Publication Server of Hochschule Darmstadt
Not a member yet
259 research outputs found
Sort by
MCLFIQ: Mobile Contactless Fingerprint Image Quality
Get PDFWe propose MCLFIQ: Mobile Contactless Fingerprint Image Quality, the first quality assessment algorithm for mobile contactless fingerprint samples. To this end, we re-trained the NIST Fingerprint Image Quality (NFIQ) 2 method, which was originally designed for contact-based fingerprints, with a synthetic contactless fingerprint database. We evaluate the predictive performance of the resulting MCLFIQ model in terms of Error-vs.-Discard Characteristic (EDC) curves on three real-world contactless fingerprint databases using three recognition algorithms. In experiments, the MCLFIQ method is compared against the original NFIQ 2 method, a sharpness-based quality assessment algorithm developed for contactless fingerprint images and the general purpose image quality assessment method BRISQUE. Furthermore, benchmarks on four contact-based fingerprint datasets are also conducted. Obtained results show that the fine-tuning of NFIQ 2 on synthetic contactless fingerprints is a viable alternative to training on real databases. Moreover, the evaluation shows that our MCLFIQ method works more accurately and is more robust compared to all baseline methods on contactless fingerprints. We suggest considering the proposed MCLFIQ method as a starting point for the development of a new standard algorithm for contactless fingerprint quality assessment
Soot and Flame Structures in Turbulent Partially Premixed Jet Flames of Pre-Evaporated Diesel Surrogates with Admixture of OMEn
Get PDFIn this study, the soot formation and oxidation processes in different turbulent, pre-evaporated and partially premixed diesel surrogate flames are experimentally investigated. For this purpose, a piloted jet flame surrounded by an air co-flow is used. Starting from a defined diesel surrogate mixture, different fuel blends with increasing blending ratios of poly(oxymethylene) dimethyl ether (OME) are studied. The Reynolds number, equivalence ratio, and vaporization temperature are kept constant to ensure the comparability of the different fuel mixtures. The effects of OME addition on flame structures, soot precursors, and soot are investigated, showing soot reduction when OME is added to the diesel surrogate. Using chemiluminescence images of C2 radicals (line of sight) and subsequent Abel-inversion, flame lengths and global flame structure are analyzed. The flame structure is visualized by means of planar laser-induced fluorescence (PLIF) of hydroxyl radicals (OH). The spatial distribution of soot precursors, such as polycyclic aromatic hydrocarbons (PAHs), is simultaneously measured by PLIF using the same excitation wavelength. In particular, aromatic compounds with several benzene rings (e.g., naphthalene or pyrene), which are known to be actively involved in soot formation and growth, have been visualized. Spatially distributed soot particles are detected by using laser-induced incandescence (LII), which allows us to study the onset of soot clouds and its structures qualitatively. Evident soot formation is observed in the pure diesel surrogate flame, whereas a significant soot reduction with changing PAH and soot structures can be identified with increasing OME addition
A Paradigm for Modeling Infectious Diseases: Assessing Malware Spread in Early-Stage Outbreaks
Get PDFAs digitalization and artificial intelligence advance, cybersecurity threats intensify, making malware—a type of software installed without authorization to harm users—an increasingly urgent concern. Due to malware’s social and economic impacts, accurately modeling its spread has become essential. While diverse models exist for malware propagation, their selection tends to be intuitive, often overlooking the unique aspects of digital environments. Key model choices include deterministic vs. stochastic, planar vs. spatial, analytical vs. simulation-based, and compartment-based vs. individual state-tracking models. In this context, our study assesses fundamental infection spread models to determine those most applicable to malware propagation. It is organized in two parts: the first examines principles of deterministic and stochastic infection models, and the second provides a comparative analysis to evaluate model suitability. Key criteria include scalability, robustness, complexity, workload, transparency, and manageability. Using consistent initial conditions, control examples are analyzed through Python-based numerical methods and agent-based simulations in NetLogo. The findings yield practical insights and recommendations, offering valuable guidance for researchers and cybersecurity professionals in applying epidemiological models to malware spread
Broad‐scale phenotyping in Arabidopsis reveals varied involvement of RNA interference across diverse plant‐microbe interactions
