openHSU (HSU Hamburg)
Not a member yet
7291 research outputs found
Sort by
Gasoline prices and presidential approval ratings of the United States
No full textThis article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (https://creativecommons.org/licenses/by-nc/4.0/).We use random forests, a machine-learning technique, to formally examine the link between real gasoline prices and presidential approval ratings of the United States (US). Random forests make it possible to study this link in a completely data-driven way, such that nonlinearities in the data can easily be detected and a large number of control variables, in line with the extant literature, can be considered. Our empirical findings show that the link between real gasoline prices and the presidential approval ratings is indeed nonlinear, and that the former even has predictive value in an out-of-sample exercise for the latter. We argue that our findings are in line with the so-called pocketbook mechanism, which stipulates that the presidential approval ratings depend on gasoline prices because the latter have sizable impact on personal economic situations of voters.Vo
Early warning of hypoglycemia via sensor-agnostic machine learning: a clinical app design for type 1 diabetes
No full textType 1 diabetes is an incurable chronic disease with an increasing incidence, especially in highly developed countries. The main challenge in living with type 1 diabetes is blood glucose management, which requires lifelong insulin treatment, dietary adjustments as well as placing restrictions on physical activities. Incorrect insulin dosages or errors in diet and physical activity can lead blood glucose levels to drop too far, causing hypoglycemia, a severe health risk. In this work we introduce our framework and proof of concept app DiApp, to assist patients with diabetes in managing their blood sugar levels. It does this by merging data from blood glucose and heart rate sensors and using this data to offer an early detection of hypoglycemia. If hypoglycemia is detected to occur in the near future, action recommendations, such as resting and eating carbohydrate-rich food, are suggested as appropriate. This developed proof-of-concept app, DiApp, uses the Apple Health interface to collect sensor data, as this interface is implemented by a wide variety of commercially available sensors. While this implementation still has some limitations, it is capable of evaluating the data from different sensors, and alerting users if hypoglycemia is predicted to occur. In the future, we aim to implement specific glucose and heart rate sensors interfaces and conduct a clinical trial with real patients to investigate how effective the app is at reducing unhealthy glucose levels.Vo
A deep learning-based approach for the detection of cucumber diseases
No full textThis is an open access article distributed under the terms of the Creative Commons Attribution License CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).Cucumbers play a significant role as a greenhouse crop globally. In numerous countries, they are fundamental to dietary practices, contributing significantly to the nutritional patterns of various populations. Due to unfavorable environmental conditions, they are highly vulnerable to various diseases. Therefore the accurate detection of cucumber diseases is essential for maintaining crop quality and ensuring food security. Traditional methods, reliant on human inspection, are prone to errors, especially in the early stages of disease progression. Based on a VGG19 architecture, this paper uses an innovative transfer learning approach for detecting and classifying cucumber diseases, showing the applicability of artificial intelligence in this area. The model effectively distinguishes between healthy and diseased cucumber images, including Anthracnose, Bacterial Wilt, Belly Rot, Downy Mildew, Fresh Cucumber, Fresh Leaf, Pythium Fruit Rot, and Gummy Stem Blight. Using this novel approach, a balanced accuracy of 97.66% on unseen test data is achieved, compared to a balanced accuracy of 93.87% obtained with the conventional transfer learning approach, where fine-tuning is employed. This result sets a new benchmark within the dataset, highlighting the potential of deep learning techniques in agricultural disease detection. By enabling early disease diagnosis and informed agricultural management, this research contributes to enhancing crop productivity and sustainability.Vo
Applications of numerical simulations for impact echo and ultrasonic testing on concrete structures
No full textNon-destructive testing methods are effective tools to assess the condition of concrete structures without causing additional damage. As concrete is the most frequently used construction material in the world and all concrete structures are degrading over time, such testing methods are a crucial aspect of inspections of public infrastructure, which aim at ensuring public safety. Many non-destructive testing methods, such as impact echo and ultrasonic testing, utilize waves as the primary information carrier. These waves are reflected at all material interfaces, which poses some unique challenges for when the methods are applied to concrete with its heterogeneous mesostructure consisting of aggregates, cement matrix, and pores. A numerical simulation code based on the elastodynamic finite integration technique was implemented to investigate wave propagation for these two non-destructive testing methods. A heterogeneous concrete model was used to increase the realism of the simulations. This thesis investigates four different applications of the simulation code related to noise simulation during impact echo and ultrasonic testing.
During ultrasonic testing, strong noise amplitudes superpose with reflection signals originating from defects inside the inspected structure and might cause these defects to be overlooked during inspections. It is, therefore, necessary to investigate the capabilities of ultrasonic testing inspection systems. Typically, such investigations are performed by using probability of detection analyses. However, such analyses require data from defects, that can only be detected under certain boundary conditions. Under different circumstances these defects might get overlooked. Therefore, numerical simulations are used to emulate real-world inspection results and estimate the detectability of defects with varying sizes and depths. These results are then used to design a concrete specimen with artificial defects implemented inside it. Measurement results showed good qualitative agreement to the simulations. Approximately 68% of the defects implemented into the specimen could be detected, proving the high degree of realism in the simulations and enabling probability of detection analysis for ultrasonic testing data from a concrete structure.
