Refine
Document Type
- Master's Thesis (32) (remove)
Has Fulltext
- yes (32) (remove)
Keywords
- Artificial Intelligence (1)
- Classification (1)
- Clustering (1)
- Computersimulation (1)
- Criteria Matrix (1)
- Data Mining (1)
- Data analysis (1)
- Datenanalyse (1)
- Dynamische Analyse (1)
- Echtzeitphysik (1)
Institute
- Fachbereich 4 (32) (remove)
In this thesis the possibilities for real-time visualization of OpenVDB
files are investigated. The basics of OpenVDB, its possibilities, as well
as NanoVDB and its GPU port, were studied. A system was developed
using PNanoVDB, the graphics API port of OpenVDB. Techniques were
explored to improve and accelerate a single ray approach of ray tracing.
To prove real-time capability, two single scattering approaches were
also implemented. One of these was selected, further investigated and
optimized to achieve interactive real-time rendering.
It is important to give artists immediate feedback on their adjustments, as
well as the possibility to change all parameters to ensure a user friendly
creation process.
In addition to the optical rendering, corresponding benchmarks were
collected to compare different improvement approaches and to prove
their relevance. Attention was paid to the rendering times and memory
consumption on the GPU to ensure optimal use. A special focus, when
rendering OpenVDB files, was put on the integrability and extensibility of
the program to allow easy integration into an existing real-time renderer
like U-Render.
On-screen interactive presentations have got immense popularity in the domain of attentive interfaces recently. These attentive screens adapt their behavior according to the user's visual attention. This thesis aims to introduce an application that would enable these attentive interfaces to change their behavior not just according to the gaze data but also facial features and expressions. The modern era requires new ways of communications and publications for advertisement. These ads need to be more specific according to people's interests, age, and gender. When advertising, it's important to get a reaction from the user but not every user is interested in providing feedback. In such a context more, advance techniques are required that would collect user's feedback effortlessly. The main problem this thesis intends to resolve is, to apply advanced techniques of gaze and face recognition to collect data about user's reactions towards different ads being played on interactive screens. We aim to create an application that enables attentive screens to detect a person's facial features, expressions, and eye gaze. With eye gaze data we can determine the interests and with facial features, age and gender can be specified. All this information will help in optimizing the advertisements.
Remote rendering services offer the possibility to stream high quality images to lower powered devices. Due to the transmission of data the interactivity of applications is afflicted with a delay. A method to reduce delay of the camera manipulation on the client is called 3d-warping. This method causes artifacts. In this thesis different approaches of remote rendering setups will be shown. The artifacts and improvements of the warping method will be described. Methods to reduce the artifacts will be implemented and analyzed.
Aufgrund des branchenweiten Bedarfs den Konkurrenzkampf zu umgehen, entwickelten Kim und Mauborgne die Blue Ocean Strategy, um neue Märkte zu ergründen. Diese bezeichnen sie als einzigartig. Da jedoch weitere Strategien zur Ergründung neuer Märkte existieren, ist es das Ziel dieser Arbeit herauszufinden, anhand welcher Charakterisierungsmerkmale die Blue Ocean Strategy als einzigartig angesehen werden kann.
Die Strategie von Kim und Mauborgne soll daher mit Schumpeters schöpferischen Zerstörung, Ansoffs Diversifikationsstrategie, Porters Nischenstrategie und Druckers Innovationsstrategien verglichen werden. Für den Vergleich werden die Charakterisierungsmerkmale herangezogen, nach denen Kim und Mauborgne die Blue Ocean Strategy als einzigartig beurteilen. Auf Basis dieser Kriterien wird ein Metamodell entwickelt, mit dessen Hilfe die Untersuchung durchgeführt wird.
