Kaiserslautern - Fachbereich Informatik
Refine
Year of publication
Document Type
- Doctoral Thesis (229) (remove)
Has Fulltext
- yes (229)
Keywords
- Visualisierung (20)
- Visualization (8)
- Deep Learning (7)
- Computergraphik (5)
- Evaluation (4)
- Robotik (4)
- Artificial Intelligence (3)
- Bildverarbeitung (3)
- Geoinformationssystem (3)
- Machine Learning (3)
- Mustererkennung (3)
- Optische Zeichenerkennung (3)
- Ray casting (3)
- Recommender Systems (3)
- Scientific Visualization (3)
- Semantic Web (3)
- Unsicherheit (3)
- computer graphics (3)
- document analysis (3)
- machine learning (3)
- optical character recognition (3)
- verification (3)
- visualization (3)
- AG-RESY (2)
- Activity recognition (2)
- Algorithmus (2)
- Automatische Differentiation (2)
- Code Generation (2)
- Computational Fluid Dynamics (2)
- Datenanalyse (2)
- Datenbank (2)
- Dreidimensionale Bildverarbeitung (2)
- Effizienter Algorithmus (2)
- Endlicher Automat (2)
- Implementierung (2)
- Interaction (2)
- Interaktion (2)
- Layout (2)
- Machine learning (2)
- Mensch-Maschine-Kommunikation (2)
- Merkmalsextraktion (2)
- Model Checking (2)
- Modellierung (2)
- Natural Language Processing (2)
- Network Calculus (2)
- Optimierung (2)
- Pattern Recognition (2)
- Raumakustik (2)
- Room acoustics (2)
- SDL (2)
- Scattered-Data-Interpolation (2)
- Simulation (2)
- Softwareentwicklung (2)
- Topology (2)
- UML (2)
- Uncertain Data (2)
- Uncertainty Visualization (2)
- Visual Analytics (2)
- Voronoi-Diagramm (2)
- Wearable computing (2)
- WiFi (2)
- artificial intelligence (2)
- computer vision (2)
- deep learning (2)
- image processing (2)
- layout analysis (2)
- semantic web (2)
- virtual acoustics (2)
- 3D Gene Expression (1)
- 3D Point Data (1)
- ASM (1)
- AUTOSAR (1)
- Ableitungsschätzung (1)
- Abrechnungsmanagement (1)
- Abstraction (1)
- Abstraction-Based Controller Design (1)
- Abstraktion (1)
- Accelerometer (1)
- Accounting Agent (1)
- Ad-hoc networks (1)
- Ad-hoc-Netz (1)
- Adaptive Data Structure (1)
- Affine Arithmetic (1)
- Akquisition (1)
- Algorithm (1)
- Algorithmic Differentiation (1)
- Amharic, Attention, Factored Convolutional Neural Network, OCR (1)
- Anomaly Detection (1)
- Application Framework (1)
- Association (1)
- Ausdrucksfähig (1)
- Ausdrucksfähigkeit (1)
- Automat <Automatentheorie> (1)
- Automatic Differentiation (1)
- Automatic Image Captioning (1)
- Automation (1)
- Automatisches Beweisverfahren (1)
- Autonomer Agent (1)
- Autonomer Roboter (1)
- Bahnplanung (1)
- Befahrbarkeitsanalyse (1)
- Benutzer (1)
- Beschränkte Arithmetik (1)
- Bibligraphic References (1)
- Bio-inspired (1)
- Bioinformatik (1)
- Bipedal Locomotion (1)
- Bluetooth (1)
- Boosting (1)
- CBR (1)
- Caching (1)
- Carbon footprint (1)
- Channel Hopping (1)
- Channel Scheduling (1)
- Channel hopping (1)
- Channel sensing (1)
- Classification (1)
- Classification of biomedical signals (1)
- Cluster-Analyse (1)
- Clustering (1)
- Cognitive Amplification (1)
- Collaboration (1)
- Combinatorial Testing (1)
- Computer Graphic (1)
- Computer Supported Cooperative Work (1)
- Computer graphics (1)
- Computer-Aided Diagnosis (1)
- Computerphysik (1)
- Computersimulation (1)
- Computertomographie (1)
- Computervision (1)
- Concurrent data structures (1)
- Congitive Radio Networks (1)
- Containertypen (1)
- Containertypes (1)
- Controller Synthesis (1)
- Correct-by-Design Controller Synthesis (1)
- Cyber-Physical Systems (1)
- Cycle Accuracy (1)
- DCE <Programm> (1)
- DFG (1)
- DPN (1)
- Data Modeling (1)
- Dataset (1)
- Datenbanken (1)
- Datenfusion (1)
- Decision Support Systems (1)
- Delaunay-Triangulierung (1)
- Derivative Estimation (1)
- Dienstgüte (1)
- Dienstschnittstellen (1)
- Direct Numerical Simulation (1)
- Discrete Event Simulation (DES) (1)
- Distributed Rendering (1)
- Distributed system (1)
- Dreidimensionale Rekonstruktion (1)
- Dreidimensionale Strömung (1)
- Duplicate Identification (1)
- Duplikaterkennung (1)
- Dynamischer Test (1)
- ESTELLE (1)
- Effizienz (1)
- Effizienzsteigerung (1)
- Empfehlungssysteme (1)
- Energie (1)
- Engineering 4.0 (1)
- Ensemble Visualization (1)
- Entwurf (1)
- Eventual consistency (1)
- Evolutionary Algorithm (1)
- Experiment (1)
- Experimentation (1)
- Explainability (1)
- Explainable Artificial Intelligence (1)
- Exposed Datapath Architectures (1)
- Expressiveness (1)
- Extrapolation (1)
- Eyewear Computing (1)
- FERAL (1)
- Fahrtkostenmodelle (1)
- Fast Mode-Signaling (1)
- Fault Tree Analysis (1)
- Feasibility study (1)
- Feature (1)
- Feature Detection (1)
- Feature Extraction (1)
- Fehlerbaumanalyse (1)
- Fließanalyse (1)
- Flow Visualization (1)
- Formale Beschreibungstechnik (1)
- Formale Grammatik (1)
- Formale Sprache (1)
- Formaler Beweis (1)
- Framework (1)
- Frequenzsprungverfahren (1)
- Functional Safety (1)
- Funktionale Sicherheit (1)
- Fusion (1)
- Future Internet (1)
- GPU (1)
- Gateway (1)
- Gauß-Filter (1)
- Gefahren- und Risikoanalyse (1)
- Gene expression programming (1)
- Generierung (1)
- Geovisualization (1)
- Google Earth (1)
- HPC (1)
- Hazard Analysis (1)
- Heterogeneous (1)
- Hitting families (1)
- Human Pose (1)
- Human-Computer Interaction (1)
- Human-Robot-Coexistence (1)
- Human-Robot-Cooperation (1)
- Hybrid CBR (1)
- Hypergraph (1)
- IEEE 802.15.4 (1)
- IMU (1)
- IP Address (1)
- IP Traffic Accounting (1)
- IP-XACT (1)
- ISO 26262 (1)
- ISO26262 (1)
- Image Processing (1)
- Imote2 (1)
- Implementation (1)
- Incremental recomputation (1)
- Induktive logische Programmierung (1)
- Information Extraction (1)
- Information Visualization (1)
- Intensity estimation (1)
- Intermediate Composition (1)
- Interpolation (1)
- Invariante (1)
- Invariante Momente (1)
- Kellerautomat (1)
- Klassifikation (1)
- Klima (1)
- Knowledge Management (1)
- Knuth-Bendix completion (1)
- Knuth-Bendix-Vervollständigung (1)
- Kommunikation (1)
- Komponentenmodell (1)
- Kompression (1)
- Kontextbezogenes System (1)
- Korrelationsanalyse (1)
- Kurve (1)
- Künstliche Intelligenz (1)
- LIDAR (1)
- LIR-Tree (1)
- LOADBAL (1)
- Large Data (1)
- Large High-Resolution Displays (1)
- Laser Wakefield Particle Accelerator (1)
- Learning Analytics (1)
- Leistungsmessung (1)
- Link Metric (1)
- Linked Data (1)
- Linking Data Analysis and Visualization (1)
- Lokalisierung (1)
- Lärmbelastung (1)
- Lärmimmission (1)
- MAC protocols (1)
- MIP-Emissionsspektroskopie (1)
- MIP-Massenspektrometrie (1)
- Magnetfeldbasierter Lokalisierung (1)
- Magnetfelder (1)
- Magnetometer (1)
- Manufacturing (1)
- MapReduce (1)
- Maschinelle Übersetzung (1)
- Matrix Completion (1)
- Maximum Intensity Projection (1)
- Medical Image Analysis (1)
- Mehragentensystem (1)
- Memory Architecture (1)
- Memory Consistency (1)
- Mensch-Roboter-Koexistenz (1)
- Mensch-Roboter-Kooperation (1)
- Menschenmenge (1)
- Merkmalsraum (1)
- Mesh-Free (1)
- Meter (1)
- Microelectromechanical Systems (1)
- Mikroskopie (1)
- Minimal Cut Set Visualization (1)
- Minimal training (1)
- Mobile Computing (1)
- Mobile system (1)
- Mobiler Roboter (1)
- Mode-Based Scheduling with Fast Mode-Signaling (1)
- Modellgetriebene Entwicklung (1)
- Modelling (1)
- Modularisierung (1)
- Modusbasierte Signalisierung (1)
- Molekulare Bioinformatik (1)
- Moment Invariants (1)
- Moment-Generating Functions (1)
- Mood-based Music Recommendations (1)
- Multi-Edge Graph (1)
- Multi-Field (1)
- Multi-Variate Data (1)
- Multidisciplinary Optimization (1)
- Multifield Data (1)
- Nachhaltigkeit (1)
- Natural Neighbor (1)
- Natural Neighbor Interpolation (1)
- Natürliche Nachbarn (1)
- Navigation (1)
- Network Architecture (1)
- Netzwerk (1)
- Neural Architecture Search (1)
- Neural Networks (1)
- NoSQL (1)
- Node-Link Diagram (1)
- Numerische Strömungssimulation (1)
- OCR (1)
- OWL (1)
- Object-orientation (1)
- Objektorientierung (1)
- Off-road Robotics (1)
- Off-road Robotik (1)
- Online chain partitioning (1)
- Open Estelle (1)
- Optimierender Compiler (1)
- Optimization (1)
- Paralleler Hybrid (1)
- Pareto Optimality (1)
- Partial Differential Equations (1)
- Partially ordered sets (1)
- Participatory Sensing (1)
- Performance (1)
- Personalisation (1)
- Pervasive health (1)
- Physical activity monitoring (1)
- Planar Pressure (1)
- Privacy (1)
- Process Data (1)
- Processor Architecture (1)
- Processor Architectures (1)
- Protocol Composition (1)
- Prototyp (1)
- Prototype (1)
- Prozessvisualisierung (1)
- Quality (1)
- Qualität (1)
- Quantitative Bildanalyse (1)
- Quartz (1)
- Quicksort (1)
- Random testing (1)
- Rate Gyro (1)
- Raumordnung (1)
- Ray tracing (1)
- Rechtecksgitter (1)
- Rectilinear Grid (1)
- Redundanzvermeidung (1)
- Regelung (1)
- Reinforcement Learning (1)
- Rekonstruktion (1)
- Representation (1)
- Risikobewertung (1)
- Risk Assessment (1)
- Roboter (1)
- Robotics (1)
- Routing (1)
- Räumliche Statistik (1)
- SAHARA (1)
- SCAD (1)
- SDL extensions (1)
- SIMERO (1)
- SPARQL (1)
- SPARQL query learning (1)
- Safety Analysis (1)
- Scalar (1)
- Scheduling (1)
- Schema <Informatik> (1)
- Schematisation (1)
- Schematisierung (1)
- Scientific Community Analysis (1)
- Scientific Computing (1)
- Self-supervised Learning (1)
- Semantic Index (1)
- Semantic Wikis (1)
- Semantische Modellierung (1)
- Semantisches Datenmodell (1)
- Sensing (1)
- Sensor (1)
- Sensor Fusion (1)
- Sensors (1)
- Service Access Points (1)
- Shared Resource Modeling (1)
- Sicherheitsanalyse (1)
- Similarity Join (1)
- Similarity Joins (1)
- Simulationen (1)
- Skalar (1)
- Smart City (1)
- Smart Device (1)
- Smart Mobile Device (1)
- Smart Textile (1)
- Smartphone (1)
- Smartwatch (1)
- Socio-Semantic Web (1)
- Software (1)
- Software Comprehension (1)
- Software Dependencies (1)
- Software Engineering (1)
- Software Evolution (1)
- Software Maintenance (1)
- Software Measurement (1)
- Software Testing (1)
- Software Visualization (1)
- Software engineering (1)
- Software transactional memory (1)
- Software-Architektur (1)
- Softwaremetrie (1)
- Softwarespezifikation (1)
- Softwarewartung (1)
- Sprachdefinition (1)
- Sprache (1)
- Sprachprofile (1)
- Statistical Independence (1)
- Stimmungsbasierte Musikempfehlungen (1)
- Stochastic Dependence (1)
- Stochastic Network Calculus (1)
- Stokes Equations (1)
- Streaming (1)
- Strukturiertes Gitter (1)
- Strömung (1)
- Surface Reconstruction (1)
- Symbolic Methods (1)
- Synchronous Control Asynchronous Dataflow (1)
- SystemC (1)
- Systemarchitektur (1)
- Systemdesign (1)
- Task-based (1)
- Temporal Decoupling (1)
- Temporal Logic (1)
- Temporal data processing (1)
- Tensor (1)
- Tensorfeld (1)
- Tesselation (1)
- Tetraeder (1)
- Tetraedergi (1)
- Tetrahedral Grid (1)
- Tetrahedral Mesh (1)
- Textual CBR (1)
- Themenbasierte Empfehlungen von Ressourcen (1)
- Time Series (1)
- Time series classification (1)
- Time-motion-Ultraschallkardiographie (1)
- Time-slotted (1)
- Topic-based Resource Recommendations (1)
- Topologie (1)
- Topology visualization (1)
- Transaction Level Modeling (TLM) (1)
- Transport Protocol (1)
- Traversability Analysis (1)
- UML Activity (1)
- Ubiquitous system (1)
- Ultraschallkardiographie (1)
- Umweltinformatik (1)
- Uncertainty Estimation (1)
- Unobtrusive instrumentations (1)
- Unorganized Data (1)
- Unstrukturiertes Gitter (1)
- Urban sprawl (1)
- UrbanSim (1)
- User Model (1)
- VIACOBI (1)
- Validierung (1)
- Vector (1)
- Vector Field (1)
- Vektor (1)
- Vektorfelder (1)
- Verifikation (1)
- Verzerrungstensor (1)
- Virtual Prototyping (1)
- Visual Queries (1)
- Visualization Theory (1)
- Volume rendering (1)
- Volumen-Rendering (1)
- Voronoi diagram (1)
- Weak Memory Model (1)
- Wearable Computing (1)
- Wi-Fi (1)
- Wide-column stores (1)
- Wissenschaftliches Rechnen (1)
- XDBMS (1)
- XEC (1)
- XML (1)
- XML query estimation (1)
- XML summary (1)
- Yaroslavskiy-Bentley-Bloch Quicksort (1)
- Zielverfolgung (1)
- acoustic modeling (1)
- affective user interface (1)
- affine arithmetic (1)
- analysis of algorithms (1)
- artificial neural network (1)
- associations (1)
- autoencoder (1)
- automated theorem proving (1)
- automotive (1)
- average-case analysis (1)
- behaviour-based system (1)
- binary countdown protocol (1)
- black bursts (1)
- body-IMU calibration (1)
- classification (1)
- climate (1)
- clustering (1)
- collaborative information visualization (1)
- collaborative mobile sensing (1)
- collective intelligence (1)
- computational biology (1)
- computational fluid dynamics (1)
- computer-supported cooperative work (1)
- concurrent (1)
- constraint exploration (1)
- content-and-structure summary (1)
- continuous master theorem (1)
- crowd condition estimation (1)
- crowd density estimation (1)
- crowd scanning (1)
- crowd sensing (1)
- crowdsourcing (1)
- curves and surfaces (1)
- data race (1)
- data sets (1)
- data-flow (1)
- dataset (1)
- decidability (1)
- dependable systems (1)
- design (1)
- deterministic arbitration (1)
- directed graphs (1)
- distributed (1)
- distributed real-time systems (1)
- distributed tasks (1)
- embedded (1)
- embedding (1)
- emotion visualization (1)
- end-to-end learning (1)
- ensemble (1)
- environmental noise (1)
- evolutionary algorithm (1)
- explainability (1)
- firewall (1)
- flow visualization (1)
- foundational translation validation (1)
- gaussian filter (1)
- geographic information systems (1)
- geology (1)
- graph drawing algorithm (1)
- graph embedding (1)
- graph layout (1)
- hand pose, hand shape, depth image, convolutional neural networks (1)
- historical documents (1)
- human body motion tracking (1)
- hypergraph (1)
- iB2C (1)
- image analysis (1)
- implementation (1)
- inertial sensors (1)
- information systems (1)
- interference (1)
- interpolation (1)
- interval arithmetic (1)
- invariant (1)
- inverses Pendel (1)
- language definition (1)
- language modeling (1)
- language profiles (1)
- linked data (1)
- long short-term memory (1)
- long tail (1)
- machine-checkable proof (1)
- magnetic field based localization (1)
- magnetometer calibration (1)
- matrix visualization (1)
- message-passing (1)
- metadata (1)
- microelectronics ontology (1)
- mixed-signal (1)
- modularisation (1)
- moment (1)
- multicore (1)
- multidimensional datasets (1)
- multinomial regression (1)
- multithreading (1)
- multitype code coupling (1)
- multiway partitioning (1)
- nestable tangibles (1)
- optimization correctness (1)
- oscillating magnetic fields (1)
- out-of-order (1)
- parallel (1)
- participatory sensing (1)
- path cost models (1)
- pattern recognition (1)
