Computer Vision (CV) problems, such as image classification and segmentation, have traditionally been solved by manual construction of feature hierarchies or incorporation of other prior knowledge. However, noisy images, varying viewpoints and lighting conditions of images, and clutters in real-world images make the problem challenging. Such tasks cannot be efficiently solved without learning from data. Therefore, many Deep Learning (DL) approaches have recently been successful for various CV tasks, for instance, image classification, object recognition and detection, action recognition, video classification, and scene labeling. The main focus of this thesis is to investigate a purely learning-based approach, particularly, Multi-Dimensional LSTM (MD-LSTM) recurrent neural networks to tackle the challenging CV tasks, classification and segmentation on 2D and 3D image data. Due to the structural nature of MD-LSTM, the network learns directly from raw pixel values and takes the complex spatial dependencies of each pixel into account. This thesis provides several key contributions in the field of CV and DL.
Several MD-LSTM network architectural options are suggested based on the type of input and output, as well as the requiring tasks. Including the main layers, which are an input layer, a hidden layer, and an output layer, several additional layers can be added such as a collapse layer and a fully connected layer. First, a single Two Dimensional LSTM (2D-LSTM) is directly applied on texture images for segmentation and show improvement over other texture segmentation methods. Besides, a 2D-LSTM layer with a collapse layer is applied for image classification on texture and scene images and have provided an accurate classification results. In addition, a deeper model with a fully connected layer is introduced to deal with more complex images for scene labeling and outperforms the other state-of-the-art methods including the deep Convolutional Neural Networks (CNN). Here, several input and output representation techniques are introduced to achieve the robust classification. Randomly sampled windows as input are transformed in scaling and rotation, which are integrated to get the final classification. To achieve multi-class image classification on scene images, several pruning techniques are introduced. This framework provides a good results in automatic web-image tagging. The next contribution is an investigation of 3D data with MD-LSTM. The traditional cuboid order of computations in Multi-Dimensional LSTM (MD-LSTM) is re-arranged in pyramidal fashion. The resulting Pyramidal Multi-Dimensional LSTM (PyraMiD-LSTM) is easy to parallelize, especially for 3D data such as stacks of brain slice images. PyraMiD-LSTM was tested on 3D biomedical volumetric images and achieved best known pixel-wise brain image segmentation results and competitive results on Electron Microscopy (EM) data for membrane segmentation.
To validate the framework, several challenging databases for classification and segmentation are proposed to overcome the limitations of current databases. First, scene images are randomly collected from the web and used for scene understanding, i.e., the web-scene image dataset for multi-class image classification. To achieve multi-class image classification, the training and testing images are generated in a different setting. For training, images belong to a single pre-defined category which are trained as a regular single-class image classification. However, for testing, images containing multi-classes are randomly collected by web-image search engine by querying the categories. All scene images include noise, background clutter, unrelated contents, and also diverse in quality and resolution. This setting can make the database possible to evaluate for real-world applications. Secondly, an automated blob-mosaics texture dataset generator is introduced for segmentation. Random 2D Gaussian blobs are generated and filled with random material textures. These textures contain diverse changes in illumination, scale, rotation, and viewpoint. The generated images are very challenging since they are even visually hard to separate the related regions.
Overall, the contributions in this thesis are major advancements in the direction of solving image analysis problems with Long Short-Term Memory (LSTM) without the need of any extra processing or manually designed steps. We aim at improving the presented framework to achieve the ultimate goal of accurate fine-grained image analysis and human-like understanding of images by machines.
Most of today’s wireless communication devices operate on unlicensed bands with uncoordinated spectrum access, with the consequence that RF interference and collisions are impairing the overall performance of wireless networks. In the classical design of network protocols, both packets in a collision are considered lost, such that channel access mechanisms attempt to avoid collisions proactively. However, with the current proliferation of wireless applications, e.g., WLANs, car-to-car networks, or the Internet of Things, this conservative approach is increasingly limiting the achievable network performance in practice. Instead of shunning interference, this thesis questions the notion of „harmful“ interference and argues that interference can, when generated in a controlled manner, be used to increase the performance and security of wireless systems. Using results from information theory and communications engineering, we identify the causes for reception or loss of packets and apply these insights to design system architectures that benefit from interference. Because the effect of signal propagation and channel fading, receiver design and implementation, and higher layer interactions on reception performance is complex and hard to reproduce by simulations, we design and implement an experimental platform for controlled interference generation to strengthen our theoretical findings with experimental results. Following this philosophy, we introduce and evaluate a system architecture that leverage interference.
