Welcome to our book review site go-pdf.online!
You may have to Search all our reviewed books and magazines, click the sign up button below to create a free account.
Principles of Catastrophic Forgetting for Continual Semantic Segmentation in Automated Driving
Deep learning excels at extracting complex patterns but faces catastrophic forgetting when fine-tuned on new data. This book investigates how class- and domain-incremental learning affect neural networks for automated driving, identifying semantic shifts and feature changes as key factors. Tools for quantitatively measuring forgetting are selected and used to show how strategies like image augmentation, pretraining, and architectural adaptations mitigate catastrophic forgetting.
Distributed Planning for Self-Organizing Production Systems
In dieser Arbeit wird ein Ansatz entwickelt, um eine automatische Anpassung des Verhaltens von Produktionsanlagen an wechselnde Aufträge und Rahmenbedingungen zu erreichen. Dabei kommt das Prinzip der Selbstorganisation durch verteilte Planung zum Einsatz. - Most production processes are rigid not only by way of the physical layout of machines and their integration, but also by the custom programming of the control logic for the integration of components to a production systems. Changes are time- and resource-expensive. This makes the production of small lot sizes of customized products economically challenging. This work develops solutions for the automated adaptation of production systems based on self-organisation and distributed planning.
Adaptive Umweltmodellierung für kognitive Systeme in offener Welt durch dynamische Konzepte und quantitative Modellbewertung
In dieser Arbeit wird ein Ansatz zur adaptiven Umweltmodellierung betrachtet. Umweltmodelle für kognitive Systeme enthalten oft vorab definierte Domänenmodelle, die für unvorhergesehene Entitäten unzureichend sein können. Der vorgestellte Ansatz adressiert eine adaptive Erweiterung solcher Modelle unter Beachtung der Relevanz von Modellanpassungen. Grundlage ist eine quantitative Modellbewertung, die die Repräsentationsfähigkeit des Modells bezüglich des beobachteten Umgebungszustands bewertet. - In this work, an approach for adaptive world modeling is proposed. World models for cognitive systems often employ predefined domain models, which may become insufficient when encountering unforeseen entities. The presented approach addresses an adaptive extension of such domain models, considering the relevance of proposed model adaptations. As a basis, a quantitative model evaluation is devised, rating the ability of a domain model to represent the currently observed environment state.
Proceedings of the 2022 Joint Workshop of Fraunhofer IOSB and Institute for Anthropomatics, Vision and Fusion Laboratory
In August 2022, Fraunhofer IOSB and IES of KIT held a joint workshop in a Schwarzwaldhaus near Triberg. Doctoral students presented research reports and discussed various topics like computer vision, optical metrology, network security, usage control, and machine learning. This book compiles the workshop's results and ideas, offering a comprehensive overview of the research program of IES and Fraunhofer IOSB.
Multimodal Panoptic Segmentation of 3D Point Clouds
The understanding and interpretation of complex 3D environments is a key challenge of autonomous driving. Lidar sensors and their recorded point clouds are particularly interesting for this challenge since they provide accurate 3D information about the environment. This work presents a multimodal approach based on deep learning for panoptic segmentation of 3D point clouds. It builds upon and combines the three key aspects multi view architecture, temporal feature fusion, and deep sensor fusion.
Self-learning Anomaly Detection in Industrial Production
Configuring an anomaly-based Network Intrusion Detection System for cybersecurity of an industrial system in the absence of information on networking infrastructure and programmed deterministic industrial process is challenging. Within the research work, different self-learning frameworks to analyze passively captured network traces from PROFINET-based industrial system for protocol-based and process behavior-based anomaly detection are developed, and evaluated on a real-world industrial system.
Trustworthy Distributed Usage Control Enforcement in Heterogeneous Trusted Computing Environments
Distributed usage control allows to regulate the use of data even after sharing. However, existing solutions are susceptible to manipulation by dishonest data receivers. This work investigates the use of trusted computing to achieve a trustworthy usage control enforcement process. For this, a suitable system architecture and several remote attestation protocols are designed and implemented. The resulting usage control framework is evaluated using a smart manufacturing application scenario.
Nonlinear Gaussian Filtering : Theory, Algorithms, and Applications
By restricting to Gaussian distributions, the optimal Bayesian filtering problem can be transformed into an algebraically simple form, which allows for computationally efficient algorithms. Three problem settings are discussed in this thesis: (1) filtering with Gaussians only, (2) Gaussian mixture filtering for strong nonlinearities, (3) Gaussian process filtering for purely data-driven scenarios. For each setting, efficient algorithms are derived and applied to real-world problems.
Nutzung von Blickbewegungen für die Mensch-Computer-Interaktion mit dynamischen Bildinhalten am Beispiel der Videobildauswertung
Interaktion mit dynamischen Bildinhalten ist für Systemnutzer herausfordernd bezüglich Wahrnehmung, Kognition und Motorik. Die vorliegende Arbeit identifiziert geeignete blickbasierte Interaktionstechniken zur Selektion bewegter Objekte in Bildfolgen mithilfe mehrerer Querschnitt- und einer Längsschnittstudie. Sie untersucht, wie blickbasierte Interaktion und automatische Verfahren bei der Videobildauswertung unterstützen und ob blickbasierte Klassifikation der Benutzertätigkeit möglich ist. - Interaction with dynamic image content is challenging for user perception, cognition and motor action. This dissertation identifies appropriate gaze-based interaction techniques for moving object selection in image sequences using multiple cross-sectional and one longitudinal user study. Further investigations evaluate how gaze-based interaction and automated image exploitation algorithms support human video exploitation as well as whether gaze-based user task classification is feasible.
Anomaliedetektion in räumlich-zeitlichen Datensätzen
Die Unterstützung des Menschen bei Überwachungsaufgaben ist aufgrund der überwältigenden Menge an Sensordaten von entscheidender Bedeutung. Diese Arbeit konzentriert sich auf die Entwicklung von Datenfusionsmethoden am Beispiel des maritimen Raums. Es werden verschiedene Anomalien untersucht, anhand realer Schiffsverkehrsdaten bewertet und mit Experten erprobt. Dazu werden Situationen von Interesse und Anomalien basierend auf verschiedenen maschinellen Lernverfahren modelliert und evaluiert. - Human support in surveillance tasks is crucial due to the overwhelming amount of sensor data. This work focuses on the development of data fusion methods using the maritime domain as an example. Various anomalies are investigated, evaluated using real vessel traffic data and tested with experts. For this purpose, situations of interest and anomalies are modelled and evaluated based on different machine learning methods.