- pivot sampling (1)
- point cloud (1)
- proof generating optimizer (1)
- quality assurance (1)
- ray casting (1)
- ray tracing (1)
- real-time tasks (1)
- reasoning (1)
- relaxed memory models (1)
- risk management (1)
- robustness (1)
- scalar field (1)
- scene flow (1)
- secondary structure prediction (1)
- sensor fusion (1)
- small-multiples node-link visualization (1)
- social media (1)
- software comprehension (1)
- software engineering (1)
- software engineering task (1)
- sparse-to-dense (1)
- spatial statistics (1)
- static software structure (1)
- stationary sensing (1)
- statistics (1)
- structural summary (1)
- subjective evaluation (1)
- symbolic simulation (1)
- synchronous (1)
- system architecture (1)
- tabletop (1)
- task sequence (1)
- tensor (1)
- tensorfield (1)
- terrain rendering (1)
- time-dependent (1)
- time-varying flow fields (1)
- touch surfaces (1)
- translation contract (1)
- ultrasound signals (1)
- urban planning (1)
- user-centered design (1)
- vector field visualization (1)
- vectorfield (1)
- virtual reality (1)
- visual analytics (1)
- visual structure (1)
- wearable systems (1)
- weighted finite-state transducers (1)
- wireless signal (1)
- xai (1)
- zeitabhängige Strömungen (1)
- Ähnlichkeit (1)
- Übersetzung (1)
Faculty / Organisational entity
Weak memory consistency models capture the outcomes of concurrent
programs that appear in practice and yet cannot be explained by thread
interleavings. Such outcomes pose two major challenges to formal
methods. First, establishing that a memory model satisfies its
intended properties (e.g., supports a certain compilation scheme) is
extremely error-prone: most proposed language models were initially
broken and required multiple iterations to achieve soundness. Second,
weak memory models make verification of concurrent programs much
harder, as a result of which there are no scalable verification
techniques beyond a few that target very simple models.
This thesis presents solutions to both of these problems.
First, it shows that the relevant metatheory of weak memory
models can be effectively decided (sparing years of manual proof
efforts), and presents Kater, a tool that can answer metatheoretic
queries in a matter of seconds. Second, it presents GenMC, the first
(and only) scalable stateless model checker that is parametric in the
choice of the memory model, often improving the prior state of the art
by orders of magnitude.
In one-dimensional (1-D) Ultrasound (US) measurements, signals are
acquired that form the basis of more sophisticated two-dimensional (2-D) or
three-dimensional (3-D) US imaging. These 1-D signals contain a lot of raw
information about the US wave propagation and interaction with the
medium that is only processed in parts during image generation. While
image representations are easy to interpret for humans, the analysis of US
wave signals is hard to perform without applying algorithms to extract
desired features.
This work investigates reliable and fast 1-D US signal classifications to
distinguish between different stages or states in biomedical US scenarios and
shows how the new field of Machine Learning (ML) on raw US wave data
provides advantages and different applications. To achieve good results, the
input signals are treated as time series, which requires the deployment of
comparatively complex Time Series Classification (TSC) algorithms.
The literature shows that a lot of research efforts have previously only
tackled the classification and segmentation of US Brightness mode (B-Mode)
images, while neglecting approaches to classify 1-D signals to a large extent.
This research contributes by developing, deploying and evaluating
classification approaches for three distinct biomedical US classification tasks
and finds that respective signal classifications for different scenarios are
possible with varying degrees of accuracies. It entails the comparison of
several combinations of data types (e.g. temporal, spectral and statistical
features or raw signals), ML models and pre-processing steps to provide a
strong foundation for robust, binary classifications of 1-D US signals for
scenarios based on low-cost wearable, mobile and stationary devices. This
research addresses scientific questions not answered before by informing on
detailed descriptions of beneficial domain specific knowledge (domain specific
knowledge (DSK)), achieved accuracies and times needed for training and
evaluation of the examined ML models.
The resulting ML pipelines includes solutions based on data acquired from
custom experimental setups or clinical trials. Possible real-world applications
might include muscle contraction trackers, muscle fatigue detectors,
epiphyseal radius bone closure detectors or devices providing information
about advanced liver disease stages.
Automated machine-assisted
classifications requiring as little DSK as possible from the end user enable
application scenarios ranging from fitness or rehabilitation trackers as
consumer devices to solutions providing diagnostic support without requiring
extensive knowledge from professional medical practitioners. For example,
decision support systems for bone age assessments in clinical use or liver
health assessment systems for gastroenterologists.
This work shows that reliable, robust and fast classifications based on 1-D
US signals are possible with high degrees of accuracies depending on the
examined scenario with achieved F 1 -scores ranging from ≈ 70% to ≈ 87%.
These results prove that real-life applications for recreational purposes are
already possible and that critical applications for clinical use are highly likely
to be achieved once the presented approaches are further optimized in the future.
The field of 3D reconstruction is one of the most important areas in computer
vision. It is not only of theoretical importance, but it is also increasingly
used in practical applications, be it in reverse engineering, quality control or
robotics. In practical applications, where high precision reconstructions are
required for a large variety of different objects, structured light reconstruction
is often the method of choice. It allows to achieve accurate and dense
point correspondences over the entire scene, regardless of object texture or
features. Techniques that project phase-shifted sinusoidals are widely used
because, based on the harmonic addition theorem, they theoretically allow
surface encoding in full camera resolution invariant to the object’s coloring.
In this thesis, a fully-automatic reconstruction pipeline based on the sinusoidal
structured light technique is presented. From the projection of the
fringe patterns for encoding the object’s surface, the robust matching of the
point correspondences in sub-pixel accuracy, the auto-calibration of the setup
including the active device, up to the fully-automatic alignment of the partial
reconstructions, all steps will be described and examined in detail. During
that, improvements will be achieved in the area of matching, obtaining highly
accurate and topologically consistent correspondences in sub-pixel precision
between all the devices used. Furthermore, the auto-calibration from point
correspondences, based on the epipolar geometry of the structured light system
is improved. Weaknesses of previous methods in the extraction of focal
lengths from the fundamental matrices are discovered and addressed. The partial
point clouds, reconstructed from the auto-calibrated devices, are finally
pre-aligned using a neural network approach, based on light-resistant optical
flow estimation and subsequently refined using a global approach.
The weaknesses of the structured light method itself will also be addressed
and partially fixed during the course of this work. Since it is an active reconstruction method, certain surface properties can affect the quality of the
reconstruction. It will be shown how these problems can be eliminated or at
least be reduced using an iterative approach that combines fringe patterns with
an inverse texture. Another weakness of the method is its time-consuming acquisition procedure. Typically, a large number of horizontal and vertical fringe
patterns are projected onto the scene to achieve high-precision encoding despite
the limited dynamic range and resolution of the projector. Therefore, a
method will be presented which allows to combine the horizontal and vertical
patterns for a simultaneous two dimensional surface encoding.
During our daily lives, we are confronted with vast amounts of data, the processing of which can dramatically influence our lives, both positively and negatively. The enormous amount of data (images, texts, tables, and time series), its variety and possible applications are not always obvious. Due to advancements in the internet of things (IoT), there exist billions of sensors that produce time series which can be found everywhere, whether in medicine, the financial sector or the agricultural economy. This incredible amount of time series data has many hidden features which are useful for industry as well as for daily use, e.g. improving the cancer prediction can save real human lives. Recently, several deep learning methods have been proposed for analyzing this time series data. However, due to their black box nature, their applicability is limited in critical sectors like medicine, finance, and communication. In addition, it is now a compulsion as per artificial intelligence (AI) Act and per General Data Protection Regulation (GDPR) to protect any sensitive data and provide explanations in safety-critical domains. To enable use of DNNs in a broader domain scope, this thesis presents a framework for privacy-preserved and interpretable time series analysis. TimeFrame consists of four main components, namely, post-hoc interpretability, intrinsic interpretability, direct privacy, and indirect privacy. Interpretability is indispensable to avoid damaging people or the infrastructure. In the past years, the development mostly focused on image data, which prevented the full potential of DNNs in time series processing from being exploited. To overcome this limitation, TimeFrame introduces five (Time to Focus, TSViz, TimeREISE, TSInsight, Data Lens) novel post-hoc and two (PatchX, P2ExNet) novel intrinsic interpretability components. TimeFrame addresses multiple perspectives such as attribution, compression, visualization, influence, prototyping, and hierarchical splitting. Compared to existing methods, the components show better explanations, robustness, and scalability. Another crucial factor is the privacy when dealing with sensitive data and deep learning. In this context, TimeFrame introduces two (PPML, PPML x XAI) components for direct and one (From Private to Public) component for indirect privacy. These components benchmark privacy approaches, their effect on interpretability, and the synthetic generation of data to overcome privacy concerns. TimeFrame offers a large set of interpretability and privacy components that can be combined and consider numerous different aspects. Furthermore, the novel approaches have shown to consistently outperform twenty existing state-of-the-art methods across up to 20 different datasets. To guarantee the fairness, various metrics were used including performance change, Sensitivity, Infidelity, Continuity, runtime, model dependency, compression rate, and others. This broad set of metrics makes it possible to provide guidelines for a more appropriate use of existing state-of-the-art approaches as well as the novel components included in TimeFrame.
Highly Automated Driving (HAD) vehicles represent complex and safety critical systems. They are deployed in an open context i.e., an intricate environment which undergoes continual changes. The complexity of these systems and insufficiencies in sensing and understanding the open context may result in unsafe and uncertain behaviour. The safety critical nature of the HAD vehicles requires modelling of root causes for unsafe behaviour and their mitigation to argue sufficient reduction of residual risk.
Standardization activities such as ISO 21448 provide guidelines on the Safety Of The Intended Functionality (SOTIF) and focus on the analysis of performance limitations under the influence of triggering conditions that can lead to hazardous behaviour. SOTIF references traditional safety analyses methods e.g., Failure Mode and Effect Analysis (FMEA) and Fault Tree Analysis (FTA) to perform safety analysis. These analyses methods are based on certain assumptions e.g., single point failure in FMEA and independence of basic events in FTA. Moreover, these analyses are generally based on expert knowledge i.e., data-based models or hybrid approaches (expert and data) are seldom practised. The resulting safety model is fixed i.e., it is generally seen as a one-time artefact. Open context environment may contain triggering conditions which may not be evident to the expert. Open context also evolves over time and new phenomena may emerge.
This thesis explores the applicability of the traditional safety analyses techniques to provide safety models for HAD vehicles operating in the open context, under the light of modelling assumptions taken by traditional safety analyses techniques. Moreover, incorporating uncertainties into safety analyses models is also explored. An explicit distinction between the inherent uncertainty of a probabilistic event (aleatory) and uncertainty due to lack of knowledge (epistemic) is made to formalize models to perform SOTIF analysis. A further distinction is made for conditions of complete ignorance and termed as ontological uncertainty. The distinction is important as for HAD vehicles operating in open context the ontological uncertainty can never be completely disregarded.
This thesis proposes a novel framework of SOTIF to model, estimate and dis cover triggering conditions relevant to performance limitations. The framework provides the ability to model uncertainties while also providing a hybrid approach i.e., supporting inclusion of expert knowledge as well as data driven engineering processes. Two representative algorithms are provided to support the framework. Bayesian Network (BN) and p-value hypothesis testing are utilised in this regard. The framework is implemented on a real-world case study in which LIDARs based perception systems are used as vehicle detection system.