Der Vergleich zeigt, dass die Konzepte von Schumpeter, Ansoff, Porter und Drucker in einigen Kriterien der Blue Ocean Strategy ähneln. Keine der Strategien verhält sich jedoch in allen Punkten so wie das Konzept von Kim und Mauborgne. Während die Blue Ocean Strategy ein Differenzierung und Senkung der Kosten anstrebt, orientieren sich die meisten Konzepte entweder an einer Differenzierung oder an einer Kostenreduktion. Auch die Betretung des neuen Marktes wird unterschiedlich interpretiert. Während die Blue Ocean Strategy auf einen Markt abzielt, der unergründet ist und somit keinen Wettbewerb vorweist, werden bei den anderen Strategien oft bestehende Märkte als neu interpretiert, auf denen das Unternehmen bisher nicht agiert hat. Dies schließt die vorherige Existenz der Märkte jedoch nicht aus.
Auf Basis der durch den Vergleich gezogenen Erkenntnisse, kann somit die Blue Ocean Strategy als einzigartig bezeichnet werden.
Data Mining im Fußball
(2014)
The term Data Mining is used to describe applications that can be applied to extract useful information from large datasets. Since the 2011/2012 season of the german soccer league, extensive data from the first and second Bundesliga have been recorded and stored. Up to 2000 events are recorded for each game.
The question arises, whether it is possible to use Data Mining to extract patterns from this extensive data which could be useful to soccer clubs.
In this thesis, Data Mining is applied to the data of the first Bundesliga to measure the value of individual soccer players for their club. For this purpose, the state of the art and the available data are described. Furthermore, classification, regression analysis and clustering are applied to the available data. This thesis focuses on qualitative characteristics of soccer players like the nomination for the national squad or the marks players get for their playing performance. Additionally this thesis considers the playing style of the available players and examines if it is possible to make predictions for upcoming seasons. The value of individual players is determined by using regression analysis and a combination of cluster analysis and regression analysis.
Even though not all applications can achieve sufficient results, this thesis shows that Data Mining has the potential to be applied to soccer data. The value of a player can be measured with the help of the two approaches, allowing simple visualization of the importance of a player for his club.
Systems to simulate crowd-behavior are used to simulate the evacuation of a crowd in case of an emergency. These systems are limited to the moving-patterns of a crowd and are generally not considering psychological and/or physical conditions. Changing behaviors within the crowd (e.g. by a person falling down) are not considered.
For that reason, this thesis will examine the psychological behavior and the physical impact of a crowd- member on the crowd. In order to do so, this study develops a real-time simulation for a crowd of people, adapted from a system for video games. This system contains a behavior-AI for agents. In order to show physical interaction between the agents and their environment as well as their movements, the physical representation of each agent is realized by using rigid bodies from a physics-engine. The movements of the agents have an additional navigation mesh and an algorithm for collision avoidance.
By developing a behavior-AI a physical and psychological state is reached. This state contains a psychological stress-level as well as a physical condition. The developed simulation is able to show physical impacts such as crowding and crushing of agents, interaction of agents with their environment as well as factors of stress.
By evaluating several tests of the simulation, this thesis examines whether the combination of physical and psychological impacts is implementable successfully. If so, this thesis will be able to give indications of an agent- behavior in dangerous and/or stressful situations as well as a valuation of the complex physical representation.
Ziel dieser Ausarbeitung ist es, das Wippe-Experiment gemäß dem Aufbau innerhalb der AG Echtzeitsysteme unter Leitung von Professor Dr. Dieter Zöbel mithilfe eines LEGO Mindstorms NXT Education-Bausatzes funktionsfähig nachzubauen und das Vorgehen zu dokumentieren. Der dabei entstehende Programmcode soll didaktisch aufbereitet und eine Bauanleitung zur Verfügung gestellt werden. Dies soll gewährleisten, dass Schülerinnen und Schüler auch ohne direkten Zugang zu einer Hochschule oder ähnlichem Institut den Versuchsaufbau Wippe möglichst unkompliziert im Klassenraum erleben können.