Get PDFRNA interference (RNAi) is a crucial mechanism in immunity against infectious microbes through the action of DICER‐LIKE (DCL) and ARGONAUTE (AGO) proteins. In the case of the taxonomically diverse fungal pathogen Botrytis cinerea and the oomycete Hyaloperonospora arabidopsidis , plant DCL and AGO proteins have proven roles as negative regulators of immunity, suggesting functional specialization of these proteins. To address this aspect in a broader taxonomic context, we characterized the colonization pattern of an informative set of DCL and AGO loss‐of‐function mutants in Arabidopsis thaliana upon infection with a panel of pathogenic microbes with different lifestyles, and a fungal mutualist. Our results revealed that, depending on the interacting pathogen, AGO1 acts as a positive or negative regulator of immunity, while AGO4 functions as a positive regulator. Additionally, AGO2 and AGO10 positively modulated the colonization by a fungal mutualist. Therefore, analyzing the role of RNAi across a broader range of plant‐microbe interactions has identified previously unknown functions for AGO proteins. For some pathogen interactions, however, all tested mutants exhibited wild‐type‐like infection phenotypes, suggesting that the roles of AGO and DCL proteins in these interactions may be more complex to elucidate
Partitioning Point Clouds and Meshes with Deep Learning
Get PDFThe overall objective of this dissertation is to enable a more efficient and effective point cloud and mesh partition for artists and 3D application developers. In this dissertation, 3D scans are assumed as the source data material of the 3D application development, reducing the manual and time-consuming modelling of virtual objects. Furthermore, the scanned data is assumed to be processed to a point cloud and reconstructed to a polygon mesh. The mesh has to be partitioned into the objects of interest to design specific interactions with a game engine. Interviews revealed that the partition is manually conducted on a mesh with a 3D manipulation software, which is time-consuming. The partition creation should be automated to increase efficiency and effectiveness. Freely available point cloud and mesh partition algorithms require an expert with appropriate programming skills and field knowledge, which makes them difficult to use. More precisely, the algorithms cannot be used in existing workflows as they are not implemented in a common graphical 3D manipulation software. Beneath these problems, the partition automation should work on real-world data and have a low runtime to raise efficiency. Different sub-research objectives were formulated from these problems and requirements, leading to novel approaches in the domains of: (a) sequential partition creation with deep reinforcement and imitation learning, (b) episodic partition creation with graph neural networks, (c) match-based reward calculation and (d) synthetic scene generation. One sub-research objective is the replacement of a human expert with an agent. In this context, a novel deep reinforcement learning (DRL) partition framework is presented. Experiments were conducted using this framework combined with the region growing algorithm and synthetic scenes created by a self-developed scene generator. The maximum reward could almost be achieved with a fine-tuned PointNet and by evaluating the wall and non-wall objects separately. This approach is not applicable to real-world scenes, which is necessary to achieve the efficiency and effectiveness objective. Therefore, another DRL partition approach is introduced, where an agent unifies superpoints in the so-called superpoint growing environment. The point cloud is divided into superpoints, which will be unified into the objects of interest by an agent. The experimental results show that this approach can be applied to real-world scenes. Beneath the application of DRL, an imitation learning approach was developed, increasing the agent’s performance in the superpoint growing environment. The runtime in the sequential superpoint growing environment is poor, as each union decision requires a neural network call. Hence, a further sub-research objective is to improve the runtime. An episodic environment was developed as a solution, only requiring one graph neural network call. Similarities between superpoints are estimated in this environment and passed to a union algorithm. The differences between two graph neural network architectures and two union algorithms were experimentally investigated. According to the results, calculating the superpoint similarities with a correlation of the embedded node features is more robust than the similarity estimation with a sigmoid activation function. The reward function, used in the DRL partition approaches, was realised by a matching procedure. As this function influences the partition quality, another sub-research objective is to investigate the differences between various match types. Matching functions from the literature were compared, and another match type was introduced. The usage of different match types in the learning process was experimentally evaluated. Although an agent gets more feedback with all match types, the best results (visual and in terms of the partition size) were achieved by only using first-order matches in the reward function. The synthetic scenes of the region growing approach lack realism as the lighting information is ignored, which can be important to train networks for the partition task. Therefore, a further sub-research objective is to develop a scene generator where the lighting is taken into account. After its development, the generated scenes were experimentally evaluated in a pre-training task. It turned out that the lighting information is important for a pre-training as larger accuracies were achieved. Furthermore, a faster convergence can be achieved with the pre-trained network instead of training a network on a target data set from scratch.