This doctoral dissertation is comprised of nine published articles covering different
methods for ‘Fast, Robust Rigid and Non-Rigid Registration for Globally Consistent
3D Scene and Shape Reconstruction’. Overall the contributing articles are separated
and discussed in three stages – The first part of the thesis i.e., chapter 2 explains
three novel method classes of rigid point set registration namely Gravitational Approach (GA), Fast Gravitational Approach (FGA), and RPSRNet. GA was introduced as the first physics-based rigid point set registration. It includes elegant modeling of rigid by dynamics using Newtonian mechanics. The method proposed many new avenues for other types of pattern matching tasks thank point set registration. Next, FGA method, published 4 years after GA presented as an extension that breaks the algorithmic complexity of GA from O(M N ) to O(M log N ) using Barnes-Hut tree representation of point cloud. It also eliminates the requirement of heuristic optimization parameter settings by GA, and achieve state-of-the-art alignment accuracy on LiDAR odometry. Finally, RPSRNet presents deep learning version of FGA, with custom convolution layers for hierarchical point feature embedding. RPSRNet is robust and the fastest among SoA methods for LiDAR data registration. The second part, i.e., chapter 3, of the thesis introduces NRGA as the fist physics-based non-rigid point set
registration method which is computationally slow but robust against noisy and partial inputs. NRGA preserves structural consistency as it coherently regularize motion of deformable vertices. For articulated hand shape reconstruction, a tailored version of NRGA -- Articulated-NRGA -- is effective to refine final hand shape. Collision and penetration avoidance between source and target surfaces are tackled by constrained optimization in NRGA. This setting has improved hand and object interaction reconstruction. Next contribution FoldMatch method remodels the shape deformation by introducing wrinkle vector field (WVF) for capturing complex clothing and garment details while fitting body models onto 3D Scans. Quantitative evaluation of FoldMatch and NRGA shows their effectiveness in geometrically consistent surface modeling and reconstruction tasks. Finally, the third part of the thesis explains globally consistent outdoor scene reconstruciton, odometry estimation, and uncertainty guided pose-graph optimization in a novel LiDAR-based localization and map building method, called Deep Evidential LiDAR Odometry (DELO). This is the first Odometry method to use predictive uncertainty modeling for sensor pose prediction network.
From industrial fault detection to medical image analysis or financial fraud prevention: Anomaly detection—the task of identifying data points that show significant deviations from the majority of data—is critical in industrial and technological applications. For efficient and effective anomaly detection, a rich set of semantic features are required to be automatically extracted from the complex data. For example, many recent advances in image anomaly detection are based on self-supervised learning, which learns rich features from a large amount of unlabeled complex image data by exploiting data augmentations. For image data, predefined transformations such as rotations are used to generate varying views of the data. Unfortunately, for data other than images, such as time series, tabular data, graphs, or text, it is unclear what are suitable transformations. This becomes an obstacle to successful self-supervised anomaly detection on other data types.
This thesis proposes Neural Transformation Learning, a self-supervised anomaly detection method that is applicable to general data types. In contrast to previous methods relying on hand-crafted transformations, neural transformation learning learns the transformations from data and uses them for detection. The key ingredient is a novel objective that encourages learning diverse transformations while preserving the relevant semantic content of the data. We prove theoretically and empirically that it is more suited than existing objectives for transformation learning.
We also introduce the extensions of neural transformation learning for anomaly detection within time series and graph-level anomaly detection. The extensions combine transformation learning and other learning paradigms to incorporate vital prior knowledge about time series and graph data. Moreover, we propose a general training strategy for deep anomaly detection with contaminated data. The idea is to infer the unlabeled anomalies and utilize them for updating parameters alternatively. In setups where expert feedback is available, we present a diverse querying strategy based on the seeding algorithm of K-means++ for active anomaly detection.
Our extensive experiments and analysis demonstrate that neural transformation learning achieves remarkable and robust anomaly detection performance on various data types. Finally, we outline specific paths for future research.
Semi-structured data is a common data format in many domains.
It is characterized by a hierarchical structure and a schema that is not fixed.
Efficient and scalable processing of this data is therefore challenging, as many existing indexing and processing techniques are not well-suited for this data format.
This dissertation presents a novel approach to processing large JSON datasets.
We describe a new data processor, JODA, that is designed to process semi-structured data by using all available computing resources and state-of-the-art techniques.
Using a custom query language and a vertically-scaling pipeline query execution engine, JODA can process large datasets with high throughput.
We optimize JODA by using a novel optimization for iterative query workloads called delta trees, which succinctly represent the changes between two documents.
This allows us to process iterative and exploratory queries efficiently.
We improve the filtering performance of JODA by implementing a holistic adaptive indexing approach that creates and improves structural and content indices on the fly, depending on the query load.
No prior knowledge about the data is required, and the indices are automatically improved over time.
JODA is also modularized and can be extended with new user-defined predicates, functions, indices, import, and export functionalities.
These modules can be written in an external programming language and integrated into the query execution pipeline at runtime.
To evaluate this system against competitors, we introduce a benchmark generator, coined BETZE, which aims to simulate data scientists exploring unknown JSON datasets.
The generator can be tweaked to generate query workload with different characteristics, or predefined presets can be used to quickly generate a benchmark.
We see that JODA outperforms competitors in most tasks over a wide range of datasets and use-cases.
3D joint angles based human pose is needed for applications like activity recognition, musculoskeletal health, sports biomechanics and ergonomics. The microelectromechanical systems (MEMS) based magnetic-inertial measurement units (MIMUs) can estimate 3D orientation. Due to small size, MIMUs can be attached to the body as wearable sensors for obtaining full 3D human pose and this system is termed as inertial motion capture (i-Mocap). But the MIMUs suffer from sensor errors and disturbances, due to which orientation estimated from individual MIMUs can be erroneous. Accurate sensor calibration is essential and subsequently alignment of these sensors to body segments must also be precisely known, which is called sensor-to-segment calibration. Sensor fusion is employed to address the disturbances and noise in MIMUs. Many state-of-art inertial motion capture approaches ignore the magnetometer and only use IMUs to reduce the error arising from inhomogeneous magnetic field. These algorithms rely on kinematic constraints and assumptions regarding joints and are based on IMUs located on the adjacent body segments. The full body coverage requires 13-17 such units and can be quite obtrusive. The setting up and calibration of so many wearable sensors also take time.
This thesis focuses on 3D human pose estimation from a reduced number of MIMUs and deals with this problem systematically. First we propose an accurate simultaneous calibration of multiple MIMUs, which also learns the uncertainty of individual sensors. We then describe a novel sensor fusion algorithm for robust orientation estimation from an MIMU and for updating sensors calibration online. The residual errors in both sensor calibration and fusion can result in drift error in the joint angles. Therefore, we present anatomical (sensor-to-segment) calibration in which an orientation offset correction term is updated and used for online correction of residual drift in individual joint angles. Subsequently we demonstrate that 3D human joint angle constraints can be learned using a data-driven approach in a high dimensional latent space. Owing to temporal and joint angle constraints, it is possible to use only a reduced set of sensors (as opposed to one sensor per segment) and still obtain 3D human pose. But the spatial and temporal prior learning from data is often limited due to finite set of movement patterns in most datasets. This introduces uncertainty while estimating 3D human pose from sparse MIMU sensors. We propose a magnetometer robust orientation parameterization and a data-driven deep learning framework to predict 3D human pose with associated uncertainty from sparse MIMUs. The model is evaluated on real MIMU data and we show that the uncertainty predicted by the trained model is well-correlated with actual error and ambiguity.
Though Computer Aided Design (CAD) and Simulation software are mature, well established, and in wide professional use, modern design and prototyping pipelines are challenging the limits of these tools. Advances in 3D printing have brought manufacturing capability to the general public. Moreover, advancements in Machine Learning and sensor technology are enabling enthusiasts and small companies to develop their own autonomous vehicles and machines. This means that many more users are designing (or customizing) 3D objects in CAD, and many are testing machine autonomy in Simulation. Though Graphical User Interfaces (GUIs) are the de-facto standard for these tools, we find that these interfaces are not robust and flexible. For example, designs made using GUI often break when customized, and setting up large simulations can be quite tedious in GUI. Though programmatic interfaces do not suffer from these limitations, they are generally quite difficult to use, and often do not provide appropriate abstractions and language constructs.
In this Thesis, we present our work on bridging the ease of use of GUI with the robustness and flexibility of programming. For CAD, we propose an interactive framework that automatically synthesizes robust programs from GUI-based design operations. Additionally, we apply program analysis to ensure customizations do not lead to invalid objects. Finally, for simulation, we propose a novel programmatic framework that simplifies building of complex test environments, and a test generation mechanism that guarantees good coverage over test parameters. Our contributions help bring some of the advantages of programming to traditionally GUI-dominant workflows. Through novel programmatic interfaces, and without sacrificing ease of use, we show that the design and customization of 3D objects can be made more robust, and that the creation of parameterized simulations can be simplified.
Faces deliver invaluable information about people. Machine-based perception can be of a great benefit in extracting that underlying information in face images if the problem is properly modeled. Classical image processing algorithms may fail to handle the diverse data available today due to several challenges related to varying capturing locations, and conditions. Advanced machine learning methods and algorithms are now highly beneficial due to the rapid development of powerful hardware, enabling feasible advanced solutions based on data learning and summarization into powerful models. In this thesis, novel solutions are provided to the problems of head orientation estimation and gender prediction. Initially, classical machine learning algorithms were used to address head orientation estimation but were limited by their inability to handle large datasets and poor generalization. To overcome these challenges, a new highly accurate head pose dataset was acquired to tackle the identified problems. Novel trained deep neural networks have been exploited, that use the acquired data and provide novel architectures. The information about head pose is then represented in the network weights, thus, allowing predicting the head orientation angles given a new unseen face. The acquired dataset, named AutoPOSE opens the door for further studies in the field of computer vision and especially, face analysis. The problem of gender prediction has also been explored, but unlike humans who can easily identify gender from a face, computers face difficulties due to facial similarities. Therefore, hand-crafted features are not effective for generalization. To address this, a new deep learning method was developed and evaluated on multiple public datasets, with identified challenges in both still images and videos addressed. Finally, the effect of facial appearance changes due to head orientation variation has been investigated on gender prediction accuracy. A novel orientation-guided feature maps recalibration method is presented, that significantly increased the accuracy of gender prediction.
In conclusion, two problems have been addressed in this thesis, independently and joined together. Existing methods have been enhanced with intelligent pre-processing methods and new approaches have been introduced to tackle existing challenges, that arise from pose, illumination, and occlusion variations. The proposed methods have been extensively evaluated, showing that head orientation and gender prediction can be estimated with high accuracy using machine learning-based methods. Also, the evaluations showed that the use of head orientation information consistently improved the gender prediction accuracy. Scientific contributions have been presented, and the new acquired highly accurate dataset motivates the research community to push the state-of-the-art forward.
Undocumented enterprise data can easily pile up in companies in form of datasets and personal information. In absence of a data management strategy, such data becomes rather messy and may not fit for its intended use. Since there is often no documentation available, only a limited number of domain experts are aware of its contents. Therefore, for companies it becomes increasingly difficult to use such data to its full potential. To provide a solution, this PhD thesis investigates the construction of enterprise and personal knowledge graphs by semantically enriching messy data with meaning using semantic technologies. Since real world entities and their interrelations are organized in a graph, knowledge graphs serve as a semantic bridge between domain conceptualization and raw data. Spreadsheets are a prominent example of such enterprise data, since they are widely used by knowledge workers in the industrial sector. Two distinct approaches are investigated to construct knowledge graphs from them: a global extraction & annotation method and a local mapping technique. The latter is further complemented with a predictor of mapping rules on messy data. Different human-in-the-loop strategies are considered to include experts depending on their user group. Since non-technical users usually lack understanding of semantic technologies, they need appropriate tools to be able to give feedback. In case of developers, approaches are proposed to close the technology gap between industry and Semantic Web related concepts. Semantic Web practitioners participate with ontology modeling and linked data applications. Enterprise and personal data is typically confidential which is why it cannot be shared with a research community to discuss its challenges. However, for evaluation and reproducibility reasons publicly available datasets are mandatory. The thesis proposes ways to generate synthetic datasets with the goal to be as authentic as possible. Besides that, for internal evaluations a crawler of personal data on desktops is implemented. There are further contributions related to this thesis in diverse domains. One is about the motivation to support users in their daily work using personal knowledge assistants. Others are the agricultural field and the data science domain which also benefit from knowledge graph approaches. In conclusion, this PhD thesis contributes to the construction of knowledge graphs from especially messy enterprise data, while users from different groups take part in this process in various ways.
This thesis focuses on novel methods to establish the utility of wearable devices along with machine learning and pattern recognition methods for formal education and address the open research questions posed by existing methods. Firstly, state-of-the-art methods are proposed to analyse the cognitive activities in the learning process, i.e., reading, writing, and their correlation. Furthermore, this thesis presents real-time applications in wearable space as an experimental tool in Physics education, and an air-writing system.
There are two critical components in analysing the reading behaviour, i.e., WHERE a person looks at (gaze analysis) and WHAT a person looks at (content analysis). This thesis proposes novel methods to classify the reading content to address the WHAT AT component. The proposed methods are based on a hybrid approach, which fuses the traditional computer vision methods with deep neural networks. These methods, when evaluated on publicly available datasets, yield state-of-the-art results to define the structure of the document images. Moreover, extensive efforts were made to refine and correct ICDAR2017-POD dataset along with a completely new FFD dataset.
Traditionally, handwriting research focuses on character and number recognition without looking into the type of writing, i.e. text, math, and drawing. This thesis reports multiple contributions for on-line handwriting classification. First, it presents a public dataset for on-line handwriting classification OnTabWriter, collected using iPen and an iPad. In addition, a new feature set is introduced for on-line handwriting classification to establish the benchmark on the proposed dataset to classify handwriting as plain text, mathematical expression, and plot/graph. An ablation study is made to evaluate the performance of the proposed feature set in comparison to existing feature sets. Lastly, this thesis evaluates the importance of context for on-line handwriting classification.
Analysing reading and writing activities individually is not enough to provide insights to identify the student's expertise unless their correlations are analysed. This thesis presents a study where reading data from wearable eye-trackers and writing data from sensor pen are analysed together in correlation to correlate the expertise of the users in Physics education with their actual knowledge. Initial results show a strong correlation between individual's expertise and understanding of the subject.