Im Rahmen des "Design Thinking"-Prozesses kommen unterschiedliche Varianten kreativitätsfördernder Techniken zum Einsatz. Aufgrund der zunehmenden Globalisierung ergeben sich immer häufiger Kollaborationen, bei denen sich die jeweiligen Projektteilnehmer an verteilten Standorten befinden. Somit erweist sich die Digitalisierung des Design-Prozesses als durchaus erstrebenswert. Ziel der hier vorliegenden Untersuchung ist daher die Erstellung eines Bewertungsschemas, welches die Eignung digitaler Kreativitätstechniken in Bezug auf das "Entrepreneurial Design Thinking" misst. Des Weiteren soll geprüft werden, inwiefern sich der Einsatz von e-Learning-Systemen in Kombination mit der Verwendung digitaler Kreativitätstechniken eignet. Diese Prüfung soll am Beispiel der e-Learning Software "WebCT" konkretisiert werden. Hieraus ergibt sich die folgende Fragestellung: Welche digitalen Kreativitätstechniken eignen sich für die Anwendung im Bereich des "Entrepreneurial Design Thinkings" unter Einsatz der e-Learning Plattform "WebCT"? Zunächst wird eine Literaturanalyse bezüglich des "Entrepreneurial Design Thinkings", der klassische und digitale Kreativitätstechniken sowie des Arbeitens in Gruppen, was auch das Content Management, e-Learning-Systeme und die Plattform "WebCT" beinhaltet, durchgeführt. Im Anschluss daran wird eine qualitative Untersuchung durchgeführt. Auf Basis bereits bestehender Literatur, soll ein Bewertungsschema erstellt werden, welches misst, welche der behandelten digitalen Kreativitätstechniken für den Einsatz im "Entrepreneurial Design Thinking" am besten geeignet ist. Darauf aufbauend erfolgt die Verknüpfung des digitalisierten "Design Thinking"-Prozesses mit der e-Learning Plattform "WebCT". Abschließend wird diskutiert, in wie fern diese Zusammenführung als sinnvoll erachtet werden kann.
ERP market analysis
(2013)
Der aktuelle ERP Markt wird dominiert von den fünf größten Anbietern SAP, Oracle, Microsoft, Infor und Sage. Da der Markt und die angebotenen Lösungen vielfältig sind, bedarf es einer fundierten Analyse der Systeme. Die Arbeit beleuchtet dabei anhand ausgesuchter Literatur und Kennzahlen der verschiedenen Unternehmen die theoretische Seite der angebotenen Lösungen der fünf großen ERP Anbieter. Daneben wird die Nutzung der Systeme in der Praxis anhand der Befragung von sechs Anwendern analysiert und die Systeme miteinander verglichen.
Ziel der Arbeit ist es, dass die Forschungsfragen beantwortet werden und dass es bezogen auf die Systeme dem Leser der Arbeit ersichtlich wird, welches ERP System für welche Unternehmensbranche und Unternehmensgröße am besten geeignet ist.
Des Weiteren gibt die Arbeit Aufschluss darüber, welche Trends für ERP Systeme für die Zukunft zu erwarten sind und welche Herausforderungen sich dadurch für die Unternehmen stellen.
Large amounts of qualitative data make the utilization of computer-assisted methods for their analysis inevitable. In this thesis Text Mining as an interdisciplinary approach, as well as the methods established in the empirical social sciences for analyzing written utterances are introduced. On this basis a process of extracting concept networks from texts is outlined and the possibilities of utilitzing natural language processing methods within are highlighted. The core of this process is text processing, to whose execution software solutions supporting manual as well as automated work are necessary. The requirements to be met by these solutions, against the background of the initiating project GLODERS, which is devoted to investigating extortion racket systems as part of the global fiσnancial system, are presented, and their fulσlment by the two most preeminent candidates reviewed. The gap between theory and pratical application is closed by a prototypical application of the method to a data set of the research project utilizing the two given software solutions.