Another sub-research objective targets the development of a usable partition interface. In this context, the Blender add-on OpenXtract was developed, containing five open-source point cloud partition algorithms. The partition algorithms were extended by approximating geodesic distances so that the edges of meshes are used. An experiment has shown that the extended algorithms produce larger accuracies, which is considered an increase in effectiveness. Moreover, unstructured interviews revealed that OpenXtract can improve the effectiveness and efficiency of the partition creation
How to design a value-based Chatbot for the manufacturing industry: An empirical study of an internal assistance for employees
Get PDFWith regard to AI as a key technology, this scientific paper deals with the identification of user drivers on the purchase decision of a cooperative AI (as explainable AI—XAI), as well as the analysis of the willingness to pay in the context of value-based pricing. Besides the economic dimension with regard to usefulness and usability of the system, the focus is mainly on the (innovative) explainable character. The analysis is carried out by a choice-based conjoint analysis (CBC) using the example of an intelligent assistance system for employees that supports internal business processes and workflows in business organizations. For this purpose, fictitious purchase offers were created under which decision-makers in manufacturing business organizations in Germany made simulated purchase decisions. The analysis shows that the target group attach great utility value to transparency in the sense of explanatory content, in addition to a high degree of interactivity and a high level of reliability
Zwischen Authentizität und Anpassung. Perspektiven von Frauen in Führungspositionen auf das Spannungsverhältnis zwischen der weiblichen Rolle und der Rolle einer Führungskraft
Get PDFDie soziale Rolle der Frau und die soziale Rolle der Führungskraft gelten als unvereinbar (siehe Rollenkongruenztheorie nach Eagly & Karau (2002)). Frauen in Führungspositionen stehen daher vor der Herausforderung, gegensätzliche Rollenanforderungen in ihrem Verhalten umzusetzen. Diese Studie hat zum Ziel, das Spannungsverhältnis dieser Rollenerwartungen in der Wahrnehmung weiblicher Führungskräfte nachzuweisen und darüber hinaus Verhaltensmaßnahmen zur Bewältigung dieses Spannungsverhältnisses aufzuzeigen. Zur Beantwortung dieser Forschungsfragen wird eine qualitative Befragung mit weiblichen Führungskräften durchgeführt. Die Auswertung der leitfadenbasierten Interviews erfolgt mithilfe der Inhaltsanalyse nach Kuckartz (2022). Die Ergebnisse der Inhaltsanalyse belegen, dass zwischen der weiblichen Rolle und der Rolle einer Führungskraft ein Spannungsverhältnis existiert, das jedoch nicht alle Frauen wahrnehmen. Obwohl sich das Spannungsverhältnis nicht in der individuellen Wahrnehmung aller weiblichen Führungskräfte nachweisen lässt, ist es anhand der bewussten Verhaltensänderungen der Probandinnen erkennbar. Diese Verhaltensänderungen haben zum Ziel, sich von stereotypen weiblichen Eigenschaften und Verhaltensweisen zu entfernen, um als Führungskraft wahr- und ernst genommen zu werden. Diese Anpassungsleistung birgt folglich das Risiko, die wahrgenommene Weiblichkeit sowie die Authentizität der betreffenden Frauen zu gefährden, was gleichzeitig zur Schmälerung der Führungsautorität beiträgt. Für weibliche Führungskräfte resultiert daraus, dass es weder eine Auflösung noch eine ideale Verhaltensstrategie für den Umgang mit diesem Spannungsverhältnis gibt. Vielmehr besteht die Notwendigkeit, diese Unauflösbarkeit zu akzeptieren und einen individuellen Führungsstil zu entwickeln, der die eigene Authentizität weitestgehend aufrechterhält.The traditional social role of women and the role of being a manager are considered incompatible (according to the Role congruity theory of prejudice toward female leaders by Eagly and Karau, 2002). Women taking on management positions are therefore faced with the challenge of mirroring conflicting role expectations. The purpose of the study is to prove the tension between two role expectations from the perspective of female managers and to describe behavioral measures to cope with the pressure associated with it. A qualitative survey of female executives provided the answers to the study questions. Using