Augmented reality & virtual applications can play a vital role in making classroom environments more interactive and engaging both for teachers and learners. To validate the hypothesis, different applications are developed and evaluated. First, smart glasses are used as an experimental tool in Physics education to help the learners perform experiments by providing assistance and feedback on head mounted display in understanding acoustics concepts. Second, a real-time application of air-writing with the finger on an imaginary canvas using a single IMU as the FAirWrite system is also presented. FAirWrite system is further equipped with DL methods to classify the air-written characters.
Due to its performance, the field of deep learning has gained a lot of attention, with neural networks succeeding in areas like \( \textit{Computer Vision} \) (CV), \( \textit{Neural Language Processing} \) (NLP), and \( \textit{Reinforcement Learning} \) (RL). However, high accuracy comes at a computational cost as larger networks require longer training time and no longer fit onto a single GPU. To reduce training costs, researchers are looking into the dynamics of different optimizers, in order to find ways to make training more efficient. Resource requirements can be limited by reducing model size during training or designing more efficient models that improve accuracy without increasing network size.
This thesis combines eigenvalue computation and high-dimensional loss surface visualization to study different optimizers and deep neural network models. Eigenvectors of different eigenvalues are computed, and the loss landscape and optimizer trajectory are projected onto the plane spanned by those eigenvectors. A new parallelization method for the stochastic Lanczos method is introduced, resulting in faster computation and thus enabling high-resolution videos of the trajectory and second-order information during neural network training. Additionally, the thesis presents the loss landscape between two minima along with the eigenvalue density spectrum at intermediate points for the first time.
Secondly, this thesis presents a regularization method for \( \textit{Generative Adversarial Networks} \) (GANs) that uses second-order information. The gradient during training is modified by subtracting the eigenvector direction of the biggest eigenvalue, preventing the network from falling into the steepest minima and avoiding mode collapse. The thesis also shows the full eigenvalue density spectra of GANs during training.
Thirdly, this thesis introduces ProxSGD, a proximal algorithm for neural network training that guarantees convergence to a stationary point and unifies multiple popular optimizers. Proximal gradients are used to find a closed-form solution to the problem of training neural networks with smooth and non-smooth regularizations, resulting in better sparsity and more efficient optimization. Experiments show that ProxSGD can find sparser networks while reaching the same accuracy as popular optimizers.
Lastly, this thesis unifies sparsity and \( \textit{neural architecture search} \) (NAS) through the framework of group sparsity. Group sparsity is achieved through \( \ell_{2,1} \)-regularization during training, allowing for filter and operation pruning to reduce model size with minimal sacrifice in accuracy. By grouping multiple operations together, group sparsity can be used for NAS as well. This approach is shown to be more robust while still achieving competitive accuracies compared to state-of-the-art methods.
In recent years, the formal methods community has made significant progress towards the development of industrial-strength static analysis tools that can check properties of real-world production code. Such tools can help developers detect potential bugs and security vulnerabilities in critical software before deployment. While the potential benefits of static analysis tools are clear, their usability and effectiveness in mainstream software development workflows often comes into question and can prevent software developers from using these tools to their full potential. In this dissertation, we focus on two major challenges that can limit their ability to be incorporated into software development workflows.
The first challenge is unintentional unsoundness. Static program analyzers are complicated tools, implementing sophisticated algorithms and performance heuristics. This makes them highly susceptible to undetected unintentional soundness issues. These issues in program analyzers can cause false negatives and have disastrous consequences e.g., when analyzing safety critical software. In this dissertation, we present novel techniques to detect unintentional unsoundness bugs in two foundational program analysis tools namely SMT solvers and Datalog engines. These tools are used extensively by the formal methods community, for instance, in software verification, systematic testing, and program synthesis. We implemented these techniques as easy-to-use open source tools that are publicly available on Github. With the proposed techniques, we were able to detect more than 55 unique and confirmed critical soundness bugs in popular and widely used SMT solvers and Datalog engines in only a few months of testing.
The second challenge is finding the right balance between soundness, precision, and perfor- mance. In an ideal world, a static analyzer should be as precise as possible while maintaining soundness and being sufficiently fast. However, to overcome undecidability issues, these tools have to employ a variety of techniques to be practical for example, compromising on the sound- ness of the analysis or approximating code behavior. Static analyzers therefore are not trivial to integrate into any usage scenario with different program sizes, resource constraints and SLAs. Most of the times, these tools also don’t scale to large industrial code bases containing millions of lines of code. This makes it extremely challenging to get the most out of these analyzers and integrate them into everyday development activities, especially for average software develop- ment teams with little to no knowledge or understanding of advanced static analysis techniques. In this dissertation we present an approach to automatically tailor an abstract interpreter to the code under analysis and any given resource constraints. We implemented our technique as an open source framework, which is publicly available on Github. The second contribution of this dissertation in this challenge area is a technique to horizontally scale analysis tools in cloud-based static analysis platforms by splitting the input to the analyzer into partitions and analyzing the partitions independently. The technique was developed in collaboration with Amazon Web Services and is now being used in production in their CodeGuru service.
The rising demand for machine learning (ML) models has become a growing concern for stakeholders who depend on automatic decisions. In today's world, black-box solutions (in particular deep neural networks) are being continuously implemented for more and more high-stake scenarios like medical diagnosis or autonomous vehicles. Unfortunately, when these opaque models make predictions that do not align with our expectations, finding a valid justification is simply not possible.
Explainable Artificial Intelligence (XAI) has emerged in response to our need for finding reasons that justify what a machine sees, but we don't. However, contributions in this field are mostly centered around local structures such as individual neurons or single input samples. Global characteristics that govern the behavior of a model are still poorly understood or have not been explored yet. An aggravating factor is the lack of a standard terminology to contextualize and compare contributions in this field. Such lack of consensus is depriving the ML community from ultimately moving away from black-boxes, and start creating systematic methods to design models that are interpretable by design.
So, what are the global patterns that govern the behavior of modern neural networks, and what can we do to make these models more interpretable from the start?
This thesis delves into both issues, unveiling patterns about existing models, and establishing strategies that lead to more interpretable architectures. These include biases coming from imbalanced datasets, quantification of model capacity, and robustness against adversarial attacks. When looking for new models that are interpretable by design, this work proposes a strategy to add more structure to neural networks, based on auxiliary tasks that are semantically related to the main objective. This strategy is the result of applying a novel theoretical framework proposed as part of this work. The XAI framework is meant to contextualize and compare contributions in XAI by providing actionable definitions for terms like "explanation" and "interpretation."
Altogether, these contributions address dire demands for understanding more about the global behavior of modern deep neural networks. More importantly, they can be used as a blueprint for designing novel, and more interpretable architectures. By tackling issues from the present and the future of XAI, results from this work are a firm step towards more interpretable models for computer vision.
Processing data streams is a classical and ubiquitous problem.
A query is registered against a potentially endless data stream and continuously delivers results as tuples stream in.
Modern stream processing systems allow users to express queries in different ways.
However, when a query involves joins between multiple input streams, the order of these joins is not transparently optimized.
In this thesis, we explore ways to optimize multi-way theta joins, where the join predicates are not limited to equality and multiple inputs are referenced.
We put forward a novel operator, MultiStream, which joins multiple input streams using iterative probing and bringing minimal materialization effort in.
The order in which tuples are sent inside a MultiStream operator is optimized using a cost-based model.
Further, a query can be answered using an multi-way tree comprising multiple MultiStream operators where each inner operator represents a materialized intermediate result.
We integrate equi-joins in MultiStream to reduce communication, such that mixed queries of theta and equality predicates are supported.
Streaming queries are long-standing and thus multiple queries might be registered at the system at the same time.
Hence, we research joint answering of multiple multi-way join queries and optimize the global ordering using integer linear programming.
All these approaches are implemented in CLASH, a system for generating Apache Storm topologies including runtime components that enables users to pose queries in a declarative way and let the system craft the suitable topology.
An Efficient Automated Machine Learning Framework for Genomics and Proteomics Sequence Analysis
(2023)
Genomics and Proteomics sequence analyses are the scientific studies of understanding the language of Deoxyribonucleic Acid (DNA), Ribonucleic Acid (RNA) and protein biomolecules with an objective of controlling the production of proteins and understanding their core functionalities. It helps to detect chronic diseases in early stages, root causes of clinical changes, key genetic targets for pharmaceutical development and optimization of therapeutics for various age groups. Most Genomics and Proteomics sequence analysis work is performed using typical wet lab experimental approaches that make use of different genetic diagnostic technologies. However, these approaches are costly, time consuming, skill and labor intensive. Hence, these approaches slow down the process of developing an efficient and economical sequence analysis landscape essential to demystify a variety of cellular processes and functioning of biomolecules in living organisms. To empower manual wet lab experiment driven research, many machine learning based approaches have been developed in recent years. However, these approaches cannot be used in practical environment due to their limited performance. Considering the sensitive and inherently demanding nature of Genomics and Proteomics sequence
analysis which can have very far-reaching as well as serious repercussions on account of misdiagnosis, the main
objective of this research is to develop an efficient automated computational framework for Genomics and Proteomics sequence analysis using the predictive and prescriptive analytical powers of Artificial Intelligence (AI) to significantly improve healthcare operations.
The proposed framework is comprised of 3 main components namely sequence encoding, feature engineering and
discrete or continuous value predictor. The sequence encoding module is equipped with a variety of existing and newly developed sequence encoding algorithms that are capable of generating a rich statistical representation of DNA, RNA and protein raw sequences. The feature engineering module has diverse types of feature selection and dimensionality reduction approaches which can be used to generate the most effective feature space. Furthermore, the discrete and/or continuous value predictor module of the proposed framework contains a wide range of existing machine learning and newly developed deep learning regressors and classifiers. To evaluate the integrity and generalizability of the proposed framework, we have performed a large-scale experimentation over diverse types of Genomics and Proteomics sequence analysis tasks (i.e., DNA, RNA and proteins).
In Genomics analysis, Epigenetic modification detection is one of the key component. It helps clinical researchers and practitioners to distinguish normal cellular activities from malfunctioned ones, which can lead to diverse genetic disorders such as metabolic disorders, cancers, etc. To support this analysis, the proposed framework is used to solve the problem of DNA and Histone modification prediction where it has achieved state-of-the-art performance on 27 publicly available benchmark datasets of 17 different species with best accuracy of 97%. RNA sequence analysis is another vital component of Genomics sequence analysis where the identification of different coding and non-coding RNAs as well as their subcellular localization patterns help to demystify the functions of diverse RNAs, root causes of clinical changes, develop precision medicine and optimize therapeutics. To support this analysis, the proposed framework is utilized for non-coding RNA classification and multi-compartment RNA subcellular localization prediction. Where it achieved state-of-the-art performance on 10 publicly available benchmark datasets of Homo sapiens and Mus Musculus species with best accuracy of 98%.
Proteomics sequence analysis is essential to demystify the virus pathogenesis, host immunity responses, the way
proteins affect or are affected by the cell processes, their structure and core functionalities. To support this analysis, the proposed framework is used for host protein-protein and virus-host protein-protein interaction prediction. It has achieved state-of-the-art performance on 2 publicly available protein protein interaction datasets of Homo Sapiens and Mus Musculus species with best accuracy of 96% and 7 viral host protein protein interaction datasets of multiple hosts and viruses with best accuracy of 94%. Considering the performance and practical significance of proposed framework, we believe proposed framework will help researchers in developing cutting-edge practical applications for diverse Genomic and Proteomic sequence analyses tasks (i.e., DNA, RNA and proteins).
Scientific research plays a crucial role in the development of a society. With ever-increasing volumes of scientific publications are now making it extremely challenging to analyze and maintain insights into the scientific communities like collaboration or citation trends and evolution of interests etc. This thesis is an effort towards using scientific publications to provide detailed insights into a scientific community from a range of aspects. The contribution of this thesis is five-fold.
Firstly, this thesis proposes approaches for automatic information extraction from scientific publications. The proposed layout-based approach for this purpose is inspired by how human beings perceive individual references relying only on visual queues. The proposed approach significantly outperforms the existing text-based techniques and is independent of any domain or language.
Secondly, this thesis tackles the problem of identifying meaningful topics from a given publication as the keywords provided in the publication are not always accurate representatives of the publication topic. To rectify this problem, this thesis proposes a state-of-the-art keywords extraction approach that employs a domain ontology along with the detected keywords to perform topic modeling for a given set of publications.
Thirdly, this thesis analyses the disposition of each citation to understand its true essence. For this purpose, we proposes a transformer-based approach for analyzing the impact of each citation appearing in a scientific publication. The impact of a citation can be determined by the inherent sentiment and intent of a citation, which refers to the assessment and motive of an author towards citing a scientific publication.
Furthermore, this thesis quantifies the influence of a research contributor in a scientific community by introducing a new semantic index for researchers that takes both quantitative and qualitative aspects of a citation into account to better represent the prestige of a researcher in a scientific community. Semantic Index is also evaluated for conformity to the guidelines and recommendations of various research funding organizations to assess the impact of a researcher.
In this thesis, all of the aforementioned aspects are packaged together in a single framework called Academic Community Explorer (ACE) 2.0, which automatically extracts and analyzes information from scientific publications and visualizes the insights using several interactive visualizations. These visualizations provide an instant glimpse into the scientific communities from a wide range of aspects with different granularity levels.
The generally unsupervised nature of autoencoder models implies that the main training metric is formulated as the error between input images and their corresponding reconstructions. Different reconstruction loss variations and latent space regularization have been shown to improve model performances depending on the tasks to solve and to induce new desirable properties like disentanglement. Nevertheless, measuring the success in, or enforcing properties by, the input pixel space is a challenging endeavor. In this work, we want to make more efficient use of the available data and provide design choices to be considered in the recording or generation of future datasets to implicitly induce desirable properties during training. To this end, we propose a new sampling technique which matches semantically important parts of the image while randomizing the other parts, leading to salient feature extraction and a neglection of unimportant details. Further, we propose to recursively apply a previously trained autoencoder model, which can then be interpreted as a dynamical system with desirable properties for generalization and uncertainty estimation.
The proposed methods can be combined with any existing reconstruction loss. We give a detailed analysis of the resulting properties on various datasets and show improvements on several computer vision tasks: image and illumination normalization, invariances, synthetic to real generalization, uncertainty estimation and improved classification accuracy by means of simple classifiers in the latent space.
These investigations are adopted in the automotive application of vehicle interior rear seat occupant classification. For the latter, we release a synthetic dataset with several fine-grained extensions such that all the aforementioned topics can be investigated in isolation, or together, in a single application environment. We provide quantitative evidence that machine learning, and in particular deep learning methods cannot readily be used in industrial applications when only a limited amount of variation is available for training. The latter can, however, often be the case because of constraints enforced by the application to be considered and financial limitations.