Today you can find smartphones everywhere. This situation created a hype for Augmented Reality and AR Apps. The big question is: Do these applications provide a real added value? To make AR pratically it is important to add the computational power of a computer to the advantages of AR. An easy and fast way of interaction is essential.
A Poker-Assistance-Software is an ideal test area for an AR Application with real added value. The estimation of the winning probability and a fast automated tracking of the playing cards is the perfect field of investigation.
In this discussion it is interesting to evaluate the added value of AR Applications in common.
We present the conceptual and technological foundations of a distributed natural language interface employing a graph-based parsing approach. The parsing model developed in this thesis generates a semantic representation of a natural language query in a 3-staged, transition-based process using probabilistic patterns. The semantic representation of a natural language query is modeled in terms of a graph, which represents entities as nodes connected by edges representing relations between entities. The presented system architecture provides the concept of a natural language interface that is both independent in terms of the included vocabularies for parsing the syntax and semantics of the input query, as well as the knowledge sources that are consulted for retrieving search results. This functionality is achieved by modularizing the system's components, addressing external data sources by flexible modules which can be modified at runtime. We evaluate the system's performance by testing the accuracy of the syntactic parser, the precision of the retrieved search results as well as the speed of the prototype.
This master thesis deals basically with the design and implementation of a path planning system based on rapidly exploring search trees for general-n-trailers. This is a probabilistic method that is characterized by a fast and uniform exploration. The method is well established, however, has been applied only to vehicles with simple kinematics to date. General-n-trailers represent a particular challenge as their controllability is limited. For this reason the focus of this thesis rests on the application of the mentioned procedure to general-n-trailers. In this context systematic correlations between the characteristics of general-n-trailers and the possibilities for the realization and application of the method are analyzed.
This thesis deals with the development and evaluation of a concept of novel interaction with ubiquitous user interfaces. To accomplish the evaluation of this interaction concept, a prototype was implementated by using an existing head-mounted display solution and an android smartphone.
Furthermore, in the course of this thesis, a concrete use case for this prototype " the navigation through a city block with the aid of an electronic map " was developed and built as an executable application to help evaluate the quality of the interaction concept. Therefore, fundamental research results were achieved.
Im Rahmen dieser Masterarbeit wird ein umfassender Überblick über die Vielfalt der Sicherheitsmodelle gegeben, indem ausgewählte Sicherheitsmodelle beschrieben, klassifiziert und miteinander verglichen werden.
Sicherheitsmodelle beschreiben in einer abstrakten Weise die sicherheitsrelevanten
Komponenten und Zusammenhänge eines Systems. Mit den Sicherheitsmodellen können komplexe Sachverhalte veranschaulicht und analysiert werden.
Da Sicherheitsmodelle unterschiedliche Sicherheitsaspekte behandeln, beschäftigt
sich diese Arbeit mit der Ausarbeitung eines Klassifizierungsschemas, welches
die strukturelle und konzeptuelle Besonderheiten der Modelle in Bezug auf die zugrundeliegenden Sicherheitsaspekte beschreibt. Im Rahmen des Klassifizierungsschemas werden die drei grundlegenden Modellklassen gebildet: Zugriffskontrollmodelle, Informationsflussmodelle und Transaktionsmodelle.
Sicherheitsmodelle werden in einem direkten und indirekten Vergleich gegenüber gestellt. Im letzten Fall werden sie einer oder mehrerer Modellklassen des Klassifizierungsschemas zugeordnet. Diese Klassifizierung erlaubt, Aussagen über die betrachteten Sicherheitsaspekte und die strukturellen bzw. konzeptuellen Besonderheiten eines Sicherheitsmodells in Bezug auf die anderen Sicherheitsmodelle
zu machen.
Beim direkten Vergleich werden anhand der ausgewählten Kriterien die Eigenschaften
und Aspekte der Sicherheitsmodelle orthogonal zu den Modellklassen
betrachtet.
