Refine
Year of publication
- 2024 (23) (remove)
Document Type
- Article (8)
- Conference Proceeding (4)
- Report (4)
- Master's Thesis (2)
- Bachelor Thesis (1)
- Book (1)
- Journal (Complete Issue of a Journal) (1)
- Study Thesis (1)
- Working Paper (1)
Keywords
- Accelerometer (1)
- Artificial Intelligence (1)
- Augmented Reality (1)
- Ballistocardiography (1)
- Bauwesen (1)
- Bayesian neural network (1)
- Bestandsgebäude (1)
- Bildung (1)
- Contactless Measurement (1)
- Data Model (1)
Institute
- Fakultät Architektur und Gestaltung (1)
- Fakultät Bauingenieurwesen (3)
- Fakultät Elektrotechnik und Informationstechnik (1)
- Fakultät Informatik (1)
- Fakultät Maschinenbau (1)
- Fakultät Wirtschafts-, Kultur- und Rechtswissenschaften (2)
- Institut für Angewandte Forschung - IAF (6)
- Institut für Optische Systeme - IOS (2)
- Konstanzer Institut für Prozesssteuerung - KIPS (1)
Evaluation of a Contactless Accelerometer Sensor System for Heart Rate Monitoring During Sleep
(2024)
The monitoring of a patient's heart rate (HR) is critical in the diagnosis of diseases. In the detection of sleep disorders, it also plays an important role. Several techniques have been proposed, including using sensors to record physiological signals that are automatically examined and analysed. This work aims to evaluate using a contactless HR monitoring system based on an accelerometer sensor during sleep. For this purpose, the oscillations caused by chest movements during heart contractions are recorded by an installation mounted under the bed mattress. The processing algorithm presented in this paper filters the signals and determines the HR. As a result, an average error of about 5 bpm has been documented, i.e., the system can be considered to be used for the forecasted domain.
Black-box variational inference (BBVI) is a technique to approximate the posterior of Bayesian models by optimization. Similar to MCMC, the user only needs to specify the model; then, the inference procedure is done automatically. In contrast to MCMC, BBVI scales to many observations, is faster for some applications, and can take advantage of highly optimized deep learning frameworks since it can be formulated as a minimization task. In the case of complex posteriors, however, other state-of-the-art BBVI approaches often yield unsatisfactory posterior approximations. This paper presents Bernstein flow variational inference (BF-VI), a robust and easy-to-use method flexible enough to approximate complex multivariate posteriors. BF-VI combines ideas from normalizing flows and Bernstein polynomial-based transformation models. In benchmark experiments, we compare BF-VI solutions with exact posteriors, MCMC solutions, and state-of-the-art BBVI methods, including normalizing flow-based BBVI. We show for low-dimensional models that BF-VI accurately approximates the true posterior; in higher-dimensional models, BF-VI compares favorably against other BBVI methods. Further, using BF-VI, we develop a Bayesian model for the semi-structured melanoma challenge data, combining a CNN model part for image data with an interpretable model part for tabular data, and demonstrate, for the first time, the use of BBVI in semi-structured models.
We quantify the effects of GATT/WTO membership on trade and welfare. Using an extensive database covering manufacturing trade for 186 countries over the period 1980–2016, we find that the average partial equilibrium impact of GATT/WTO membership on trade among member countries is large, positive, and significant. We contribute to the literature by estimating country-specific estimates and find them to vary widely across the countries in our sample with poorer members benefitting more. Using these estimates, we simulate the general equilibrium effects of GATT/WTO on welfare, which are sizable and heterogeneous across members. We show that countries not experiencing positive trade effects from joining GATT/WTO can still gain in terms of welfare, due to lower import prices and higher export demand.
Apnea is a sleep disorder characterized by breathing interruptions during sleep, impacting cardiorespiratory function and overall health. Traditional diagnostic methods, like polysomnography (PSG), are unobtrusive, leading to noninvasive monitoring. This study aims to develop and validate a novel sleep monitoring system using noninvasive sensor technology to estimate cardiorespiratory parameters and detect sleep apnea. We designed a seamless monitoring system integrating noncontact force-sensitive resistor sensors to collect ballistocardiogram signals associated with cardiorespiratory activity. We enhanced the sensor’s sensitivity and reduced the noise by designing a new concept of edge-measuring sensor using a hemisphere dome and mechanical hanger to distribute the force and mechanically amplify the micromovement caused by cardiac and respiration activities. In total, we deployed three edge-measuring sensors, two deployed under the thoracic and one under the abdominal regions. The system is supported with onboard signal preprocessing in multiple physical layers deployed under the mattress. We collected the data in four sleeping positions from 16 subjects and analyzed them using ensemble empirical mode decomposition (EMD) to avoid frequency mixing. We also developed an adaptive thresholding method to identify sleep apnea. The error was reduced to 3.98 and 1.43 beats/min (BPM) in heart rate (HR) and respiration estimation, respectively. The apnea was detected with an accuracy of 87%. We optimized the system such that only one edge-measuring sensor can measure the cardiorespiratory parameters. Such a reduction in the complexity and simplification of the instruction of use shows excellent potential for in-home and continuous monitoring.
