Publikationsserver der Ostbayerischen Technischen Hochschule Regensburg
Not a member yet
6172 research outputs found
Sort by
Ein quantenlogisch motivierter Ansatz zur Verarbeitung von Äußerungs‐Bedeutungspaaren
No full textWir repräsentieren eine Bedeutung als Liste von Mustersignalen, und unser Ziel ist es, ein weiteres ankommendes Signal damit zu vergleichen. Die Quantenlogik motiviert die Verwendung von Orthogonalprojektoren, um die gesuchte Ähnlichkeit als Projektionswahrscheinlichkeit darzustellen. Die Ergebnisse des quantenlogischen Verfahrens hängen davon ab, in welcher Weise die Signale vorverarbeitet werden. In diesem Aufsatz untersuchen und diskutieren wir vier verschiedene Möglichkeiten der Vorverarbeitung
Octra Backend - Eine skalierbare Infrastruktur für Transkriptionsprojekte
No full textOctra Backend ist eine portable web-basierte Infrastruktur für Transkriptionsprojekte, die lokal im Feld oder geschützten Bereichen, im begrenzten Intranet oder weltweit erreichbar im Internet eingesetzt werden kann. Entwicklungsziele waren die Gewährleistung möglichst hoher Sicherheitsanforderungen, eine gute Skalierbarkeit sowie eine einfache Installation auch ohne Administratorrechte. Octra Backend ist in Node.js implementiert und für MacOS, Windows und Linux verfügbar
Transient Response of Macroscopic Deformation of Magnetoactive Elastomeric Cylinders in Uniform Magnetic Fields
No full textSignificant deformations of bodies made from compliant magnetoactive elastomers (MAE) in magnetic fields make these materials promising for applications in magnetically controlled actuators for soft robotics. Reported experimental research in this context was devoted to the behaviour in the quasi-static magnetic field, but the transient dynamics are of great practical importance. This paper presents an experimental study of the transient response of apparent longitudinal and transverse strains of a family of isotropic and anisotropic MAE cylinders with six different aspect ratios in time-varying uniform magnetic fields. The time dependence of the magnetic field has a trapezoidal form, where the rate of both legs is varied between 52 and 757 kA/(s·m) and the maximum magnetic field takes three values between 153 and 505 kA/m. It is proposed to introduce four characteristic times: two for the delay of the transient response during increasing and decreasing magnetic field, as well as two for rise and fall times. To facilitate the comparison between different magnetic field rates, these characteristic times are further normalized on the rise time of the magnetic field ramp. The dependence of the normalized characteristic times on the aspect ratio, the magnetic field slew rate, maximum magnetic field values, initial internal structure (isotropic versus anisotropic specimens) and weight fraction of the soft-magnetic filler are obtained and discussed in detail. The normalized magnetostrictive hysteresis loop is introduced, and used to explain why the normalized delay times vary with changing experimental parameters
EYE TRACKING AS TECHNOLOGY IN EDUCATION: FURTHER INVESTIGATION OF DATA QUALITY AND IMPROVEMENTS
No full textEye tracking serves as a powerful tool across a variety of empirical research areas: From usability research over cognitive research to educational research and applications in classrooms. However, data noise in eye tracking data poses a challenge to researchers and educators, as it leads to gaze positions being measured imprecisely under unfavorable conditions.
In our previous study, we systematically investigated factors that influence data quality and are easily controllable in a classroom or laboratory environment, such as illumination, sampling frequency, and head orientation. However, no recommendations regarding the light source and light orientation could be provided, as these influences could not be analyzed in sufficient detail. Yet, a further examination of these factors, eliminating human influences by using an artificial head, revealed significant differences between individual settings.
Hence, in this empirical study of eye tracking as an educational technology, we delve deeper into examining the impact of both light source and light orientation on data quality. This is investigated with an artificial head together with the Tobii Pro Spectrum eye tracking device.
To measure data quality, we use the metrics precision and standard deviation as indicators of data noise. The obtained results derive practical advice for educators and researchers, such as not to illuminate the subject from the rear, in order to gather useful data for research and future classroom applications.