Human detection is a key element for human-robot interaction. More and more robots are used in human environments, and are expected to react to the behavior of people. Before a robot can interact with a person, it must be able to detect it at first. This thesis presents a system for the detection of humans and their hands using a RGB-D camera. First, a model based hypotheses for possible positions of humans are created to recognize a person. By using the upper parts of the body are used to extract, new features based on relief and width of a person- head and shoulders are extracted. The hypotheses are checked by classifying the features with a support vector machine (SVM). The system is able to detect people in different poses. Both sitting and standing humans are found, by using the visible upper parts of the person. Moreover, the system is able to recognize if a human is facing or averting the sensor. If the human is facing the sensor, the color information and the distance between hand and body are used to detect the positions of the person- hands. This information is useful for gestures recognition and thus can further enhances human-robot interaction.
Pedestrian Detection in digital images is a task of huge importance for the development of automaticsystems and in improving the interaction of computer systems with their environment. The challenges such a system has to overcome are the high variance of the pedestrians to be recognized and the unstructured environment. For this thesis, a complete system for pedestrian detection was implemented according to a state of the art technique. A novel insight about precomputing the Color Self-Similarity accelerates the computations by a factor of four. The complete detection system is described and evaluated, and was published under an open source license.
In a software reengineering task legacy systems are adapted computer-aided to new requirements. For this an efficient representation of all data and information is needed. TGraphs are a suitable representation because all vertices and edges are typed and may have attributes. Further more there exists a global sequence of all graph elements and for each vertex exists a sequence of all incidences. In this thesis the "Extractor Description Language" (EDL) was developed. It can be used to generate an extractor out of a syntax description, which is extended by semantic actions. The generated extractor can be used to create a TGraph representation of the input data. In contrast to classical parser generators EDL support ambiguous grammars, modularization, symbol table stacks and island grammars. These features simplify the creation of the syntax description. The collected requirements for EDL are used to determine an existing parser generator which is suitable to realize the requirements.
After that the syntax and semantics of EDL are described and implemented using the suitable parser generator. Following two extractors one for XML and one for Java are created with help of EDL. Finally the time they need to process some input data is measured.
The annotation of digital media is no new area of research, instead it is widely investigated. There are many innovative ideas for creating the process of annotation. The most extensive segment of related work is about semi automatic annotation. One characteristic is common in the related work: None of them put the user in focus. If you want to build an interface, which is supporting and satsfying the user, you will have to do a user evaluation first. Whithin this thesis we want to analyze, which features an interface should or should not have to meet these requirements of support, user satisfaction and beeing intuitive. After collecting many ideas and arguing with a team of experts, we determined only a few of them. Different combination of these determined variables form the interfaces, we have to investigate in our usability study. The results of the usability leads to the assumption, that autocompletion and suggestion features supports the user. Furthermore coloring tags for grouping them into categories is not disturbing to the user, but has a tendency of being supportive. Same tendencies emerge for an interface consisting of two user interface elements. There is also an example given for the definition differences of being intuitive. This thesis leads to the concolusion that for reasons of user satisfaction and support it is allowed to differ from classical annotation interface features and to implement further usability studies in the section of annotation interfaces.
In this master thesis some new helpful features will be added to the Spanning Tree Simulator. This simulator was created by Andreas Sebastian Janke in his bachelor thesis [Jan10b] in 2010. It is possible to visualize networks which are defined in a configuration file. Each of them is a XML representation of a network consisting of switches and hosts. After loading such a file into the program it is possible to run the Spanning Tree Algorithm IEEE 802.1D. In contrast to the previous version only the switches are implemented as threads. When the algorithm is finished a spanning tree is built. This means, that messages cannot run into loops anymore. This is important because loops can cause a total breakdown of the communication in a network, if the running routing protocols cannot handle loops.