First, we identify the conditions for successful reception of concurrent transmissions in wireless networks. We focus on the inherent ability of angular modulation receivers to reject interference when the power difference of the colliding signals is sufficiently large, the so-called capture effect. Because signal power fades over distance, the capture effect enables two or more sender–receiver pairs to transmit concurrently if they are positioned appropriately, in turn boosting network performance. Second, we show how to increase the security of wireless networks with a centralized network access control system (called WiFire) that selectively interferes with packets that violate a local security policy, thus effectively protecting legitimate devices from receiving such packets. WiFire’s working principle is as follows: a small number of specialized infrastructure devices, the guardians, are distributed alongside a network and continuously monitor all packet transmissions in the proximity, demodulating them iteratively. This enables the guardians to access the packet’s content before the packet fully arrives at the receiver. Using this knowledge the guardians classify the packet according to a programmable security policy. If a packet is deemed malicious, e.g., because its header fields indicate an unknown client, one or more guardians emit a limited burst of interference targeting the end of the packet, with the objective to introduce bit errors into it. Established communication standards use frame check sequences to ensure that packets are received correctly; WiFire leverages this built-in behavior to prevent a receiver from processing a harmful packet at all. This paradigm of „over-the-air“ protection without requiring any prior modification of client devices enables novel security services such as the protection of devices that cannot defend themselves because their performance limitations prohibit the use of complex cryptographic protocols, or of devices that cannot be altered after deployment.
This thesis makes several contributions. We introduce the first software-defined radio based experimental platform that is able to generate selective interference with the timing precision needed to evaluate the novel architectures developed in this thesis. It implements a real-time receiver for IEEE 802.15.4, giving it the ability to react to packets in a channel-aware way. Extending this system design and implementation, we introduce a security architecture that enables a remote protection of wireless clients, the wireless firewall. We augment our system with a rule checker (similar in design to Netfilter) to enable rule-based selective interference. We analyze the security properties of this architecture using physical layer modeling and validate our analysis with experiments in diverse environmental settings. Finally, we perform an analysis of concurrent transmissions. We introduce a new model that captures the physical properties correctly and show its validity with experiments, improving the state of the art in the design and analysis of cross-layer protocols for wireless networks.
In this paper, we discuss the problem of approximating ellipsoid uncertainty sets with bounded (gamma) uncertainty sets. Robust linear programs with ellipsoid uncertainty lead to quadratically constrained programs, whereas robust linear programs with bounded uncertainty sets remain linear programs which are generally easier to solve.
We call a bounded uncertainty set an inner approximation of an ellipsoid if it is contained in it. We consider two different inner approximation problems. The first problem is to find a bounded uncertainty set which sticks close to the ellipsoid such that a shrank version of the ellipsoid is contained in it. The approximation is optimal if the required shrinking is minimal. In the second problem, we search for a bounded uncertainty set within the ellipsoid with maximum volume. We present how both problems can be solved analytically by stating explicit formulas for the optimal solutions of these problems.
Further, we present in a computational experiment how the derived approximation techniques can be used to approximate shortest path and network flow problems which are affected by ellipsoidal uncertainty.
Safety-related Systems (SRS) protect from the unacceptable risk resulting from failures of technical systems. The average probability of dangerous failure on demand (PFD) of these SRS in low demand mode is limited by standards. Probabilistic models are applied to determine the average PFD and verify the specified limits. In this thesis an effective framework for probabilistic modeling of complex SRS is provided. This framework enables to compute the average, instantaneous, and maximum PFD. In SRS, preventive maintenance (PM) is essential to achieve an average PFD in compliance with specified limits. PM intends to reveal dangerous undetected failures and provides repair if necessary. The introduced framework pays special attention to the precise and detailed modeling of PM. Multiple so far neglected degrees of freedom of the PM are considered, such as two types of elementwise PM at arbitrarily variable times. As shown by analyses, these degrees of freedom have a significant impact on the average, instantaneous, and maximum PFD. The PM is optimized to improve the average or maximum PFD or both. A well-known heuristic nonlinear optimization method (Nelder-Mead method) is applied to minimize the average or maximum PFD or a weighted trade-off. A significant improvement of the objectives and an improved protection are achieved. These improvements are achieved via the available degrees of freedom of the PM and without additional effort. Moreover, a set of rules is presented to decide for a given SRS if significant improvements will be achieved by optimization of the PM. These rules are based on the well-known characteristics of the SRS, e.g. redundancy or no redundancy, complete or incomplete coverage of PM. The presented rules aim to support the decision whether the optimization is advantageous for a given SRS and if it should be applied or not.
