Publikationsserver der Ostbayerischen Technischen Hochschule Regensburg
Not a member yet
6172 research outputs found
Sort by
Machine-learning-based detection and severity estimation of drought stress in plants using hyperspectral imaging data
No full textGrowing food demand due to population growth, coupled with increasingly frequent and severe droughts caused by climate change make water increasingly scarce. To address this, accurate assessment of plant water demand is essential for precise drought treatment and water conservation. Hyperspectral imaging (HSI) captures hypercubes, a combination of spectral and spatial data and offers promising capabilities for detection of plant stresses. However, most reported approaches only use selected spectral bands or indices, neglecting the full hypercube information. This is assumed to limit the detection accuracy. To overcome these limitations, we aim to develop a measurement pipeline to generate a comprehensive dataset comprising hypercubes of plants under varying drought stress levels along with selected physiological, environmental, and illumination data. This dataset will be used to train suitable data-driven models that enable improved drought stress detection as well as the non-invasive determination of physiological parameters based on HSI data
Annika Gahl-Lang: Selbstbestimmung und Lebensqualität. Eine Photovoice-Studie mit Menschen mit intellektueller Beeinträchtigung und herausforderndem Verhalten in spezialisierten Wohnangeboten der Behindertenhilfe.
No full text
Out of Tune: Demystifying Noise-Effects on Quantum Fourier Models
No full textVariational quantum algorithms have received substantial theoretical and empirical attention. As the underlying variational quantum circuit (VQC) can be represented by Fourier series that contain an exponentially large spectrum in the number of input features, hope for quantum advantage remains. Nevertheless, it remains an open problem if and how quantum Fourier models (QFMs) can concretely outperform classical alternatives, as the eventual sources of non-classical computational power (for instance, the role of entanglement) are far from being fully understood. Likewise, hardware noise continues to pose a challenge that will persist also along the path towards fault tolerant quantum computers. In this work, we study VQCs with Fourier lenses, which provides possibilities to improve their understanding, while also illuminating and quantifying constraints and challenges. We seek to elucidate critical characteristics of QFMs under the influence of noise. Specifically, we undertake a systematic investigation into the impact of noise on the Fourier spectrum, expressibility, and entangling capability of QFMs through extensive numerical simulations and link these properties to training performance. The insights may inform more efficient utilisation of quantum hardware and support the design of tailored error mitigation and correction strategies. Decoherence imparts an expected and broad detrimental influence across all Ansätze. Nonetheless, we observe that the severity of these deleterious effects varies among different model architectures, suggesting that certain configurations may exhibit enhanced robustness to noise and show computational utility
Predict and Conquer: Navigating Algorithm Trade-Offs with Quantum Design Automation
No full textCombining quantum computers with classical compute power has become a standard means for developing algorithms and heuristics that are, eventually, supposed to beat any purely classical alternatives. While in-principle advantages for solution quality or runtime are expected for increasingly many approaches, substantial challenges remain: Non-functional properties like runtime or solution quality of many suggested approaches are not yet fully understood, and need to be explored empirically. This, in turn, makes it unclear which approach is best suited for a given problem. Accurately predicting behaviour and properties of quantum-classical algorithms opens possibilities for software abstraction layers, which in turn can automate decisionmaking for algorithm selection and parametrisation. While such techniques find frequent use in classical high-performance computing, they are still mostly absent from quantum software toolchains. In this paper, we present a methodology (accompanied by a reproducible reference implementation) to perform algorithm selection based on desirable non-functional requirements. This greatly simplifies decision-making processes for end users. Based on meta-information annotations at the source code level, our framework traces key characteristics of quantum-classical heuristics and algorithms, and uses this information to predict the most suitable approach and its parameters for given computational challenges and their non-functional requirements. As combinatorial optimisation is a very extensively studied aspect of quantumclassical systems, we perform a comprehensive case study based on numerical simulations of algorithmic approaches to implement and validate our ideas. We develop statistical models to quantify the influence of various factors on non-functional properties, and establish predictions for optimal algorithmic choices without manual user effort. We argue that our methodology generalises to problem classes beyond combinatorial optimisation, such as Hamiltonian optimisation, and lays a foundation for integrated software layers for quantum design automation
Analysing Learning Behaviour in Moodle: A Heuristic and AI-Driven Approach
No full textIn this thesis, a novel evaluation method to determine the duration of learning sessions and the completion of learning material is introduced. Screen recordings of students’ browser sessions are utilised to capture detailed, low-level learner activity in a blended learning environment.The accumulated recordings are analysed to define a continuous completion state for individual learning material, focusing on the exact time a student spends on the material and their activity during this period. The evaluation method is adapted to the type of material, such as PDF documents, quizzes, or video lectures. The results demonstrate the potential of this approach, based on analysis of two distinct time frames: an exam preparation period and a lecture period. Additionally, an AI-driven approach is outlined to further refine the chosen definition of learning material completion
Evaluating smartphone-based 3D imaging techniques for clinical application in oral and maxillofacial surgery: A comparative study with the vectra M5
No full textPURPOSE
This study aimed to clarify the applicability of smartphone-based three-dimensional (3D) surface imaging for clinical use in oral and maxillofacial surgery, comparing two smartphone-based approaches to the gold standard.