Incremental one-class learning using regularized null-space training for industrial defect detection
(2024)
One-class incremental learning is a special case of class-incremental learning, where only a single novel class is incrementally added to an existing classifier instead of multiple classes. This case is relevant in industrial defect detection scenarios, where novel defects usually appear during operation. Existing rolled-out classifiers must be updated incrementally in this scenario with only a few novel examples. In addition, it is often required that the base classifier must not be altered due to approval and warranty restrictions. While simple finetuning often gives the best performance across old and new classes, it comes with the drawback of potentially losing performance on the base classes (catastrophic forgetting [1]). Simple prototype approaches [2] work without changing existing weights and perform very well when the classes are well separated but fail dramatically when not. In theory, null-space training (NSCL) [3] should retain the basis classifier entirely, as parameter updates are restricted to the null space of the network with respect to existing classes. However, as we show, this technique promotes overfitting in the case of one-class incremental learning. In our experiments, we found that unconstrained weight growth in null space is the underlying issue, leading us to propose a regularization term (R-NSCL) that penalizes the magnitude of amplification. The regularization term is added to the standard classification loss and stabilizes null-space training in the one-class scenario by counteracting overfitting. We test the method’s capabilities on two industrial datasets, namely AITEX and MVTec, and compare the performance to state-of-the-art algorithms for class-incremental learning.
AbstractSanctions encompass a wide set of policy instruments restricting cross‐border economic activities. In this paper, we study how different types of sanctions affect the export behavior of firms to the targeted countries. We combine Danish register data, including information on firm‐destination‐specific exports, with information on sanctions imposed by Denmark from the Global Sanctions Database. Our data allow us to study firms' export behavior in 62 sanctioned countries, amounting to a total of 453 country‐years with sanctions over the period 2000–2015. Methodologically, we apply a two‐stage estimation strategy to properly account for multilateral resistance terms. We find that, on average, sanctions lead to a significant reduction in firms' destination‐specific exports and a significant increase in firms' probability to exit the destination. Next, we study heterogeneity in the effects of sanctions across (i) sanction types and sanction packages, (ii) the objectives of sanctions, and (iii) countries subject to sanctions. Results confirm that the effects of sanctions on firms' export behavior vary considerably across these three dimensions.
Der Bericht dokumentiert die Aktivitäten, die ich während des Wintersemesters 2023/24 im Rahmen meines Fortbildungssemesters durchgeführt habe. Der Forschungsschwerpunkt lag dabei auf der Umsetzung eines gemeinschaftlichen Projekts zur Bewertung mündlicher Prüfungen. Zusätzlich erfolgten Aufenthalte zur Förderung der internationalen Kooperationen an Partnerhochschulen.
Diese Masterarbeit erforscht das Potenzial großer Sprachmodelle in der Bauindustrie mit einem Fokus auf digitale Transformation, Effizienzsteigerung und Nachhaltigkeit. Durch eine umfassende Literaturanalyse und qualitative Experteninterviews werden spezifische Anwendungsfälle, Herausforderungen bei der Implementierung und ethische sowie datenschutzrechtliche Überlegungen untersucht.
Die Arbeit hebt hervor, wie große Sprachmodelle die Planungsprozesse optimieren, das Risikomanagement verbessern und maßgeschneiderte Lösungen entwickeln können, um ökonomische und ökologische Vorteile zu erzielen. Zudem werden praxisorientierte Empfehlungen für eine erfolgreiche Integration dieser Technik in das Bauwesen präsentiert, die sowohl die technologische Machbarkeit als auch soziale Akzeptanz berücksichtigen.
Abschließend werden zukünftige Forschungsrichtungen aufgezeigt, die darauf abzielen, die digitale Transformation im Bauwesen unter Einbeziehung ethischer Standards und Datenschutz zu beschleunigen.
Die Ergebnisse dieser Arbeit demonstrieren das Potenzial von großen Sprachmodellen, traditionelle Bauprozesse zu revolutionieren, und betonen die Notwendigkeit einer sorgfältigen Implementierung, um die Vorteile dieser Technologie vollständig auszuschöpfen.