In recent years, deep learning has made substantial improvements in various fields like image understanding, Natural Language Processing (NLP), etc. These huge advancements have led to the release of many commercial applications which aim to help users carry out their daily tasks. Personal digital assistants are one such successful application of NLP, having a diverse userbase from all age groups. NLP tasks like Natural Language Understanding (NLU) and Natural Language Generation (NLG) are core components for building these assistants. However, like any other deep learning model, the growth of NLU & NLG models is directly coupled with tremendous amounts of training examples, which are expensive to collect due to annotator costs. Therefore, this work investigates the methodologies to build NLU and NLG systems in a data-constrained setting.
We evaluate the problem of limited training data in multiple scenarios like limited or no data available when building a new system, availability of a few labeled examples when adding a new feature to an existing system, and changes in the distribution of test data during the lifetime of a deployed system.
Motivated by the standard methods to handle data-constrained settings, we propose novel approaches to generate data and exploit latent representations to overcome performance drops emerging from limited training data.We propose a framework to generate high-quality synthetic data when few training examples are available for a newly added feature for dialogue agents. Our interpretation-to-text model uses existing training data for bootstrapping new features and improves the accuracy of downstream tasks of intent classification and slot labeling. Following, we study a few-shot setting and observe that generation systems face a low semantic coverage problem. Hence, we present an unsupervised NLG algorithm that ensures that all relevant semantic information is present in the generated text.
We also study to see if we really need all training examples for learning a generalized model. We propose a data selection method that selects the most informative training examples to train Visual Question Answering (VQA) models without erosion of accuracy. We leverage the already available inter-annotator agreement and design a diagnostic tool, called (EaSe), that leverages the entropy and semantic similarity of answer patterns.
Finally, we discuss two empirical studies to understand the feature space of VQA models and show how language model pre-training and exploiting multimodal embedding space allows for building data constrained models ensuring minimal or no accuracy losses.
In recent years, Augmented Reality has made its way into everyday devices. Most smartphones are AR-enabled, providing applications like pedestrian navigation, Point of Interest highlighting, gaming, and retail. The high-tech industry has been focused on developing smartglasses to present virtual elements directly in front of the viewers’ eyes, allowing more immersive AR experiences. Smartglasses can also be deployed while driving for an enhanced and more safe experience. A 3D registered augmentation of the real world with navigation arrows, lane highlighting, or warnings can decrease the duration of inattentiveness regarding driving due to glancing at other screens. Enabling HMDs’ usage inside cars requires knowing its exact position and orientation (6-DoF pose) in the car. This necessitates sensors either built inside the AR glasses or the car. In a car, the latter option called outside-in tracking is more attractive due to two reasons. First, AR glasses containing different sensor sets exist, hampering finding one single solution for different HMDs. Second, the view from the driver’s perspective combines static interior and dynamic exterior features, complicating finding a reliable set of features. Nowadays, tracking methods utilize Deep Learning for a more generalizable and accurate derivation of the 6-DoF pose. They achieve outstanding results for head and object pose estimation. In this thesis, we present Deep Learning-based in-car 6-DoF AR glasses pose estimation approaches. The goal of the work is an exploration of accurate HMD pose estimation with the help of neural networks. The thesis achieves this by investigating numerous pose estimation techniques. Evaluations on the recorded HMDPose dataset constitute the foundation for this, consisting of infrared images of drivers wearing different HMD models. First, algorithms based on images are derived and evaluated on the dataset. For comparison, we carried out an evaluation on image-based methods considering time information. Further, pose estimation based on point clouds, generated out of infrared images, are analyzed. An investigation of various head pose estimation methods to derive its potential use are conducted. In conclusion, we introduce several highly accurate AR glasses pose estimators. The HMD pose alone achieves better results than the head pose and the combination of the head and HMD. Especially our image-based methods with optional usage of time information can efficiently and accurately regress the AR glasses pose. Our algorithms show excellent estimation results on live data when deployed inside a car, making seamless in-car HMD usage possible in the future.
The wireless spectrum is already a scarce good, shared by multiple competing technologies such as Bluetooth, ZigBee and Wi-Fi, and the hunger for traffic is only increasing. Due to the heterogeneity of the existing wireless technologies and the real threat that interference poses to network performance, sophisticated techniques must be developed to ensure acceptable levels of quality of service.
In this thesis, we present a passive channel sensing scheme based on both energy and signal detection, that primarily considers the spectrum occupation of foreign traffic while allowing for additional complementary information such as the signal-to-noise ratio. The resulting channel quality metric is first corrected for the spectrum occupation of internal transmissions and later aggregated with help of a moving average followed by an exponential weighted moving average. This aggregation keeps the metric both sufficiently stable and adaptive to significant changes in channel usage. Moreover, the channel quality metric is made volatility-aware by penalizing qualities proportionally to their downward volatility. This yields a conservative metric and allows to differentiate channels with similar aggregated qualities but different volatility behavior.
Our second main contribution is in the form of a schedule-based channel sensing protocol, in which nodes possess two network interfaces, one for communication and one for channel sensing. Channel sensing schedules are derived from communication schedules, i.e. channel hopping sequences used for communication, with help of a stochastic local search-based heuristic that attempts to minimize channel sensing bias, the channel overlap between both schedules and to maximize overlap fairness. This minimizes the effect of internal transmissions in the resulting channel quality metric, allowing nodes to derive channel quality primarily based on foreign traffic in an unbiased manner.
Finally, we propose and implement a stabilization protocol for keeping nodes in an ad-hoc network tick-synchronized and schedule-consistent w.r.t. a communication schedule. This stabilization protocol makes use of special messages, namely tick frames for synchronization, channel quality reports for sharing local views of channel conditions and schedule reports for disseminating the global communication hopping sequence. The communication schedules are computed by a master node based on an aggregation of local channel quality views and the re-computation of these schedules is triggered by significant changes in channel conditions. The resulting protocol is robust against changes in topology and channel conditions.
Several applications have emerged and benefited from the recent advancements in wireless communication technologies. In the case of industrial automation, the wireless networks substituted wired networks to control and monitor the production systems and the factory environment. In such use cases, a common requirement is communication reliability. Technologies based on IEEE 802.15.4, such as WirelessHart and ZigBee developed for industrial applications, offer deterministic guarantees using reservation-based medium access. However, it is becoming more challenging for these technologies to guarantee their sufficiently predictable behavior, as the number of consumer electronics equipped with wireless communication technologies operating in the 2.4 GHz ISM band shared by IEEE 802.15.4 is increasing day by day.
Meanwhile, developments in WiFi technology opened the opportunity to use WiFi for industrial applications. Compared to the technologies based on IEEE 802.15.4, WiFi offers significantly higher transmission rates, and the off-the-shelf commodity WiFi hardwares are available at a low cost. However, when using a contention-based technology such as WiFi for industrial applications, additional measures are required to guarantee the specified statistical reliability.
This thesis lays the foundations for developing a multi-hop wireless control network using off-the-shelf IEEE 802.11 (WiFi) hardware operating in contention mode that can satisfy the specified reliability requirements of the applications. In a multi-hop wireless network, the communication reliability between the nodes depends on the routes determined by the routing protocol and managing these routes. We introduce a novel Quality-of-Service (QoS) routing protocol for contention-based wireless technologies such as WiFi that prioritizes reliability as the QoS requirement for route selection. The proposed routing protocol relies on different aspects of the network to determine and manage the routes. For instance, it requires algorithms and protocols to monitor and measure link quality, available bandwidth, or medium overload. Further, the determined routes require certain statistical link properties for the successful operation of the routes. We develop and evaluate different protocols, algorithms, and metrics to monitor and measure different aspects of the network in this thesis.
This dissertation describes the implementation, validation, and troubleshooting of ``Digital Twins'' in assembly processes of thin structures like parts from the automotive and aerospace industry. As requirements in terms of cost, weight, and human (pedestrian) safety are increasing for modern vehicles, thinner materials are used for exterior components. By that, components become softer but less stable which is challenging for the assembly processes and impacts the resulting quality. The most critical quality measures are gap and flushness as these are affecting aesthetics, wind noise, and fuel consumption of the final vehicle. To compensate for geometrical deviations, parts have adjustable mechanical interfaces which are used to tune in gaps and flushness for each individual assembly. For the components being assembled, individual process parameters depending on the geometry of the actual physical part must be defined. This is a challenging task that cannot be solved in a straightforward manner. However, assembly quality can be predicted by setting up individual Finite Element Method (FEM) simulation models for each part being assembled. These simulation models are called Digital Twin (DTs) as they are enriched with measured properties from the actual physical part. By that, precise predictions can be made and optimal assembly parameters for automated processes are derived. The demonstration use case in this dissertation is the assembly process of exterior car components made from sheet metals. For this kind of process, the geometrical deviations of individual components are crucial and have to be considered by the DT. To capture geometrical deviations, 3D-scanning is employed which provides a high-resolution point cloud representation of the actual physical part. This point cloud is processed further to obtain the DT that preserves the measured geometry. This dissertation tackles the following challenges: (a) setting up DTs on different level of details, (b) correctly post-processing 3D-scanned data to remove systematical measurement errors, (c) automatically morphing meshes to derive simulation models from measured point clouds, and (d) troubleshooting DTs with human-in-the-loop approaches. For all approaches, validations are provided that underline applicability and benefits. All methods and results are discussed on a high-level perspective and connections as well as the interplay between methods are elaborated. Each method either improves or extends existing approaches or provides benefits, i.e. higher precision, compared to existing solutions.
This PhD thesis is concerned with the visual analysis of time-dependent scalar field ensembles as occur in climate simulations.
Modern climate projections consist of multiple simulation runs (ensemble members) that vary in parameter settings and/or initial values, which leads to variations in the resulting simulation data.
The goal of ensemble simulations is to sample the space of possible futures under the given climate model and provide quantitative information about uncertainty in the results.
The analysis of such data is challenging because apart from the spatiotemporal data, also variability has to be analyzed and communicated.
This thesis presents novel techniques to analyze climate simulation ensembles visually.
A central question is how the data can be aggregated under minimized information loss.
To address this question, a key technique applied in several places in this work is clustering.
The first part of the thesis addresses the challenge of finding clusters in the ensemble simulation data.
Various distance metrics lend themselves for the comparison of scalar fields which are explored theoretically and practically.
A visual analytics interface allows the user to interactively explore and compare multiple parameter settings for the clustering and investigate the resulting clusters, i.e. prototypical climate phenomena.
A central contribution here is the development of design principles for analyzing variability in decadal climate simulations, which has lead to a visualization system centered around the new Clustering Timeline.
This is a variant of a Sankey diagram that utilizes clustering results to communicate climatic states over time coupled with ensemble member agreement.
It can reveal
several interesting properties of the dataset, such as:
into how many inherently similar groups the ensemble can be divided at any given time,
whether the ensemble diverges in general,
whether there are different phases in the time lapse, maybe periodicity, or outliers.
The Clustering Timeline is also used to compare multiple climate simulation models and assess their performance.
The Hierarchical Clustering Timeline is an advanced version of the above.
It introduces the concept of a cluster hierarchy that may group the whole dataset down to the individual static scalar fields into clusters of various sizes and densities recording the nesting relationship between them.
One more contribution of this work in terms of visualization research is, that ways are investigated how to practically utilize a hierarchical clustering of time-dependent scalar fields to analyze the data.
To this end, a system of different views is proposed which are linked through various interaction possibilities.
The main advantage of the system is that a dataset can now be inspected at an arbitrary level of detail without having to recompute a clustering with different parameters.
Interesting branches of the simulation can be expanded to reveal smaller differences in critical clusters or folded to show only a coarse representation of the less interesting parts of the dataset.
The last building block of the suit of visual analysis methods developed for this thesis aims at a robust, (largely) automatic detection and tracking of certain features in a scalar field ensemble.
Techniques are presented that I found can identify and track super- and sub-levelsets.
And I derive “centers of action” from these sets which mark the location of extremal climate phenomena that govern the weather (e.g. Icelandic Low and Azores High).
The thesis also presents visual and quantitative techniques to evaluate the temporal change of the positions of these centers; such a displacement would be likely to manifest in changes in weather.
In a preliminary analysis with my collaborators, we indeed observed changes in the loci of the centers of action in a simulation with increased greenhouse gas concentration as compared to pre-industrial concentration levels.
One of the biggest social issues in mature societies such as Europe and Japan
is the aging population and declining birth rate. These societies have a serious
problem with the retirement of the expert workers, doctors, and engineers etc.
Especially in the sectors that require long time to make experts in fields like medicine and industry; the retirement and injuries of the experts, is a serious problem. The technology to support the training and assessment of skilled workers (like doctors, manufacturing
workers) is strongly required for the society. Although there are some solutions for
this problem, most of them are video-based which violates the privacy of the subjects.
Furthermore, they are not easy to deploy due to the need for large training data.
This thesis provides a novel framework to recognize, analyze, and assess human
skills with minimum customization cost. The presented framework tackles this problem
in two different domains, industrial setup and medical operations of catheter-based
cardiovascular interventions (CBCVI).
In particular, the contributions of this thesis are four-fold. First, it proposes an
easy-to-deploy framework for human activity recognition based on zero-shot learning
approach, which is based on learning basic actions and objects. The model recognizes
unseen activities by combinations of basic actions learned in a preliminary way and involved objects. Therefore, it is completely configurable by the user and can be used to detect completely new activities.
Second, a novel gaze-estimation model for attention driven object detection task is
presented. The key features of the model are: (i) usage of the deformable convolutional
layers to better incorporate spatial dependencies of different shapes of objects and
backgrounds, (ii) formulation of the gaze-estimation problem in two different way, as a
classification as well as a regression problem. We combine both formulations using a
joint loss that incorporates both the cross-entropy as well as the mean-squared error in
order to train our model. This enhanced the accuracy of the model from 6.8 by using only
the cross-entropy loss to 6.4 for the joint loss.
The third contribution of this thesis targets the area of quantification of quality of
i
actions using wearable sensor. To address the variety of scenarios, we have targeted two
possibilities: a) both expert and novice data is available , b) only expert data is available,
a quite common case in safety critical scenarios.
Both of the developed methods from these scenarios are deep learning based. In the
first one, we use autoencoders with OneClass SVM, and in the second one we use the
Siamese Networks. These methods allow us to encode the expert’s expertise and to learn
the differences between novice and expert workers. This enables quantification of the
performance of the novice in comparison to the expert worker.
The fourth contribution, explicitly targets medical practitioners and provides a
methodology for novel gaze-based temporal spatial analysis of CBCVI data. The developed
methodology allows continuous registration and analysis of gaze data for analysis
of the visual X-ray image processing (XRIP) strategies of expert operators in live-cases scenarios and may assist in transferring experts’ reading skills to novices.
