6949 research outputs found
Sort by
Machine Learning Approaches to Predicting Energy Expenditure in Preschool Children: Insights from Accelerometry, Gyroscope Data, and Cross-National Validation
No full textAs highlighted by the World Health Organization, physical inactivity has been recognized as a public health crisis affecting not only adults, but also children and adolescents. To address this alarming trend, it is essential to establish a reliable and robust measure of physical activity (PA) to better understand its underlying determinants. For this purpose, wearable sensors are often used, offering an indirect measure to predict/estimate the energy expenditure (EE) of PA. With the adoption of wearable sensors, numerous researchers are implementing more sophisticated machine learning approaches in their analyses that are better equipped to model complex relationships. The overarching aim of this doctoral research was to develop and refine machine learning models to predict the EE of preschool children. Across four studies, key aspects of the modeling process were explored, including model selection, preprocessing strategies, feature selection, sensor integration, the influence of metabolic equivalent (METs) definitions, and external validation. Two calibration datasets, one consisting of Canadian preschool children and the other of German preschool children, were used to develop and evaluate models using accelerometers, gyroscopes, and portable metabolic units during semi-structured activity protocols. The findings indicated that while deep learning models achieved the lowest error on the training datasets, feature-based models demonstrated superior performance in external validation. Furthermore, preprocessing techniques, specifically frequency-based filtering, and the inclusion of frequency-domain features and participant characteristics (age, sex, height, and weight) contributed to reduced prediction error. When comparing models built using gyroscope data, accelerometer data, and a combination of both, the dual-sensor models consistently outperformed single-sensor models, yielding lower error rates. Finally, after identifying the optimal feature set, the models were applied to a large cohort of Canadian children to generate and compare PA estimates based on different METs definitions. Notably, it was found that measuring the resting period, rather than estimating it using predictive approaches, resulted in higher estimates of sedentary time and lower estimates of overall PA. Collectively, this thesis advances the field of movement behavior research by contributing validated machine learning models for estimating EE in preschool children and addressing key methodological questions relevant to this domain
Vom Spielmobil zum Lernmobil. Perspektiven für temporäre Lern-Erlebnisstationen. Tagungsdokumentation
No full textDer vorliegende Tagungsband dokumentiert die Beiträge des Fachtags „Vom Spielmobil zum Lernmobil“ im Haus der Wissenschaft in Bremen im September 2025. Vorgestellt werden konzeptionelle Ansätze und Praxisbeispiele aus den Feldern Spielmobil, Outreach von Museen, Wissenschaftskommunikation, Lifestyle-Kampagnen und Umweltbildung. Das Projekt MOBILE, in dessen Rahmen die Tagung stattfand, zielt auf eine Beschreibung übergreifender didaktischer Modelle für temporäre Lern-Erlebnis-Arrangements und die Analyse von Perspektiven für eine transformative Bildung
On learning Hamiltonian systems using machine learning with focus on system symmetry preservation
No full textSymmetries are fundamental properties of many natural phenomena and structures. In physics, they enable the formulation of conservation laws such as the conservation of energy, momentum, and angular momentum, which simplify the mathematical description and modeling of physical systems.
In the field of machine learning, particularly in neural networks, symmetries can be utilized to improve the efficiency and accuracy of models. By incorporating symmetries into the architecture and training of neural networks, these models can become more robust and generalizable. While neural networks are widely used in areas such as image analysis, speech recognition, and text processing, the learning of nonlinear dynamical systems that consider physical laws is less explored.
To address this gap, Hamiltonian Neural Networks (HNNs) have been developed, specifically designed to learn dynamical systems while preserving the Hamiltonian structure. This approach ensures that the symplecticity of the system is maintained during data-driven modeling. However, preserving additional symmetry properties requires extra attention.
This work proposes two methods to improve HNNs by considering system symmetries. The first part integrates known symmetry information during training through the introduction of symmetry-preserving extensions to the Hamiltonian network architecture. Discrete symmetries, such as periodicity, and continuous symmetries, like translational or rotational invariance, are discussed. The second part focuses on identifying system symmetries alongside learning the Hamiltonian function. The proposed method extends the structure of HNNs with a Lie algebra framework to recognize and embed symmetries into the neural network. This allows for the simultaneous learning of the symmetry group and the total energy of the system.