Der Beitrag beschäftigt sich mit der Frage, ob Schildkröten alleine anhand der Musterung bzw. Struktur ihres Bauch- Rückenpanzers eindeutig identifiziert werden können. Dabei sollen sinnvolle Identifizierungsmerkmale entwickelt werden, die auf der Basis von Fotos ausgewertet werden. Das Besondere an diesem Problem ist, dass es mit Lernenden ganz unterschiedlicher Altersstufen bearbeitet werden kann und dass es eine unheimliche Vielfalt an mathematischen Methoden gibt, die auf dem Weg zu einer Lösung hilfreich sind: Dies reicht von einfachen geometrischen Überlegungen über Analysis (Integration, Kurvendiskussion) bis hin zu mathematischer Bildverarbeitung und Fragen der Robustheit. Genauso breit wie das Spektrum der einsetzbaren mathematischen Werkzeuge ist die Altergruppe, mit der ein derartiges Projekt durchführbar ist: Vom Grundschulalter bis hin zur Masterarbeit ist eine Bearbeitung möglich, und die benötigte Zeitspanne reicht von wenigen Stunden bis hin zu mehreren Monaten. Im Beitrag wird die angesprochene Vielfalt exemplarisch gezeigt, so dass die Leser im Idealfall das Projekt genau an die Bedürfnisse ihrer Lerngruppe anpassen können.
Um Spielkarten zu mischen gibt es unterschiedliche Techniken, die sich sowohl in ihrem Zeitaufwand, als auch in der Güte der Durchmischung unterscheiden. Der folgende Artikel vermittelt, wie man die Frage nach einer besonders guten Mischtechnik nutzen kann, um mathematische Modellierung anhand einer alltagsnahen Fragestellung in den Unterricht einzubinden. Dabei können verschiedene Aspekte der Stochastik angesprochen werden, und es bietet sich ein breites Potential, auf unterschiedlichen Niveaus Computer zum Generieren von Zufallsexperimenten zu verwenden.
Die Planung von Bushaltestellen in Innenstädten ist ein authentisches Thema, welches sich für den Einsatz in einem realitätsbezogenen Unterricht in unterschiedlichen Klassenstufen eignet. Verschiedene Interessen und Gegebenheiten müssen in einem Modell und in einer Lösungsstrategie vereint werden. Durch eine sehr offen gewählte Fragestellung sind verschiedene Ansätze und Modelle möglich. Somit wird mathematisches Modellieren trainiert und das Durchlaufen eines Modellierungsprozesses in einem interessanten Projekt ermöglicht. Die mathematischen Hintergründe sowie das vielseitige Lösungsspektrum von Schülerinnen und Schülern unterschiedlicher Jahrgangsstufen zu derselben Fragestellung werden im Folgenden vorgestellt.
Die Autoren befassen sich mit der Ableitung und Bearbeitung eines Modellierungsprojektes aus der populären Sportart Fußball: Ein Freistoß wird unter Beachtung der gegebenen physikalischen Effekte mathematisch modelliert und simuliert. Der Fokus liegt auf der möglichen Durchführung dieses Modellierungsprojekts mit Schülerinnen und Schülern der Sekundarstufe II.
Memory accesses are the bottleneck of modern computer systems both in terms of performance and energy. This barrier, known as "the Memory Wall", can be break by utilizing memristors. Memristors are novel passive electrical components with varying resistance based on the charge passing through the device . In this abstract, the term "memristor" covers also an extension of the definition, memristive devices, which vary their resistance depending on a state variable . While memristors are naturally used as memory cells, they can also be used for other applications, such as logic circuits .
We present a novel architecture that redefines the relationship between the memory and the processor by enabling data processing within the memory itself. Our architecture is based on a memristive memory array, in which we perform two basic logic operations: Imply (material implication)  and False.
This study presents an energy-efficient ultra-low voltage standard-cell based memory in 28nm FD-SOI. The storage element (standard-cell latch) is replaced with a full- custom designed latch with 50 % less area. Error-free operation is demonstrated down to 450mV @ 9MHz. By utilizing body bias (BB) @ VDD = 0.5 V performance spans from 20 MHz @ BB=0V to 110MHz @ BB=1V.