Thereby, this study serves as a complement to our previous research, answering open questions regarding best practices for researchers and educators when using eye trackers. It aims to provide valuable insights into producing data of the highest quality possible when using eye trackers, both in laboratory settings and in future classrooms applications
Short and sweet: multiple mini case studies as a form of rigorous case study research
No full textCase study research is one of the most widely used research methods in Information Systems (IS). In recent years, an increasing number of publications have used case studies with few sources of evidence, such as single interviews per case. While there is much methodological guidance on rigorously conducting multiple case studies, it remains unclear how researchers can achieve an acceptable level of rigour for this emerging type of multiple case study with few sources of evidence, i.e., multiple mini case studies. In this context, we synthesise methodological guidance for multiple case study research from a cross-disciplinary perspective to develop an analytical framework. Furthermore, we calibrate this analytical framework to multiple mini case studies by reviewing previous IS publications that use multiple mini case studies to provide guidelines to conduct multiple mini case studies rigorously. We also offer a conceptual definition of multiple mini case studies, distinguish them from other research approaches, and position multiple mini case studies as a pragmatic and rigorous approach to research emerging and innovative phenomena in IS
A data-driven approach for the partial reconstruction of individual human molar teeth using generative deep learning
No full textBackground and objectiveDue to the high prevalence of dental caries, fixed dental restorations are regularly required to restore compromised teeth or replace missing teeth while retaining function and aesthetic appearance. The fabrication of dental restorations, however, remains challenging due to the complexity of the human masticatory system as well as the unique morphology of each individual dentition. Adaptation and reworking are frequently required during the insertion of fixed dental prostheses (FDPs), which increase cost and treatment time. This article proposes a data-driven approach for the partial reconstruction of occlusal surfaces based on a data set that comprises 92 3D mesh files of full dental crown restorations.MethodsA Generative Adversarial Network (GAN) is considered for the given task in view of its ability to represent extensive data sets in an unsupervised manner with a wide variety of applications. Having demonstrated good capabilities in terms of image quality and training stability, StyleGAN-2 has been chosen as the main network for generating the occlusal surfaces. A 2D projection method is proposed in order to generate 2D representations of the provided 3D tooth data set for integration with the StyleGAN architecture. The reconstruction capabilities of the trained network are demonstrated by means of 4 common inlay types using a Bayesian Image Reconstruction method. This involves pre-processing the data in order to extract the necessary information of the tooth preparations required for the used method as well as the modification of the initial reconstruction loss.ResultsThe reconstruction process yields satisfactory visual and quantitative results for all preparations with a root mean square error (RMSE) ranging from 0.02 mm to 0.18 mm. When compared against a clinical procedure for CAD inlay fabrication, the group of dentists preferred the GAN-based restorations for 3 of the total 4 inlay geometries.ConclusionsThis article shows the effectiveness of the StyleGAN architecture with a downstream optimization process for the reconstruction of 4 different inlay geometries. The independence of the reconstruction process and the initial training of the GAN enables the application of the method for arbitrary inlay geometries without time-consuming retraining of the GAN
Layer-selective deep representation to improve esophageal cancer classification
No full textEven though artificial intelligence and machine learning have demonstrated remarkable performances in medical image computing, their accountability and transparency level must be improved to transfer this success into clinical practice. The reliability of machine learning decisions must be explained and interpreted, especially for supporting the medical diagnosis.For this task, the deep learning techniques’ black-box nature must somehow be lightened up to clarify its promising results. Hence, we aim to investigate the impact of the ResNet-50 deep convolutional design for Barrett’s esophagus and adenocarcinoma classification. For such a task, and aiming at proposing a two-step learning technique, the output of each convolutional layer that composes the ResNet-50 architecture was trained and classified for further definition of layers that would provide more impact in the architecture. We showed that local information and high-dimensional features are essential to improve the classification for our task. Besides, we observed a significant improvement when the most discriminative layers expressed more impact in the training and classification of ResNet-50 for Barrett’s esophagus and adenocarcinoma classification, demonstrating that both human knowledge and computational processing may influence the correct learning of such a problem
KINiro, Künstliche Intelligenz für Nichtregierungsorganisationen - Bedarf, Akzeptanz und Umsetzungsmöglichkeiten. 3. Arbeitspapier: Künstliche Intelligenz in Nichtregierungsorganisationen: Quantitative Erforschung der Umsetzung in deutschen NROs
No full textHintergrund und Fragestellung
Künstliche Intelligenz (KI) stellt ein wichtiges Thema der Digitalisierung dar und hat
das Potenzial viele Arbeitsfelder grundlegend zu verändern. Mit dem Einsatz von KI
wird die Hoffnung verbunden durch Automatisierung Zeit und Geld zu sparen. Dies
könnte besonders interessant für Organisationen sein, die wenig Ressourcen zur Verfügung
haben, da sie nicht gewinnorientiert oder ehrenamtlich arbeiten. Dazu zählen
Nichtregierungsorganisationen (NRO), die einen wichtigen Bestandteil der Zivilgesellschaft
darstellen. Aufgrund der komplexer werdenden Arbeit von NROs scheint KI
Möglichkeiten zur Bewältigung aktueller und zukünftiger Herausforderungen zu bieten.
Über den Einsatz von KI in NROs ist derzeit jedoch wenig bekannt.
Methodik
Es wurde eine quantitative Querschnittstudie deutscher Nichtregierungsorganisationen
(NRO) zum Thema KI durchgeführt. Der Web-Survey baut auf den Erkenntnissen
des Scoping Reviews und der qualitativen Interviews auf und vertieft den Einblick in
die Bereiche aktuelle Nutzung von KI, Wissenstand in den Organisationen sowie vorhandene
und noch benötigte Ressourcen. Dafür wurden 343 NROs verschiedener Größen
und Handlungsfelder befragt und die Ergebnisse quantitativ analysiert.
Ergebnisse
Die meisten NROs sind der Anwendung von KI gegenüber positiv eingestellt und planen
den Einsatz für die Zukunft oder testen ihn bereits. Dafür werden meist schnell
umsetzbare, externe Anwendungen wie Schreibunterstützungen oder Textübersetzung
für Bürotätigkeiten in den Arbeitsablauf eingebaut. Diese Nutzung beschränkt
sich häufig auf erste Versuche und wird durch interessierte Einzelpersonen vorangetrieben.
Für größere Umsetzung fehlen häufig Zeit, Geld und Wissen. Da die NROs
selbst erwarten, dass die Nutzung von KI in den nächsten Jahren zunehmen wird,
werden mehr Wissensaufbau und Austausch in den Organisationen notwendig.
Schlussfolgerung
Das Interesse an der Thematik scheint in der Zielgruppe vorhanden zu sein, wie die
hohe Teilnahmebereitschaft nahelegt. Doch handelt es sich um ein eher neues Thema
für die NROs, sodass der Einsatz noch getestet wird und feste Strukturen, welche die
Anwendung systematisieren und regeln noch aufgebaut werden müssen. Um dies zu
ermöglichen, sind mehr Wissen, Richtlinien und finanzielle Ressourcen notwendig.
Die Umsetzung hat in den NROs jedoch wenig Priorität, da zunächst andere Hürden
überwunden werden müssen