This work examines the results from simulations of various physical systems to demonstrate the effectiveness of the model approaches. Illustrative examples include the simple pendulum, the cart-pendulum system, and the two-body problem in astrodynamics. The results show that incorporating symmetry into the learning process results in more robust and accurate predictions
Begründung und Erprobung eines Verfahrens zur Kompetenzfeststellung für den Einsatz an berufsbildenden Schulen in Bremen
No full textDiese Dissertation entwickelt und erprobt mittels Design-Based Research ein Kompetenzfeststellungsverfahren für An- und Ungelernte an berufsbildenden Schulen zur Vorbereitung auf die Externenprüfung. In Deutschland fehlen systematische Verfahren zur Validierung informell erworbener Kompetenzen. An- und Ungelernte, die 2020 in Bremen 66% aller Arbeitslosen ausmachten, besitzen oft umfangreiche Berufserfahrung, haben jedoch ohne formalen Abschluss schlechtere Arbeitsmarktchancen. Die Externenprüfung ermöglicht den Erwerb eines Berufsabschlusses, doch Erfahrungen aus den Bremer Modellprojekten NQVorE und LWmBB (2013-2017) zeigen, dass es an geeigneten Kompetenzfeststellungsverfahren und qualifizierten Anbietern mangelt. Literaturrecherchen in FIS- und BIBB-Datenbanken belegen, dass die systematische Einbindung berufsbildender Schulen bislang nicht wissenschaftlich untersucht wurde. Das Forschungsziel besteht in der Entwicklung und Erprobung eines Verfahrens, mit dem Lehrkräfte an berufsbildenden Schulen, die über Fachkenntnisse, Prüfungsvertrautheit und pädagogische Kompetenz verfügen, den Status quo der beruflichen Handlungskompetenz von An- und Ungelernten erfassen können. Die Arbeit untersucht erstens, welche Verfahrensschritte und Instrumente sich für die Kompetenzerfassung im Hinblick auf die Externenprüfung eignen, und zweitens, welche strukturellen Rahmenbedingungen für eine nachhaltige, flächendeckende Implementierung an berufsbildenden Schulen erfüllt sein müssen. Die Dissertation leistet damit einen Beitrag zur Professionalisierung der Kompetenzfeststellung in Deutschland und zur Weiterentwicklung berufsbildender Schulen als Kompetenzzentren
Theoretical modelling of semiconductor nanolasers: On the influence of many-body effects on the optical properties of semiconductor nanostructures
No full textThe rapid rise in global data centre energy consumption poses a critical challenge in the age of digitalisation and artificial intelligence. A promising strategy for improving energy efficiency involves the use of on-chip integrated photonic components to enable high-speed and low-loss optical links and photonic computing platforms. Realising this vision requires integrable laser sources on the nanoscale. This dissertation develops a comprehensive theoretical framework for semiconductor nanolasers, with a particular emphasis on the influence of quantum many-body effects on the optical properties of low-dimensional gain media.
This work is structured to consist of three interdependent chapters.
The first chapter investigates a metallic-cavity nanolaser incorporating multiple quantum wells. A fully quantised electromagnetic field formalism is employed and Quantum Laser Equations are derived on the quadruplet-level. These equations enable access to key observables including the input-output characteristics, coherence time, and second-order correlation function at zero delay. Theoretical predictions are shown to be in substantial agreement with experimental data from a silver-coated MQW nanolaser device. A spectral lineshape anomaly is identified: A transition from a Lorentzian to a Gaussian emission profile at the lasing threshold, attributed to intrinsic non-linear effects and partial mode-locking in the open cavity system.
The second chapter establishes a microscopic model of monolayer molybdenum disulfide, a transition metal dichalcogenide, as a gain material. Based on a tight-binding approach, material-specific dipole and Coulomb matrix elements are derived and energy renormalisations via the screened-exchange Coulomb-hole approximation are introduced to account for the influence of screening due to the presence of excited carriers. Absorption spectra are generated using the Semiconductor Bloch Equations, demonstrating the importance of screening effects in shaping the optical response. The impact of the Brillouin zone sampling density on computational efficiency and spectral accuracy is systematically analysed and a minimum carrier density for the expectation of gain is quantified.
The third and final chapter combines insights gained in the previous two and proposes a theoretical nanolaser device consisting of a molybdenum disulfide monolayer integrated with a photonic crystal cavity. A material-oriented doublet-level formulation of the Quantum Laser Equations is developed to manage the multiscale dynamics, spanning femtosecond Coulomb processes to nanosecond lasing behaviour. The theory predicts electron-hole-plasma-based lasing at room temperature, characterised by an S-shaped input-output curve, hole burning at the K- and K′-valley, and spectral clamping. These features are indicative of lasing driven by plasma gain at high carrier densities and emerge naturally from the material-oriented quantum-optical treatment of the device.
