com sistemas híbridos de mesma finalidade. Através destes testes, é mostrado que dispositivos consolidados, mesmo que de naturezas distintas, podem ser utilizados de maneira colaborativa, permitindo não só calibrar um compromisso entre o tempo de resposta e a taxa de acerto, mas também evoluir de acordo com o histórico de problemas processados. / This work presents a formulation of a hybrid decision-making system that employs adaptive techniques as a way to coordinate multiple devices in order to make a collaborative decision. The strategy proposed here is to restrict the use of slower devices, based on how difficult the specific problem is - easier problems may be solved on faster devices. Each device is able to learn through solutions given by the others, aggregating new knowledge with the aid of adaptive techniques. In order to evaluate and minimize the negative impact those new solutions may have, a study concerning machine learning under uncertainty is carried out. A particular application of this system has been tested and compared, not only to each individual device that is part of the system itself, but to similar hybrid systems as well. It is shown that even devices of distinct natures may be reused in a collaborative manner, making it possible to calibrate the trade-off between hit rate and response time, and to evolve according to the input stimuli received as well.

Este trabalho apresenta a formulação de um sistema híbrido de apoio à decisão que, através de técnicas adaptativas, permite que múltiplos dispositivos sejam utilizados de forma colaborativa para encontrar uma solução para um problema de tomada de decisão. É proposta uma estratégia particular para o trabalho colaborativo que restringe o acesso aos dispositivos mais lentos com base na dificuldade encontrada pelos dispositivos mais rápidos para solucionar um problema específico. As soluções encontradas por cada dispositivo são propagadas aos demais, permitindo que cada um deles agregue estas novas soluções com o auxílio de técnicas adaptativas. É feito um estudo sobre aprendizagem de máquina mediante incertezas para verificar e minimizar os impactos negativos que uma nova solução, possivelmente errônea, possa ter. O sistema híbrido proposto é apresentado numa aplicação particular, utilizando testes padronizados para compará-lo com os dispositivos individuais que o compõem e com sistemas híbridos de mesma finalidade. Através destes testes, é mostrado que dispositivos consolidados, mesmo que de naturezas distintas, podem ser utilizados de maneira colaborativa, permitindo não só calibrar um compromisso entre o tempo de resposta e a taxa de acerto, mas também evoluir de acordo com o histórico de problemas processados. / This work presents a formulation of a hybrid decision-making system that employs adaptive techniques as a way to coordinate multiple devices in order to make a collaborative decision. The strategy proposed here is to restrict the use of slower devices, based on how difficult the specific problem is - easier problems may be solved on faster devices. Each device is able to learn through solutions given by the others, aggregating new knowledge with the aid of adaptive techniques. In order to evaluate and minimize the negative impact those new solutions may have, a study concerning machine learning under uncertainty is carried out. A particular application of this system has been tested and compared, not only to each individual device that is part of the system itself, but to similar hybrid systems as well. It is shown that even devices of distinct natures may be reused in a collaborative manner, making it possible to calibrate the trade-off between hit rate and response time, and to evolve according to the input stimuli received as well.