The development of machine learning algorithms and novel sensing modalities has boosted the exploration of human activity recognition(HAR) in recent years. In this work, we explored field-based sensing solutions and different machine learning models for HAR tasks to address the shortcomings of existing HAR sensing solutions, like the weak robustness of RF-based solution, environment-dependency of the optic-based solution, etc., aiming to supply a competitive and alternative sensing approach for HAR tasks.
Field, in physics, describes a region in which each point will be affected by force. Field sensing is potentially a low-cost, low-power, non-intrusive, privacy-respecting HAR solution that is ideal for long-term, wearable activity recording. By directly/indirectly monitoring the field strength or other field variation caused variables, some unsolved HAR problems could be addressed when other sensing solutions fail. An example is the social distance monitoring problem, where the most widely adopted approach is based on the Bluetooth signal strength measurement. However, the signal is so subtle that any object surrounding the signal emitter will cause signal attenuation. To guarantee the accuracy of social distance monitoring, we developed an induced magnetic field-based social distance monitoring system with an accuracy of a sub-ten centimetre. Moreover, the system is robust and resistant to environmental variations. Like Bluetooth, other RF-wave-based sensing modalities also face the multi-path effect caused by refraction. Thus their signal is unreliable for positioning applications where higher accuracy and robustness are needed. Besides the magnetic field, we also explored a natural static passive electric field, the field between the human body and surroundings, namely the human body capacitance(HBC). HBC is a physiological parameter describing the charge distribution difference between the body and the surroundings and is seldomly explored before. We developed a few wearable, low-cost, low power consumption hardware platforms, either based on an oscillating unit or discrete components composed sensing front end followed by a high resolution analog-to-digital module, to
monitor the variation of the parameter regarding the body movement and environmental variations. Compared with the inertial sensors, the HBC could deliver full-body movement perceiving, meaning that the movement of the legs could be perceived by a wrist-worn HBC sensing unit, which is far beyond the
sensing ability of an inertial sensing unit.
To summarize, we introduced two competitive field sensing modalities for HAR tasks, the magnetic field sensing for position-related services and the passive electric field sensing for full-body action and environmental variation sensing. Both of which were still in an infant stage and not fully explored in the community. The advantages of the two field sensing modalities were demonstrated with a series of position-related and motion-related experiments.
As the usage of concurrency in software has gained importance in the last years, and is still rising, new types of defects increasingly appeared in software. One of the most prominent and critical types of such new defect types are data races. Although research resulted in an increased effectiveness of dynamic quality assurance regarding data races, the efficiency in the quality assurance process still is a factor preventing widespread practical application. First, dynamic quality assurance techniques used for the detection of data races are inefficient. Too much effort is needed for conducting dynamic quality assurance. Second, dynamic quality assurance techniques used for the analysis of reported data races are inefficient. Too much effort is needed for analyzing reported data races and identifying issues in the source code.
The goal of this thesis is to enable efficiency improvements in the process of quality assurance for data races by: (1) analyzing the representation of the dynamic behavior of a system under test. The results are used to focus instrumentation of this system, resulting in a lower runtime overhead during test execution compared to a full instrumentation of this system. (2) Analyzing characteristics and preprocessing of reported data races. The results of the preprocessing are then provided to developers and quality assurance personnel, enabling an analysis and debugging process, which is more efficient than traditional analysis of data race reports. Besides dynamic data race detection, which is complemented by the solution, all steps in the process of dynamic quality assurance for data races are discussed in this thesis.
The solution for analyzing UML Activities for nodes possibly executing in parallel to other nodes or themselves is based on a formal foundation using graph theory. A major problem that has been solved in this thesis was the handling of cycles within UML Activities. This thesis provides a dynamic limit for the number of cycle traversals, based on the elements of each UML Activity to be analyzed and their semantics. Formal proofs are provided with regard to the creation of directed acyclic graphs and with regard to their analysis concerning the identification of elements that may be executed in parallel to other elements. Based on an examination of the characteristics of data races and data race reports, the results of dynamic data race detection are preprocessed and the outcome of this preprocessing is presented to users for further analysis.
This thesis further provides an exemplary application of the solution idea, of the results of analyzing UML Activities, and an exemplary examination of the efficiency improvement of the dynamic data race detection, which showed a reduction in the runtime overhead of 44% when using the focused instrumentation compared to full instrumentation. Finally, a controlled experiment has been set up and conducted to examine the effects of the preprocessing of reported data races on the efficiency of analyzing data race reports. The results show that the solution presented in this thesis enables efficiency improvements in the analysis of data race reports between 190% and 660% compared to using traditional approaches.
Finally, opportunities for future work are shown, which may enable a broader usage of the results of this thesis and further improvements in the efficiency of quality assurance for data races.
Towards PACE-CAD Systems
(2022)
Despite phenomenal advancements in the availability of medical image datasets and the development of modern classification algorithms, Computer-Aided Diagnosis (CAD) has had limited practical exposure in the real-world clinical workflow. This is primarily because of the inherently demanding and sensitive nature of medical diagnosis that can have far-reaching and serious repercussions in case of misdiagnosis. In this work, a paradigm called PACE (Pragmatic, Accurate, Confident, & Explainable) is presented as a set of some of must-have features for any CAD. Diagnosis of glaucoma using Retinal Fundus Images (RFIs) is taken as the primary use case for development of various methods that may enrich an ordinary CAD system with PACE. However, depending on specific requirements for different methods, other application areas in ophthalmology and dermatology have also been explored.
Pragmatic CAD systems refer to a solution that can perform reliably in day-to-day clinical setup. In this research two, of possibly many, aspects of a pragmatic CAD are addressed. Firstly, observing that the existing medical image datasets are small and not representative of images taken in the real-world, a large RFI dataset for glaucoma detection is curated and published. Secondly, realising that a salient attribute of a reliable and pragmatic CAD is its ability to perform in a range of clinically relevant scenarios, classification of 622 unique cutaneous diseases in one of the largest publicly available datasets of skin lesions is successfully performed.
Accuracy is one of the most essential metrics of any CAD system's performance. Domain knowledge relevant to three types of diseases, namely glaucoma, Diabetic Retinopathy (DR), and skin lesions, is industriously utilised in an attempt to improve the accuracy. For glaucoma, a two-stage framework for automatic Optic Disc (OD) localisation and glaucoma detection is developed, which marked new state-of-the-art for glaucoma detection and OD localisation. To identify DR, a model is proposed that combines coarse-grained classifiers with fine-grained classifiers and grades the disease in four stages with respect to severity. Lastly, different methods of modelling and incorporating metadata are also examined and their effect on a model's classification performance is studied.
Confidence in diagnosing a disease is equally important as the diagnosis itself. One of the biggest reasons hampering the successful deployment of CAD in the real-world is that medical diagnosis cannot be readily decided based on an algorithm's output. Therefore, a hybrid CNN architecture is proposed with the convolutional feature extractor trained using point estimates and a dense classifier trained using Bayesian estimates. Evaluation on 13 publicly available datasets shows the superiority of this method in terms of classification accuracy and also provides an estimate of uncertainty for every prediction.
Explainability of AI-driven algorithms has become a legal requirement after Europe’s General Data Protection Regulations came into effect. This research presents a framework for easy-to-understand textual explanations of skin lesion diagnosis. The framework is called ExAID (Explainable AI for Dermatology) and relies upon two fundamental modules. The first module uses any deep skin lesion classifier and performs detailed analysis on its latent space to map human-understandable disease-related concepts to the latent representation learnt by the deep model. The second module proposes Concept Localisation Maps, which extend Concept Activation Vectors by locating significant regions corresponding to a learned concept in the latent space of a trained image classifier.
This thesis probes many viable solutions to equip a CAD system with PACE. However, it is noted that some of these methods require specific attributes in datasets and, therefore, not all methods may be applied on a single dataset. Regardless, this work anticipates that consolidating PACE into a CAD system can not only increase the confidence of medical practitioners in such tools but also serve as a stepping stone for the further development of AI-driven technologies in healthcare.
Dealing with Dependence in the End-to-End Performance Analysis in Stochastic Network Calculus
(2022)
Communication networks, in particular the Internet, have become a pivotal part of our life. Since their beginnings, a key aspect of their applicability has been the performance. Safety-critical applications, for example, can sometimes only be implemented in a responsible manner if guarantees about their end-to-end delay can be made. A mathematical modeling and performance evaluation of communication networks requires a powerful set of tools that is able to incorporate their increasing complexity.
The stochastic network calculus (SNC) is a versatile, mathematical framework that allows for a calculation of probabilistic end-to-end performance bounds of distributed systems. Its flexibility enables to incorporate a large class of different schedulers as well as different models of traffic processes beyond the assumption of Poisson arrivals that is predominant in queueing theory-based analyses. It originates in the so-called deterministic network analysis (DNC) in the 90's of the 20th century that was introduced to provide deterministic, ``hard'' guarantees that are of relevance, e.g., in the context of real-time systems. While the DNC of today can be used to calculate fast and accurate delay bounds of arbitrary feedforward networks, the SNC is still in a significantly earlier stage. In particular, method-pertinent dependencies, i.e., a phenomenon that occurs when independent flows become stochastically dependent after sharing resources in the network, can be considered a major challenge in the SNC with moment-generating functions (MGFs).
This thesis argues to contribute to the SNC in several ways. First, we show that the ``pay multiplexing only once'' (PMOO) analysis known from the DNC is also possible in the SNC. Not only does it significantly improve end-to-end delay bounds, it also needs to consider less method-pertinent dependencies. Therefore, complexity and runtimes of the calculation are greatly reduced. Second, we introduce the concept of negative dependence to the SNC with MGFs and give numerical evidence that this can further lead to better performance bounds. Third, for the larger problem of end-to-end performance bounds of tree networks, we introduce so-called ''h-mitigators'', a modification in the calculation of MGF output bounds. It is minimally invasive, all existing results and procedures are still applicable, and improves performance bounds. As a fourth contribution, we conduct extensive numerical evaluations to substantiate our claims. Moreover, we made the respective code, the ''SNC MGF toolbox'', publicly available to ensure that the results are reproducible. At last, we conduct different stochastic analyses of a popular fair scheduler, generalized processor sharing (GPS). We give an overview of the state-of-the-art analyses in the SNC and substantiate the comparison through numerical evaluations.
To support scientific work with large and complex data the field of scientific visualization emerged in computer science and produces images through computational analysis of the data. Frameworks for combination of different analysis and visualization modules allow the user to create flexible pipelines for this purpose and set the standard for interactive scientific visualization used by domain scientists.
Existing frameworks employ a thread-parallel message-passing approach to parallel and distributed scalability, leaving the field of scientific visualization in high performance computing to specialized ad-hoc implementations. The task-parallel programming paradigm proves promising to improve scalability and portability in high performance computing implementations and thus, this thesis aims towards the creation of a framework for distributed, task-based visualization modules and pipelines.
The major contribution of the thesis is the establishment of modules for Merge Tree construction and (based on the former) topological simplification. Such modules already form a necessary first step for most visualization pipelines and can be expected to increase in importance for larger and more complex data produced and/or analysed by high performance computing.
To create a task-parallel, distributed Merge Tree construction module the construction process has to be completely revised. We derive a novel property of Merge Tree saddles and introduce a novel task-parallel, distributed Merge Tree construction method that has both good performance and scalability. This forms the basis for a module for topological simplification which we extend by introducing novel alternative simplification parameters that aim to reduce the importance of prior domain knowledge to increase flexibility in typical high performance computing scenarios.
Both modules lay the groundwork for continuative analysis and visualization steps and form a fundamental step towards an extensive task-parallel visualization pipeline framework for high performance computing.
This dissertation was developed in the context of the BMBF and EU/ECSEL funded
projects GENIAL! and Arrowhead Tools. In these projects the chair examines methods
of specifications and cooperations in the automotive value chain from OEM-Tier1-Tier2.
Goal of the projects is to improve communication and collaborative planning, especially
in early development stages. Besides SysML, the use of agreed vocabularies and on-
tologies for modeling requirements, overall context, variants, and many other items, is
targeted. This thesis proposes a web database, where data from the collaborative requirements elicitation is combined with an ontology-based approach that uses reasoning
capabilities.
For this purpose, state-of-the-art ontologies have been investigated and integrated that
entail domains like hardware/software, roadmapping, IoT, context, innovation and oth-
ers. New ontologies have been designed like a HW / SW allocation ontology and a
domain-specific "eFuse ontology" as well as some prototypes. The result is a modular
ontology suite and the GENIAL! Basic Ontology that allows us to model automotive
and microelectronic functions, components, properties and dependencies based on the
ISO26262 standard among these elements. Furthermore, context knowledge that influences design decisions such as future trends in legislation, society, environment, etc. is
included. These knowledge bases are integrated in a novel tool that allows for collabo-
rative innovation planning and requirements communication along the automotive value
chain. To start off the work of the project, an architecture and prototype tool was developed. Designing ontologies and knowing how to use them proved to be a non-trivial
task, requiring a lot of context and background knowledge. Some of this background
knowledge has been selected for presentation and was utilized either in designing models
or for later immersion. Examples are basic foundations like design guidelines for ontologies, ontology categories and a continuum of expressiveness of languages and advanced
content like multi-level theory, foundational ontologies and reasoning.
Finally, at the end, we demonstrate the overall framework, and show the ontology with
reasoning, database and APPEL/SysMD (AGILA ProPErty and Dependency Descrip-
tion Language / System MarkDown) and constraints of the hardware / software knowledge base. There, by example, we explore and solve roadmap constraints that are coupled
with a car model through a constraint solver.
Controller design for continuous dynamical systems is a core algorithmic problem in the design of cyber-physical systems (CPS). When the CPS application is safety critical, additionally we require the controller to have strong correctness guarantees. One approach for this design problem is to use simpler discrete abstraction of the original continuous system, on which known reactive synthesis methods can be used to design the controller. This approach is known as the abstraction-based controller design (ABCD) paradigm.
In this thesis, we build ABCD procedures which are faster and more modular compared to the state-of-the-art, and can handle problems which were beyond the scope of the existing techniques.
Usually, existing ABCD approaches use state space discretization for computing the abstractions, for which the procedures do not scale well for larger systems. Our first contribution is a multi-layered ABCD algorithm, where we combine coarse abstractions and lazily computed fine abstractions to improve scalability. So far, we only address reach-avoid and safety specifications, for which our prototype tool (called Mascot) showed up to an order of magnitude speedup on standard benchmark examples.