METHODS
Facial surface models (SMs) were generated for 30 volunteers (15 men, 15 women) using the Vectra M5 (Canfield Scientific, USA), the TrueDepth camera of the iPhone 14 Pro (Apple Inc., USA), and the iPhone 14 Pro with photogrammetry. Smartphone-based SMs were superimposed onto Vectra-based SMs. Linear measurements and volumetric evaluations were performed to evaluate surface-to-surface deviation. To assess inter-observer reliability, all measurements were performed independently by a second observer. Statistical analyses included Bland-Altman analyses, the Wilcoxon signed-rank test for paired samples, and Intraclass correlation coefficients.
RESULTS
Photogrammetry-based SMs exhibited an overall landmark-to-landmark deviation of M = 0.8 mm (SD = ± 0.58 mm, n = 450), while TrueDepth-based SMs displayed a deviation of M = 1.1 mm (SD = ± 0.72 mm, n = 450). The mean volumetric difference for photogrammetry-based SMs was M = 1.8 cc (SD = ± 2.12 cc, n = 90), and M = 3.1 cc (SD = ± 2.64 cc, n = 90) for TrueDepth-based SMs. When comparing the two approaches, most landmark-to-landmark measurements demonstrated 95% Bland-Altman limits of agreement (LoA) of ≤ 2 mm. Volumetric measurements revealed LoA > 2 cc. Photogrammetry-based measurements demonstrated higher inter-observer reliability for overall landmark-to-landmark deviation.
CONCLUSION
Both approaches for smartphone-based 3D surface imaging exhibit potential in capturing the face. Photogrammetry-based SMs demonstrated superior alignment and volumetric accuracy with Vectra-based SMs than TrueDepth-based SMs
Artificial intelligence improves submucosal vessel detection during third space endoscopy
No full textBackground and study aims: While artificial intelligence (AI) shows high potential in decision support for diagnostic gastrointestinal endoscopy, its role in therapeutic endoscopy remains unclear. Third space endoscopic procedures pose the risk of intraprocedural bleeding. Therefore, we aimed to develop an AI algorithm for intraprocedural blood vessel detection. Patients and Methods: Using a test dataset with 101 standardized video clips containing 200 predefined submucosal blood vessels, 19 endoscopists were evaluated for the vessel detection rate (VDR) and time (VDT) with and without support of an AI algorithm. Test subjects were grouped according to experience in ESD. Results: With AI support, endoscopists VDR increased from 56.4% [CI 54.1–58.6] to 72.4% [CI 70.3–74.4]. Endoscopists‘ VDT dropped from 6.7sec [CI 6.2-7.1] to 5.2sec [CI 4.8-5.7]. False positive (FP) readings appeared in 4.5% of frames and were marked significantly shorter than true positives (6.0sec [CI 5.28-6.70] vs. 0.7sec [CI 0.55-0.87]). Conclusions: AI improved the vessel detection rate and time of endoscopists during third space endoscopy. While these data need to be corroborated by clinical trials, AI may prove to be an invaluable tool for the improvement of endoscopic interventions
Heilmittelrichtlinie des G-BA, Keine Förderung leitliniengerechter Versorgung und Behandlung
No full textDie Deklaration der Heilmittel im Heilmittelkatalog ist zu grob, um nur näherungsweise die Leitlinien korrekt abzubilden. Damit liegt es in der Verantwortung der Physiotherapeutinnen und Physiotherapeuten, leitliniengerechte therapeutische Maßnahmen durchzuführen. Hinsichtlich der Wahl der Intervention zeigte die zugrundeliegende Analyse, dass unabhängig davon, ob eine Verordnung leitliniengerecht ist oder nicht, die Behandlung sowohl leitliniengerecht als auch nicht leitliniengerecht stattfinden kann
When Work (Re)releases Energy: Five steps towards regenerative people management
No full textIn this article, we explore the innovative concept of regenerative work, which positions itself as a response to the challenges of the modern working world. At the core of our discussion are five phases of work design: from conventional (degenerative) work, often perceived as burdensome, through employee-oriented work, human-centeredness, restorative work, and finally to regenerative work, which can return energy to employees (and the environment). We examine the influence that leaders, employees, and HR professionals can have on designing working conditions, jobs, and teams. Practical examples illustrate how companies can achieve a positive energy balance for employees and teams through energy analysis. We emphasize that regenerative work is not a distant ideal but an achievable goal that can be realized through the conscious design of the workplace