Kuckartz content analysis (2022), the results of the guideline-based interviews show the tension between the female role and the role of the manager, however not everyone interviewed perceive it that way. Although not all women recognize the tension, the changes in the test subjects' conscious behavior are recognizable. For women to be perceived and taken seriously as leaders, they tend to move away from stereotypical female behaviors. While this adaptability can be beneficial, it can also endanger the perceived femininity and authenticity of women, thus diminishing their leadership authority. The solution to this tension is neither an ideal behavioral strategy nor a resolution for femle managers. Rather, there is a need to accept this indissolubility and to develop an individual leadership style that maintains one's authenticity as far as possible
Small-scale urban design interventions: A framework for deploying cities as resource for mental health and mental health literacy
Get PDFWith roughly half of the global population living in cities, urban environments become central to public health often perceived as health risk factors. Indeed, mental disorders show higher incidences in urban contexts compared to rural areas. However, shared urban environments also provide a rich potential to act as a resource for mental health and as a platform to increase mental health literacy. Based on the concepts of salutogenesis and restorative environments, we propose a framework for urban design interventions. It outlines (a) an output level, i.e., preventive and discursive potentials of such interventions to act as biopsychosocial resources, and (b) a process level, i.e., mechanisms of inter- and transdisciplinary collaboration of researchers and citizens in the design process. This approach aims at combining evidence-based, salutogenic, psychosocially-supportive design with a focus on mental health. Implementing low-threshold, resource-efficient options in the existing urban context brings this topic to the public space. Implications for the implementation of such interventions for citizens, researchers, and municipality stakeholders are discussed. This illustrates new directions of research for urban person-environment interactions, public health, and beyond
Künstliche Intelligenz im Studium Eine quantitative Befragung von Studierenden zur Nutzung von ChatGPT & Co.
Get PDFKI-basierte Tools wie ChatGPT bzw. GPT-4 verändern derzeit die Hochschullandschaft und vielerorts wird bereits über die Konsequenzen für die zukünftigen Lehr- und Prüfungsformen diskutiert. Um hier eine empirische Grundlage zu schaffen, ist eine deutschlandweite Befragung von Studierenden durchgeführt worden, in welcher das Nutzungsverhalten im Umgang mit KI-basierten Tools im Rahmen des Studiums und Alltags erfasst wurde. Hierbei wurden unter anderem diverse Funktionen der KIbasierten Tools identifiziert, die für die Studierenden als besonders wichtig eingeschätzt wurden. Das Ziel der quantitativen Befragung lag somit in der Erfassung davon, wie KI-Tools genutzt werden und welche Faktoren für die Nutzung maßgeblich sind.
Insgesamt haben sich deutschlandweit über 6300 Studierende an der anonymen Befragung beteiligt. Die Ergebnisse dieser quantitativen Analyse verdeutlichen, dass fast zwei Drittel der befragten Studierenden KI-basierte Tools im Rahmen des Studiums nutzen bzw. genutzt haben. Explizit nennen in diesem Kontext fast die Hälfte der befragten Studierenden ChatGPT bzw. GPT-4 als genutztes Tool. Am häufigsten nutzen Studierende der Ingenieurwissenschaften sowie Mathematik und Naturwissenschaften KI-basierte Tools.
Eine differenzierte Betrachtung des Nutzungsverhaltens verdeutlicht, dass die Studierenden KI-basierte Tools vielfältig einsetzen. Die Klärung von Verständnisfragen und Erläuterung fachspezifischer Konzepte zählen in diesem Kontext zu den relevantesten Nutzungsgründen