Second, we consider the problem of modular design of sound local controllers for a network of local discrete abstractions communicating via discrete/boolean variables and having local specifications. We propose a sound algorithm, where the systems negotiate a pair of local assume-guarantee contracts, in order to synchronize on a set of non-conflicting and correct behaviors. As a by-product, we also obtain a set of local controllers for the systems which ensure simultaneous satisfaction of the local specifications. We show the effectiveness of the our algorithm using a prototype tool (called Agnes) on a set of discrete benchmark examples.
Our third contribution is a novel ABCD algorithm for a more expressive model of nonlinear dynamical systems with stochastic disturbances and ω-regular specifications. This part has two subparts, which are of significant merits on their own rights. First, we present an abstraction algorithm for nonlinear stochastic systems using 2.5-player games (turn-based stochastic graph games). We show that an almost sure winning strategy in this abstract 2.5-player game gives us a sound controller for the original system for satisfying the specification with probability one. Second, we present symbolic algorithms for a seemingly different class of 2-player games with certain environmental fairness assumptions, which can also be used to efficiently compute winning strategies in the aforementioned abstract 2.5-player game. Using our prototype tool (Mascot-SDS), we show that our algorithm significantly outperforms the state-of-the-art implementation on standard benchmark examples from the literature.
Comparative Uncertainty Visualization for High-Level Analysis of Scalar- and Vector-Valued Ensembles
(2022)
With this thesis, I contribute to the research field of uncertainty visualization, considering parameter dependencies in multi valued fields and the uncertainty of automated data analysis. Like uncertainty visualization in general, both of these fields are becoming more and more important due to increasing computational power, growing importance and availability of complex models and collected data, and progress in artificial intelligence. I contribute in the following application areas:
Uncertain Topology of Scalar Field Ensembles.
The generalization of topology-based visualizations to multi valued data involves many challenges. An example is the comparative visualization of multiple contour trees, complicated by the random nature of prevalent contour tree layout algorithms. I present a novel approach for the comparative visualization of contour trees - the Fuzzy Contour Tree.
Uncertain Topological Features in Time-Dependent Scalar Fields.
Tracking features in time-dependent scalar fields is an active field of research, where most approaches rely on the comparison of consecutive time steps. I created a more holistic visualization for time-varying scalar field topology by adapting Fuzzy Contour Trees to the time-dependent setting.
Uncertain Trajectories in Vector Field Ensembles.
Visitation maps are an intuitive and well-known visualization of uncertain trajectories in vector field ensembles. For large ensembles, visitation maps are not applicable, or only with extensive time requirements. I developed Visitation Graphs, a new representation and data reduction method for vector field ensembles that can be calculated in situ and is an optimal basis for the efficient generation of visitation maps. This is accomplished by bringing forward calculation times to the pre-processing.
Visually Supported Anomaly Detection in Cyber Security.
Numerous cyber attacks and the increasing complexity of networks and their protection necessitate the application of automated data analysis in cyber security. Due to uncertainty in automated anomaly detection, the results need to be communicated to analysts to ensure appropriate reactions. I introduce a visualization system combining device readings and anomaly detection results: the Security in Process System. To further support analysts I developed an application agnostic framework that supports the integration of knowledge assistance and applied it to the Security in Process System. I present this Knowledge Rocks Framework, its application and the results of evaluations for both, the original and the knowledge assisted Security in Process System. For all presented systems, I provide implementation details, illustrations and applications.
Robotic systems are entering the stage. Enabled by advances in both hardware components and software techniques, robots are increasingly able to operate outside of factories, assist humans, and work alongside them. The limiting factor of robots’ expansion remains the programming of robotic systems. Due to the many diverse skills necessary to build a multi-robot system, only the biggest organizations are able to innovate in the space of services provided by robots.
To make developing new robotic services easier, in this dissertation I propose a program- ming model in which users (programmers) give a declarative specification of what needs to be accomplished, and then a backend system makes sure that the specification is safely and reliably executed. I present Antlab, one such backend system. Antlab accepts Linear Temporal Logic (LTL) specifications from multiple users and executes them using a set of robots of different capabilities.
Building on the experience acquired implementing Antlab, I identify problems arising from the proposed programming model. These problems fall into two broad categories, specification and planning.
In the category of specification problems, I solve the problem of inferring an LTL formula from sets of positive and negative example traces, as well as from a set of positive examples only. Building on top of these solutions, I develop a method to help users transfer their intent into a formal specification. The approach taken in this dissertation is combining the intent signals from a single demonstration and a natural language description given by a user. A set of candidate specifications is inferred by encoding the problem as a satisfiability problem for propositional logic. This set is narrowed down to a single specification through interaction with the user; the user approves or declines generated simulations of the robot’s behavior in different situations.
In the category of planning problems, I first solve the problem of planning for robots that are currently executing their tasks. In such a situation, it is unclear what to take as the initial state for planning. I solve the problem by considering multiple, speculative initial states. The paths from those states are explored based on a quality function that repeatedly estimates the planning time. The second problem is a problem of reinforcement learning when the reward function is non-Markovian. The proposed solution consists of iteratively learning an automaton representing the reward function and using it to guide the exploration.
Data-driven and Sparse-to-Dense Concepts in Scene Flow Estimation for Automotive Applications
(2022)
Highly assisted driving and autonomous vehicles require a detailed and accurate perception of the environment. This includes the perception of the 3D geometry of the scene and the 3D motion of other road users. The estimation of both based on images is known as the scene flow problem in computer vision. This thesis deals with a solution to the scene flow problem that is suitable for application in autonomous vehicles. This application imposes strict requirements on accuracy, robustness, and speed. Previous work was lagging behind in at least one of these metrics. To work towards the fulfillment of those requirements, the sparse-to-dense concept for scene flow estimation is introduced in this thesis. The idea can be summarized as follows: First, scene flow is estimated for some points of the scene for which this can be done comparatively easily and reliably. Then, an interpolation is performed to obtain a dense estimate for the entire scene. Because of the separation into two steps, each part can be optimized individually. In a series of experiments, it is shown that the proposed methods achieve competitive results and are preferable to previous techniques in some aspects. As a second contribution, individual components in the sparse-to-dense pipeline are replaced by deep learning modules. These are a highly localized and highly accurate feature descriptor to represent pixels for dense matching, and a network for robust and generic sparse-to-dense interpolation. Compared to end-to-end architectures, the advantage of deep modules is that they can be trained more effciently with data from different domains. The recombination approach applies a similar concept as the sparse-to-dense approach by solving and combining less diffcult, auxiliary sub-problems. 3D geometry and 2D motion are estimated separately, the individual results are combined, and then also interpolated into a dense scene flow. As a final contribution, the thesis proposes a set of monolithic end-to-end networks for scene flow estimation.
Today’s digital world would be unthinkable without complex data sets. Whether in private, business or industrial environments, complex data provide the basis for important and critical decisions and determine many processes, some of which are automated. This is often associated with Big Data. However, often only one aspect of the usual Big Data definitions is sufficient and a human observer can no longer capture the data completely and correctly. In this thesis, different approaches are presented in order to master selected challenges in a more effective, efficient and userfriendly way. The approaches range from easier pre-processing of data sets for later analysis and the identification of design guidelines of such assistants, new visualization techniques for presenting uncertainty, extensions of existing visualizations for categorical data, concepts for time-saving selection methods for subsets of data points and faster navigation and zoom interaction–especially in the web-based area with enormous amounts of data–to new and innovative orientation-based interaction metaphors for mobile devices as well as stationary working environments. Evaluations and appropriate use case of the individual approaches show the usability also in comparison with state-of-the-art techniques.
Sequence learning describes the process of understanding the spatio-temporal
relations in a sequence in order to classify it, label its elements or generate
new sequences. Due to the prevalence of structured sequences in nature
and everyday life, it has many practical applications including any language
related processing task. One particular such task that has seen recent success
using sequence learning techniques is the optical recognition of characters
(OCR).
State-of-the-art sequence learning solutions for OCR achieve high performance
through supervised training, which requires large amounts of transcribed
training data. On the other hand, few solutions have been proposed on how
to apply sequence learning in the absence of such data, which is especially
common for hard to transcribe historical documents. Rather than solving
the unsupervised training problem, research has focused on creating efficient
methods for collecting training data through smart annotation tools or generating
synthetic training data. These solutions come with various limitations
and do not solve all of the related problems.
In this work, first the use of erroneous transcriptions for supervised sequence
learning is introduced and it is described how this concept can be applied in
unsupervised training scenarios by collecting or generating such transcriptions.
The proposed OCR pipeline reduces the need of domain specific expertise
to apply OCR, with the goal of making it more accessible. Furthermore, an
approach for evaluating sequence learning OCR models in the absence of
reference transcriptions is presented and its different properties compared
to the standard method are discussed. In a second approach, unsupervised
OCR is treated as an alignment problem between the latent features of the
different language modalities. The outlined solution is to extract language
properties from both the text and image domain through adversarial training
and learn to align them by adding a cycle consistency constraint. The proposed
approach has some strict limitations on the input data, but the results
encourage future research into more widespread applications.
Recommender systems recommend items (e.g., movies, products, books) to users. In this thesis, we proposed two comprehensive and cluster-induced recommendation-based methods: Orthogonal Inductive Matrix Completion (OMIC) and Burst-induced Multi-armed Bandit (BMAB). Given the presence of side information, the first method is categorized as context-aware. OMIC is the first matrix completion method to approach the problem of incorporating biases, side information terms and a pure low-rank term into a single flexible framework with a well-principled optimization procedure. The second method, BMAB, is context-free. That is, it does not require any side data about users or items. Unlike previous context-free multi-armed bandit approaches, our method considers the temporal dynamics of human communication on the web and treats the problem in a continuous time setting. We built our models' assumptions under solid theoretical foundations. For OMIC, we provided theoretical guarantees in the form of generalization bounds by considering the distribution-free case: no assumptions about the sampling distribution are made. Additionally, we conducted a theoretical analysis of community side information when the sampling distribution is known and an adjusted nuclear norm regularization is applied. We showed that our method requires just a few entries to accurately recover the ratings matrix if the structure of the ground truth closely matches the cluster side information. For BMAB, we provided regret guarantees under mild conditions that demonstrate how the system's stability affects the expected reward. Furthermore, we conducted extensive experiments to validate our proposed methodologies. In a controlled environment, we implemented synthetic data generation techniques capable of replicating the domains for which OMIC and BMAB were designed. As a result, we were able to analyze our algorithms' performance across a broad spectrum of ground truth regimes. Finally, we replicated a real-world scenario by utilizing well-established recommender datasets. After comparing our approaches to several baselines, we observe that they achieved state-of-the-art results in terms of accuracy. Apart from being highly accurate, these methods improve interpretability by describing and quantifying features of the datasets they characterize.
In the past, information and knowledge dissemination was relegated to the
brick-and-mortar classrooms, newspapers, radio, and television. As these
processes were simple and centralized, the models behind them were well
understood and so were the empirical methods for optimizing them. In today’s
world, the internet and social media has become a powerful tool for information
and knowledge dissemination: Wikipedia gets more than 1 million edits per day,
Stack Overflow has more than 17 million questions, 25% of US population visits
Yahoo! News for articles and discussions, Twitter has more than 60 million
active monthly users, and Duolingo has 25 million users learning languages
online. These developments have introduced a paradigm shift in the process of
dissemination. Not only has the nature of the task moved from being centralized
to decentralized, but the developments have also blurred the boundary between
the creator and the consumer of the content, i.e., information and knowledge.
These changes have made it necessary to develop new models, which are better
suited to understanding and analysing the dissemination, and to develop new
methods to optimize them.
At a broad level, we can view the participation of users in the process of
dissemination as falling in one of two settings: collaborative or competitive.
In the collaborative setting, the participants work together in crafting
knowledge online, e.g., by asking questions and contributing answers, or by
discussing news or opinion pieces. In contrast, as competitors, they vie for
the attention of their followers on social media. This thesis investigates both
these settings.
The first part of the thesis focuses on the understanding and analysis of
content being created online collaboratively. To this end, I propose models for
understanding the complexity of the content of collaborative online discussions
by looking exclusively at the signals of agreement and disagreement expressed
by the crowd. This leads to a formal notion of complexity of opinions and
online discussions. Next, I turn my attention to the participants of the crowd,
i.e., the creators and consumers themselves, and propose an intuitive model for
both, the evolution of their expertise and the value of the content they
collaboratively contribute and learn from on online Q&A based forums. The
second part of the thesis explores the competitive setting. It provides methods
to help the creators gain more attention from their followers on social media.
In particular, I consider the problem of controlling the timing of the posts of
users with the aim of maximizing the attention that their posts receive under
the idealized setting of full-knowledge of timing of posts of others. To solve
it, I develop a general reinforcement learning based method which is shown to
have good performance on the when-to-post problem and which can be employed in
many other settings as well, e.g., determining the reviewing times for spaced
repetition which lead to optimal learning. The last part of the thesis looks at
methods for relaxing the idealized assumption of full knowledge. This basic
question of determining the visibility of one’s posts on the followers’ feeds
becomes difficult to answer on the internet when constantly observing the feeds
of all the followers becomes unscalable. I explore the links of this problem to
the well-studied problem of web-crawling to update a search engine’s index and
provide algorithms with performance guarantees for feed observation policies
which minimize the error in the estimate of visibility of one’s posts.
Data is the new gold and serves as a key to answer the five W’s (Who, What, Where, When, Why) and How’s of any business. Companies are now mining data more than ever and one of the most important aspects while analyzing this data is to detect anomalous patterns to identify critical patterns and points. To tackle the vital aspects of timeseries analysis, this thesis presents a novel hybrid framework that stands on three pillars: Anomaly Detection, Uncertainty Estimation,
and Interpretability and Explainability.
The first pillar is comprised of contributions in the area of time-series anomaly detection. Deep Anomaly Detection for Time-series (DeepAnT), a novel deep learning-based anomaly detection method, lies at the foundation of the proposed hybrid framework and addresses the inadequacy of traditional anomaly detection methods. To the best of the author’s knowledge, Convolutional Neural Network (CNN) was used for the first time in Deep Anomaly Detection for Time-series (DeepAnT) to robustly detect multiple types of anomalies in the tricky
and continuously changing time-series data. To further improve the anomaly detection performance, a fusion-based method, Fusion of
Statistical and Deep Learning for Anomaly Detection (FuseAD) is proposed. This method aims to combine the strengths of existing wellfounded
statistical methods and powerful data-driven methods.
In the second pillar of this framework, a hybrid approach that combines the high accuracy of the deterministic models with the posterior distribution approximation of Bayesian neural networks is proposed.
In the third pillar of the proposed framework, mechanisms to enable both HOW and WHY parts are presented.