Together, the three chapters establish a consistent, predictive, and microscopically realistic theoretical framework for semiconductor nanolasers. The developed models combine quantum-optical treatments with device-level properties focused on the utilised gain media, enabling a detailed understanding of the interplay between many-body physics and stimulated emission in nanoscale systems. These results contribute to the design of next-generation nanophotonic light sources and offer a pathway towards energy-efficient, on-chip integrated lasers suitable for future optical links and photonic computing platforms
Automatic design of norm-optimal structured PID-controllers for nonlinear systems
No full textDiese Arbeit soll die Verwendung von mathematisch optimalen Reglerentwurfsmethoden, insbesondere dem -Ansatz, an die Gegenheiten in der industriellen Praxis anpassen. So soll die Verwendung dieser mächtigen Werkzeuge für Prozesse, die nichtlinear sind und mit übersichtlichen Reglern gehandhabt werden sollen, ermöglicht werden, was in vielen Fällen Effizienssteigerungen ohne größeren Aufwand ermöglicht. Es wird ein automatisiertes Verfahren entwickelt, das basierend auf einem Streckenmodell selbstständig erkennt, welche Form der Nichtlinearität hier behandelt werden muss, und entwirft anschließend ein von einem existierenden linearen Regler ausgehendes nichtlineares Gesamtregelungssystem. Es sollen Kennlinien, Umschaltpunkte, Totzeiten und Split-Range-Probleme selbstständig erkannt und gelöst werden. Dabei wird darauf geachtet, ein möglichst einfaches Regelungssystem mit ausreichender Regelungsqualität zu entwerfen, und nicht zusätzliche Signalpfade, Parameter und Teilsysteme einzuführen, wenn diese keine signifikante Verbesserung des Regelungsverhaltens bieten
Generative KI als Spiegelraum: Zur Rolle dialogischer Systeme in der transformativen Bildung für nachhaltige Entwicklung (Verlagsversion)
No full textDer Beitrag untersucht die dialogischen Potenziale generativer KI-Systeme wie ChatGPT im Kontext transformativer Bildung für nachhaltige Entwicklung (BNE). Ausgehend von der These, dass KI kein neutrales Werkzeug, sondern ein kultureller Spiegelraum ist, analysiert der Text, wie kollektives Prompting zu einer neuen Kulturtechnik werden kann. Die Nutzer:innen treten dabei nicht nur als Konsumierende, sondern als Mitgestaltende epistemischer Prozesse auf. Anhand theoretischer Perspektiven (u. a. Haraway, Barad, Butler) und eines visionären Bildungskonzepts – des »KI-Flashmob für Nachhaltigkeit« – wird aufgezeigt, wie sprachliche Interaktion mit KI zur reflexiven und politischen Praxis werden kann. Der Beitrag plädiert für eine aktive Aneignung dieser neuen Form digitaler Kommunikation, in der Sprache als Medium gesellschaftlicher Gestaltung verstanden wird für eine demokratische und nachhaltige KI-Kultur von morgen.21523
Permafrost carbon mobilization into the Laptev Sea since the last deglaciation
No full textWarming in the Arctic has been observed to exceed the global average by more than four times, which can accelerate circumarctic permafrost thaw, leading to the release of carbon into the atmospheric carbon pool. Once thawed, circumarctic permafrost is transported away from the land and eventually deposited in the marginal seas of the Arctic Ocean. During these processes, carbon can be released through microbial decomposition. Several questions related to this process remain insufficiently understood: Which environmental conditions favor permafrost thawing, and through what mechanisms? How vulnerable is organic matter released from thawing permafrost to decomposition? What is the key step in carbon release during permafrost mobilization? To address these questions, this thesis investigates the dynamics of mobilized terrestrial permafrost deposited on the Laptev Sea shelf, where terrestrial sources comprise an approximately equal mix of Pleistocene and Holocene permafrost deposits.
The first manuscript (Chapter 3) evaluated the organic matter characteristics of terrestrial Pleistocene and Holocene permafrost by analyzing their bulk organic composition, isotopic signatures, and thermal reactivity properties. Terrestrial Pleistocene permafrost was found to exhibit lower reactivity and higher degradation status compared to Holocene permafrost. Surface and downcore sediments (PS51/154 and PS51/159) from the Laptev Sea shelf were subsequently analyzed for the same parameters, and lower reactivity of the Laptev Sea surface sediment was found in the region with higher input from mobilized terrestrial Pleistocene permafrost. A decrease in organic matter reactivity was observed with the increase in transport distance from the shoreline, indicating that degradation during transport is a major control for the organic matter reactivity of Laptev Sea surface sediments. A drastic decrease in organic matter reactivity near the coast was observed, indicating the nearshore region as a hotspot of rapid organic matter degradation.
The second manuscript (Chapter 4) aimed to identify potential environmental causes of rapid terrestrial permafrost thawing by examining sediment core records from the western Laptev Sea (PS51/154 and PS51/159) spanning the last deglaciation. Three periods of high mass accumulation rates of terrestrial biomarkers were identified and linked to distinct environmental conditions. Comparisons with published records from other Arctic marginal seas revealed that enhanced coastal erosion driven by accelerated sea-level rise during meltwater pulse 1A (mwp-1A) was a widespread phenomenon across different marginal seas in the Arctic. Additional periods in rapid terrestrial biomarker mobilization were identified asynchronously between regions and attributed to various region-specific factors, such as freshwater flooding events, enhanced inland warming, and prolonged sea ice-free seasons.