The second application of numerical simulations in this thesis aims to investigate physical and numerical factors influencing the outcome of numerical impact echo simulations. The goal of impact echo is to measure the frequency of the zerogroup-velocity S1 Lamb wave mode. Numerical simulations of this non-destructive testing method are rarely performed. As the wavelengths during impact echo tests are of similar size as the thickness of a structure, many simulations approximate concrete as a homogeneous medium. However, it is known that pores and aggregates affect the simulation outcome. This study investigates the influence of material heterogeneity on wave speeds and the S1 Lamb wave frequency. Additionally, material attenuation and the frequency spectrum of the source function are investigated to increase the realism of simulations. Lastly, impact echo measurements are recreated numerically with 2D and 3D simulations. Simulation and measurement results are compared in time and frequency domain. It was found that 3D simulations show a high degree of realism, whereas 2D simulations only capture some of the features found in the measurements.Vo
Temporale Ordnungen und zeittheoretische Betrachtungen wissenschaftlicher Weiterbildung
No full textDieses Werk steht unter der Lizenz Creative Commons Namensnennung - Weitergabe unter gleichen Bedingungen 4.0 International (https://creativecommons.org/licenses/by-sa/4.0/).Der Beitrag führt ein in zeittheoretisches Nachdenken über Bildung und Lernen und sondiert entsprechende Forschungszugänge und Studien für das Feld der wissenschaftlichen Weiterbildung. Wie wird Zeit für das Lernen genutzt und wie wird sie auf vielfältige, oft widersprüchliche Weise erlebt? Kann Lernen zeitliche Gegenbewegungen wie Entschleunigung oder Muße hervorrufen?
Wir wenden uns diesen Fragen zunächst in einer grundlagentheoretischen Annäherung zu und zeichnen die historische Entwicklung temporaltheoretischer Perspektiven im wissenschaftlichen Diskurs unterschiedlicher Fachdisziplinen nach, um sie anschließend auf das Lebenslange Lernen zu beziehen. Anhand von aktuellen Forschungsansätzen und Studien mit Bezug zur (wissenschaftlichen) Weiterbildung werden empirische Einsichten in Zeitphänomene eröffnet und Forschungszugänge und -desiderate freigelegt. Abschließend resümieren wir, welche Phänomene wissenschaftlicher Weiterbildung in einer zeittheoretischen Perspektive in den Blick rücken und in welcher Weise weitergehende temporalanalytische Untersuchungen das Forschungsfeld der wissenschaftlichen Weiterbildung anreichern können.Vo
A knowledge-guided hybrid learning framework for semantic constraint integration in time series models
No full textUse permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0, https://creativecommons.org/licenses/by/4.0).Current time series models often operate solely on sensor data, lacking the contextual understanding that domain knowledge provides. This limitation particularly exists in domains like maritime operations or medical monitoring, where sensor data are often noisy, incomplete, or ambiguous. To address this gap, this doctoral research proposes a hybrid learning framework that integrates semantic knowledge from ontologies, domain texts, and expert-defined rules into the modeling process as formal constraints. The framework comprises three main building blocks: (1) learning joint representations from heterogeneous sources such as time series, structured knowledge, and unstructured text; (2) extracting and formalizing semantic knowledge into symbolic or functional constraints; and (3) fusing these components into a hybrid framework, where formal constraints complement machine-learned patterns. Initial work has been conducted in the maritime domain and will be extended to medical datasets for cross-domain evaluation.Vo
Whitepaper Resonanzstabilitäts-Bewertung nach dem impedanzbasierten Stabilitätskriterium
No full textMit dem Ausbau der erneuerbaren Energien und der Netzinfrastruktur soll die elektrische Energieversorgung zukunftssicher gestaltet werden. Diese anspruchsvolle Aufgabe liegt zu großen Teilen bei den Verteilnetzbetreibern, die unter anderem dafür sorgen müssen, dass stromrichter-gekoppelte Anlagen zukünftig auch auf Resonanzstabilität geprüft werden. Dieses Whitepaper bietet einen praxisorientierten Einstieg in das Thema, orientiert am impedanzbasierten Stabilitätskriterium und Erkenntnissen der Netz- und Anlagenimpedanzmessung. Die Ergebnisse aus Messkampagnen zur Bestimmung der Netz- und Wechselrichterimpedanzen werden im Kontext des Stabilitätskriteriums interpretiert. Daraus werden grundlegende Thesen zur Resonanzstabilität abgeleitet, die eine Einschätzung der Prüfnotwendigkeit beim Netzanschluss ermöglichen. Der Beitrag schließt mit einem Vorschlag für ein Prüfschema, das auf den gewonnenen Erkenntnissen basiert.Vo
Pedagogical attachments and virtual reality
No full textThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/
licenses/by/4.0/.This paper explores what comes to matter pedagogically when introducing Virtual Reality (VR) in Vocational Education and Training (VET) schools. In spite of the general understanding that digital technologies rarely work as anticipated, research exploring how VR unfolds in schools and how it gives way to specific pedagogical forms of acting, thinking, and relating to others, is scarcer. Addressing this gap, we explicate our ethnographic fieldwork in one Belgian school through sensitivities derived from Science and Technology Studies (STS). Specifically, thinking with the concept of ‘attachment’ (and its corollary ‘detachment’) to indicate how educational ways of being are continuously in the making in classroom practices, our fieldwork shows how different types of attachments (i.e. communizing, instancing, modulating, conditioning), that emerge through concrete and heterogeneous acts, relate to different forms of ‘immersion’. Conclusively, the paper discusses the need to further unpack how technologies always come to matter pedagogically in specific ways and argues that an appropriate terminology is necessary to articulate this. To this extent, we propose the concepts of ‘pedagogical attachments’ and immersion to enunciate the pedagogical specificity of what is at stake when introducing digital technologies in the classroom.Vo