In order to improve performance or conserve energy, modern hardware implementations have adopted weak memory models; that is, models of concurrency that allow more outcomes than the classic sequentially consistent (SC) model of execution. Modern programming languages similarly provide their own language-level memory models, which strive to allow all the behaviors allowed by the various hardware-level memory models, as well as those that can occur as a result of desired compiler optimizations.
As these weak memory models are often rather intricate, it can be difficult for programmers to keep track of all the possible behaviors of their programs. It is therefore very useful to have an abstraction layer over the model that can be used to ensure program correctness without reasoning about the underlying memory model. Program logics are a way of constructing such an abstraction—one can use their syntactic rules to reason about programs, without needing to understand the messy details of the memory model for which the logic has been proven sound.
Unfortunately, most of the work on formal verification in general, and program logics in particular, has so far assumed the SC model of execution. This means that new logics for weak memory have to be developed.
This thesis presents two such logics—fenced separation logic (FSL) and weak separation logic (Weasel)—which are sound for reasoning under two different weak memory models.
FSL considers the C/C++ concurrency memory model, supporting several of its advanced features. The soundness of FSL depends crucially on a specific strengthening of the model which eliminates a certain class of undesired behaviors (so-called out-of-thin-air behaviors) that were inadvertently allowed by the original C/C++ model.
Weasel works under weaker assumptions than FSL, considering a model which takes a more fine-grained approach to the out-of-thin-air problem. Weasel's focus is on exploring the programming constructs directly related to out-of-thin-air behaviors, and is therefore significantly less feature-rich than FSL.
Using FSL and Weasel, the thesis explores the key challenges in reasoning under weak memory models, and what effect different solutions to the out-of-thin-air problem have on such reasoning. It explains which reasoning principles are preserved when moving from a stronger to a weaker model, and develops novel proof techniques to establish soundness of logics under weaker models.
Using Enhanced Logic Programming Semantics for Extending and Optimizing Synchronous System Design
(2021)
The semantics of programming languages assign a meaning to the written program syntax.
Currently, the meaning of synchronous programming languages, which are especially designed to develop programs for reactive and embedded systems, is based on a formal semantics definition similar to Fitting`s fixpoint semantics for logic programs.
Nevertheless, it is possible to write a synchronous program code that does not evaluate to concrete values with the current semantics, which means those programs are currently seen to be not constructive.
In the last decades, the theoretical knowledge and representation of semantics for logic programming has increased, but not all theoretical results and achievements have found their way to practice and application in system design.
This thesis, in a first part, focuses on extensions to the semantics of synchronous programming languages to an evaluation similar to a well-founded semantics as defined in logic programming by van Gelder, Ross and Schlipf and to the stable model semantics as defined by Gelfond and Lifschitz. Particularly, this allows an evaluation for some of the currently not constructive programs where the semantics based on Fitting`s fixpoint fails.
It is shown that the extension to well-founded semantics is a conservative extension of Fitting`s semantics, so that the meaning for programs which were already constructive does not change. Finally, it is considered how one can still generate circuits that implement the considered synchronous programs with the well-founded semantics. Again, this is a conservative approach that does not modify the circuits generated by the so-far used synthesis procedures.
Answer set programming and the underlying stable model semantics describe problems by constraints and the related answer set solvers give all solutions to that problem as so-called answers. This allows the formulation of searching and planning problems as well as efficient solutions without having the need to develop special and possibly error-prone algorithms for every single application.
The semantics of the synchronous programming language Quartz is also extended to the stable model semantics. For this extension, two alternatives are discussed: First of all, a direct extension similar to the extension to well-founded semantics is discussed. Second, a transformation of synchronous programs to the available answer set programming languages is given, as this allows to directly use answer set solvers for the synthesis and optimization of synchronous systems.
The second part of the thesis contains further examples of the use of answer set programming in system design to emphasis their benefits for system design in general. The first example is hereby the generation of optimal/minimal interconnection-networks which allow non-blocking connections between n sources and n targets in parallel. As a second example, the stable model semantics is used to build a complete compiler chain, which transforms a given program to an optimal assembler code (called move code) for the new SCAD processor architecture which was developed at the University of Kaiserslautern. As a final part, the lessons learned from the two examples are shown by the means of some enhancement ideas for the synchronous programming language paradigm.
Deep learning has achieved significant improvements in a variety of tasks in computer vision applications with an open image dataset which has a large amount of data. However, the acquisition of a large number of the dataset is a challenge in real-world applications, especially if they are new eras for deep learning. Furthermore, the distribution of class in the dataset is often imbalanced. The data imbalance problem is frequently bottlenecks of the neural network performance in classification. Recently, the potential of generative adversarial networks (GAN) as a data augmentation method on minority data has been studied.
This dissertation investigates using GAN and transfer learning to improve the performance of the classification under imbalanced data conditions. We first propose a classification enhancement generative adversarial networks (CEGAN) to enhance the quality of generated synthetic minority data and more importantly, to improve the prediction accuracy in data imbalanced condition. Our experiments show that approximating the real data distribution using CEGAN improves the classification performance significantly in data imbalanced conditions compared with various standard data augmentation methods.
To further improve the performance of the classification, we propose a novel supervised discriminative feature generation method (DFG) for minority class dataset. DFG is based on the modified structure of Generative Adversarial Network consisting of four independent networks: generator, discriminator, feature extractor, and classifier. To augment the selected discriminative features of minority class data by adopting attention mechanism, the generator for class-imbalanced target task is trained while feature extractor and classifier are regularized with the pre-trained ones from large source data. The experimental results show that the generator of DFG enhances the augmentation of label-preserved and diverse features, and classification results are significantly improved on the target task.
In this thesis, these proposals are deployed to bearing fault detection and diagnosis of induction motor and shipping label recognition and validation for logistics. The experimental results for bearing fault detection and diagnosis conclude that the proposed GAN-based framework has good performance on the imbalanced fault diagnosis of rotating machinery. The experimental results for shipping label recognition and validation also show that the proposed method achieves better performance than many classical and state-of-the-art algorithms.
Medical cyber-physical systems (MCPS) emerged as an evolution of the relations between connected health systems, healthcare providers, and modern medical devices. Such systems combine independent medical devices at runtime in order to render new patient monitoring/control functionalities, such as physiological closed loops for controlling drug infusion or optimization of alarms. Despite the advances regarding alarm precision, healthcare providers still struggle with alarm flooding caused by the limited risk assessment models. Furthermore, these limitations also impose severe barriers on the adoption of automated supervision through autonomous actions, such as safety interlocks for avoiding overdosage. The literature has focused on the verification of safety parameters to assure the safety of treatment at runtime and thus optimize alarms and automated actions. Such solutions have relied on the definition of actuation ranges based on thresholds for a few monitored parameters. Given the very dynamic nature of the relevant context conditions (e.g., the patient’s condition, treatment details, system configurations, etc.), fixed thresholds are a weak means for assessing the current risk. This thesis presents an approach for enabling dynamic risk assessment for cooperative MCPS based on an adaptive Bayesian Networks (BN) model. The main aim of the approach is to support continuous runtime risk assessment of the current situation based on relevant context and system information. The presented approach comprises (i) a dynamic risk analysis constituent, which corresponds to the elicitation of relevant risk parameters, risk metric building, and risk metric management; and (ii) a runtime risk classification constituent, which aims to analyze the current situation risk, establish risk classes, and identify and deploy mitigation measures. The proposed approach was evaluated and its feasibility proved by means of simulated experiments guided by an international team of medical experts with a focus on the requirements of efficacy, efficiency, and availability of patient treatment.
Dataflow process networks (DPNs) consist of statically defined process nodes with First-In-First-Out (FIFO) buffered point-to-point connections. DPNs are intrinsically data-driven, i.e., node actions are not synchronized among each other and may fire whenever sufficient input operands arrived at a node. In this original form, DPNs are data-driven and therefore a suitable model of computation (MoC) for asynchronous and distributed systems. For DPNs having nodes with only static consumption/production rates, however, one can easily derive an optimal schedule that can then be used to implement the DPN in a time-driven (clock-driven) way, where each node fires according to the schedule.
Both data-driven and time-driven MoCs have their own advantages and disadvantages. For this reason, desynchronization techniques are used to convert clock-driven models into data-driven ones in order to more efficiently support distributed implementations. These techniques preserve the functional specification of the synchronous models and moreover preserve properties like deadlock-freedom and bounded memory usage that are otherwise difficult to ensure in DPNs. These desynchronized models are the starting point of this thesis.
While the general MoC of DPNs does not impose further restrictions, many different subclasses of DPNs representing different dataflow MoCs have been considered over time like Kahn process networks, cyclo-static and synchronous DPNs. These classes mainly differ in the kinds of behaviors of the processes which affect on the one hand the expressiveness of the DPN class as well as the methods for their analysis (predictability) and synthesis (efficiency). A DPN may be heterogeneous in the sense that different processes in the network may exhibit different kinds of behaviors. A heterogeneous DPN therefore can be effectively used to model and implement different components of a system with different kinds of processes and therefore different dataflow MoCs.
Design tools for modeling like Ptolemy and FERAL are used to model and to design parallel embedded systems using well-defined and precise MoCs, including different dataflow MoCs. However, there is a lack of automatic synthesis methods to analyze and to evaluate the artifacts exhibited by particular MoCs. Second, the existing design tools for synthesis are usually restricted to the weakest classes of DPNs, i.e., cyclo-static and synchronous DPNs where each tool only supports a specific dataflow MoC.
This thesis presents a model-based design based on different dataflow MoCs including their heterogeneous combinations. This model-based design covers in particular the automatic software synthesis of systems from DPN models. The main objective is to validate, evaluate and compare the artifacts exhibited by different dataflow MoCs at the implementation level of embedded systems under the supervision of a common design tool. We are mainly concerned about how these different dataflow MoCs affect the synthesis, in particular, how they affect the code generation and the final implementation on the target hardware. Moreover, this thesis also aims at offering an efficient synthesis method that targets and exploits heterogeneity in DPNs by generating implementations based on the kinds of behaviors of the processes.
The proposed synthesis design flow therefore generally starts from the desynchronized dataflow models and automatically synthesizes them for cross-vendor target hardware. In particular, it provides a synthesis tool chain, including different specialized code generators for specific dataflow MoCs, and a runtime system that finally maps models using a combination of different dataflow MoCs on the target hardware. Moreover, the tool chain offers a platform-independent code synthesis method based on the open computing language (OpenCL) that enables a more generalized synthesis targeting cross-vendor commercial off-the-shelf (COTS) heterogeneous platforms.
DeepKAF: A Knowledge Intensive Framework for Heterogeneous Case-Based Reasoning in Textual Domains
(2021)
Business-relevant domain knowledge can be found in plain text across message exchanges
among customer support tickets, employee message exchanges and other business transactions.
Decoding text-based domain knowledge can be a very demanding task since traditional
methods focus on a comprehensive representation of the business and its relevant paths. Such
a process can be highly complex, time-costly and of high maintenance effort, especially in
environments that change dynamically.
In this thesis, a novel approach is presented for developing hybrid case-based reasoning
(CBR) systems that bring together the benefits of deep learning approaches with CBR advantages.
Deep Knowledge Acquisition Framework (DeepKAF) is a domain-independent
framework that features the usage of deep neural networks and big data technologies to decode
the domain knowledge with the minimum involvement from the domain experts. While
this thesis is focusing more on the textual data because of the availability of the datasets, the
target CBR systems based on DeepKAF are able to deal with heterogeneous data where a
case can be represented by different attribute types and automatically extract the necessary
domain knowledge while keeping the ability to provide an adequate level of explainability.
The main focus within this thesis are automatic knowledge acquisition, building similarity
measures and cases retrieval.
Throughout the progress of this research, several sets of experiments have been conducted
and validated by domain experts. Past textual data produced over around 15 years have
been used for the needs of the conducted experiments. The text produced is a mixture
between English and German texts that were used to describe specific domain problems
with a lot of abbreviations. Based on these, the necessary knowledge repositories were built
and used afterwards in order to evaluate the suggested approach towards effective monitoring
and diagnosis of business workflows. Another public dataset has been used, the CaseLaw
dataset, to validate DeepKAF when dealing with longer text and cases with more attributes.
The CaseLaw dataset represents around 22 million cases from different US states.
Further work motivated by this thesis could investigate how different deep learning models
can be used within the CBR paradigm to solve some of the chronic CBR challenges and be
of benefit to large-scale multi-dimensional enterprises.
Today, information systems are often distributed to achieve high availability and low latency.
These systems can be realized by building on a highly available database to manage the distribution of data.
However, it is well known that high availability and low latency are not compatible with strong consistency guarantees.
For application developers, the lack of strong consistency on the database layer can make it difficult to reason about their programs and ensure that applications work as intended.
We address this problem from the perspective of formal verification.
We present a specification technique, which allows specifying functional properties of the application.
In addition to data invariants, we support history properties.
These let us express relations between events, including invocations of the application API and operations on the database.
To address the verification problem, we have developed a proof technique that handles concurrency using invariants and thereby reduces the problem to sequential verification.
The underlying system semantics, technique and its soundness proof are all formalized in the interactive theorem prover Isabelle/HOL.
Additionally, we have developed a tool named Repliss which uses the proof technique to enable partially automated verification and testing of applications.
For verification, Repliss generates verification conditions via symbolic execution and then uses an SMT solver to discharge them.
This work presents a visual analytics-driven workflow for an interpretable and understandable machine learning model. The model is driven by a reverse
engineering task in automotive assembly processes. The model aims
to predict the assembly parameters leading to the given displacement field
on the geometries surface. The derived model can work on both measurement
and simulation data. The proposed approach is driven by the scientific
goals from visual analytics and interpretable artificial intelligence alike. First, a concept for systematic uncertainty monitoring, an object-oriented, virtual reference scheme (OOVRS), is developed. Afterward, the prediction task is solved via a regressive machine learning model using adversarial neural networks.
A profound model parameter study is conducted and assisted with an interactive visual analytics pipeline. Further, the effects of the learned
variance in displacement fields are analyzed in detail. Therefore a visual analytics pipeline is developed, resulting in a sensitivity benchmarking tool. This allows the testing of various segmentation approaches to lower the machine learning input dimensions. The effects of the assembly parameters are
investigated in domain space to find a suitable segmentation of the training
data set’s geometry. Therefore, a sensitivity matrix visualization is developed. Further, it is shown how this concept could directly compare results
from various segmentation methods, e.g., topological segmentation, concerning the assembly parameters and their impact on the displacement field variance. The resulting databases are still of substantial size for complex simulations with large and high-dimensional parameter spaces. Finally, the applicability of video compression techniques towards compressing visualization image databases is studied.