The third manuscript (Chapter 5) further traced the sources of mobilized terrestrial organic matter during these periods of rapid terrestrial biomarker mobilization. Age-depth models for the two cores were first refined using an updated estimation of the local marine reservoir age, which was determined by aligning authigenic 10Be/9Be ratio variations in marine sediment and 10Be fluxes in ice cores. Pre-depositional ages of terrigenous biomarkers, including C28:0 fatty acids, C23 + C25 n-alkanes, and C29 + C31 n-alkanes, were analyzed to assess terrestrial source changes across different periods. The pre-depositional ages of C28:0 fatty acids indicated a higher contribution of aged material during periods of rapid terrestrial biomarker mobilization, whereas inputs from younger permafrost increased during intervals of enhanced river discharge. In contrast, the pre-depositional ages of C23 + C25 and C29 + C31 n-alkanes suggested a consistently predominant input from aged terrestrial materials.
Overall, this thesis provides a multifaceted investigation of terrestrial organic matter transport into the Laptev Sea. It includes differences in reactivity between young and aged terrestrial permafrost, variations in organic matter reactivity during transport and post-burial processes, environmental factors fostering rapid terrestrial permafrost mobilization, and the sources of mobilized materials. These findings contribute to a better understanding of permafrost-derived carbon dynamics in the Arctic and their implications for global carbon cycling
New invariants in topological data analysis and their applications in material and biological sciences
No full textTopological data analysis has proven to be a versatile and powerful enrichment for data analysis of diverse selections of data in all types of quantitative analyses. Strategies include incorporating descriptors and characterizing topological properties of datasets ranging from scalar fields, point clouds, graphs, or any other scientific dataset. Furthermore, more data-oriented combinatorial constructions can be used from which the topological properties are taken. In the following thesis, the focus is on three applications: A new combinatorial structure for describing phylogenetic networks as well as general filtered spaces, surface classification via roughness computations and its comparison to topological invariants on its scalar field, and an application for the spatial analysis of microscopic images.
More specifically, the cliquegram and facegram models are established for phylogenetic models and their theoretical properties investigated, as well as their computational complexity, and algorithms for their efficient computation proposed. In addition, an approach to classify surfaces based on their roughness is presented and persistent homology is employed to extract multiscale topological features from surface data and integrate these features into machine learning models. Finally, embryonic and neural stem cells are distinguished after protein staining results in distinct spatial structures of their 3D microscopic images which are captured using cubical persistent homology and analyzed.
These different approaches demonstrate the benefits of combining new topological and combinatorial methods with existing ones, providing a more comprehensive toolkit for complex network analysis. Code for reproducing and reusing the techniques is provided
Wege in Ausbildung und Ausbildungslosigkeit: Qualitative Zusatzbefragung. Studienreport.
No full textDas Forschungsprojekt „Wege in Ausbildung und Ausbildungslosigkeit“ stellt eine qualitative Zusatzbefragung zum quantitativen Übergangspanel des Deutschen Jugendinstitut (DJI) dar. Die jugendlichen Absolventen der Hauptschule in Deutschland wurden im Übergangspanels mehrmals zu ihrem weiteren Werdegang insbesondere in Bezug auf (Aus-)Bildungswege befragt. Mittels eines theoretischen Sampling wurden zwei Teilstichproben mit jungen Erwachsenen ohne Ausbildung und jungen Erwachsenen mit begonnener oder abgeschlossener Ausbildung generiert. Insgesamt wurden 2011 damit 56 Jugendliche mittels eines Leitfadeninterviewt befragt. Von besonderem Interesse sind dabei „weiche“ Faktoren und ihr Einfluss für den erfolgreichen oder nicht erfolgreichen Übergang von schulischer Bildung in eine Ausbildung. Dazu zählen unter anderem die Rolle von anderen Personen, beispielsweise Lehrer*innen, Eltern oder Peers, im Übergang von Schule in Ausbildung, (Eigen-)Motivation der Interviewten und biographische Erlebnisse. Zentrale Erkenntnisse der Untersuchung ist das Herausarbeiten von vier Erfahrungsebenen, die für Übergangsverläufe der befragten jungen Erwachsenen eine zentrale Bedeutung einnehmen: Motivation, Agency, kritische biografische Ereignisse und soziale Interaktion. Auf Basis der Ergebnisse können verschiedene Verlaufstypen identifiziert und wichtige Erkenntnisse gewonnen werden. Die Transkripte aller 56 geführten Interviews sowie zugehörige Memos stehen bei Qualiservice für die wissenschaftliche Nachnutzung zur Verfügung