The 10 most recently published documents
Abstract
The component-specific quality of linings and prefabricated components made from refractory castables is largely determined by the processing properties during the placement of refractory castables. These are, in particular, the rheological properties, specifically the shear rate-de-pendent dynamic viscosity. Existing measurement methods for determining the shear rate-dependent dynamic viscosity of aggregate-containing suspensions such as refractory castables are prone to errors and inaccurate. From a scientific point of view, an exact determination of the shear rate-dependent dynamic viscosity as well as influencing variables on the rheological properties of refractory castables cannot be realized due to the lack of measurement methods. This leads to contradictory and unclassifiable statements on the influence of aggregate fractions on the rheological properties of refractory castables in the state of science and technology. From a technical point of view, this complicates the rheological optimization of refractory concretes.
Within this work, a measuring method was developed with which the shear rate-dependent dynamic viscosity of aggregate-containing suspensions such as refractory castables can be determined precisely. The measurement method is based on a spherical viscometer for deter-mining the dynamic viscosity and the coupling with a CFD-FEM simulation to determine the actual prevailing shear rate. The determination of the dynamic viscosity is based on the determination of the pull-out speed at a defined pull-out force (force-controlled) according to Stokes' law. It was shown that the pull-out speed and the shear rate can be correlated linearly, which made it possible to redesign the measuring method in a shear rate-controlled manner. A check showed that the shear rate-dependent dynamic viscosity of aggregate-containing suspensions such as refractory castables could be determined just as accurately. The CFD-FEM simulation downstream of the test was no longer necessary.
It was demonstrated that the shear rate-dependent dynamic viscosity of aggregate-containing suspensions increases with increasing mineral aggregate content. Furthermore, it was shown that the complex composition of refractory castables defines their rheological properties. Variations in the particle size distribution of the medium grain and the coarse grain can significantly influence the shear rate-dependent dynamic viscosity, but do not necessarily have to. This also depends on the particle size distribution of the slurry. Furthermore, the dynamic viscosity can be influenced to a different extent at different shear rates. An unpredictable influence of increasing shear rates on dynamic viscosity can therefore be observed.
This complexity complicates the rheological optimization of refractory concretes. Nevertheless, it has been proven that a rheological optimization of refractory concretes, for example the suppression of dilatancy, can also be achieved by optimizing the particle size distribution of the mineral aggregates.
Specifically designed and accurate force fields are of central importance in molecular simulations, as they are often required when investigating new or slightly modified systems. Their parameterization, referred to as force-field parameter optimization, is a complex multi-modal optimization challenge. It requires balancing the parameter’s transferability between various optimization objectives, while their interdependencies often are non-trivial and hard to unravel. This cumulative dissertation addresses this optimization challenge by systematically developing and extending an automatized multi-scale force-field parameter optimization workflow. An important feature is ist modular design that (a) allows the optimization of any amount and combination of force-field parameters towards any type and amount of target properties, and (b) facilitates the extension by additional optimization algorithms or objective functions. First, the workflow’s foundation and a proof-of-concept is provided by simultaneously optimizing the Lennard-Jones parameters towards n-octane’s liquid-phase density (i.e. a multi-molecular, thermodynamic property) and its relative conformational Energies (i.e. a single molecular structural property). By showing that the applicability of the force field was expanded, the simultaneous multi-scale optimization workflow is established. Then, the optimization workflow’s hyperparameters are fine-tuned to improveits results and it is shown that by reasonably balancing the optimization objectives using weighting factors the previously introduced errors are reduced. Next, the optimization procedure’s efficiency is increased by substituting the most time-consuming molecular dynamics simulations by machine learning surrogate models. Those simulations are required repeatedly throughout the optimization, and due to the workflow’s iterative nature, they need to be performed just-in-time. By substituting them with machine learning surrogate models the required run time is approximately reduced by 20-fold. Additionally, guidelines for the model’s training and selection are included. Subsequently, as a challenging test of the workflow, the Lennard-Jones Parameters for 1-bromobutane and 2-bromobutane are optimized. They exhibit a σ-hole allowing halogen bonding, which is subject to current research, that can be supported by accurate simulation models. It is shown that the modeling is improved by the herein presented multi-scale optimization approach, but that further refinements with respect to the workflow and the modeling are necessary. In conclusion, promising approaches to improve the optimization workflow and the modeling are suggested.
Abstract
At the outset of this dissertation, the Bavarian–Czech border region appeared to be a rather unremarkable case in the European context. Unlike at other internal EU borders, renewed border controls had barely been introduced, and there were indications of increasing crossborder integration. Given the region’s complex and historically burdened background, however, it remained unclear how far this process had actually progressed and which factors continued to hinder it. The Bavarian spatial planning instrument of designating cross-border central places was of particular interest in this regard. It suggested close ties between neighboring municipalities along the border. This formed the starting point for two of the three empirical investigations presented here. First, a focus group with local mayors was conducted to discuss the state of cross-border relations and the role of the planning instrument in their cooperation. This was complemented by qualitative interviews with residents to place their everyday lives in relation to the planning postulations. The Covid-19 pandemic, however, partly obscured the initial assumptions. The Bavarian–Czech border region quickly became a pandemic hotspot, which required adjustments to the original research design. The role of cross-border commuters in the ongoing negotiation of the border regime was included as a third relevant topic. As all three approaches address similar localities, their results can be viewed as complementary empirical spotlights and analyzed as part of the social production of a cross-border space. To do so, the dissertation develops a heuristic framework that highlights both the interaction between bordering processes and cross-border integration, and the ways in which local dynamics are embedded in wider developments of European integration. The findings point to two broad modes in the social production of the Bavarian–Czech cross-border border space, each marked by its own ambivalences. They show that cross-border integration is not a linear process, but one that is open, multi-layered and at times contradictory. Localities situated directly at the border emerge as key sites where these tensions become visible. It is in these settings that current debates on overcoming border obstacles and strengthening the resilience of border regions intersect with the everyday lives of the population.
The integration of Large Language Models (LLMs) into information retrieval sys tems has transformed the user experience by providing direct, conversational responses instead of traditional ranked lists of search results. This modification raises substantial concerns about user trust, behaviour, and the risk of misinformation, even as it improves accessibility and convenience. This thesis investigates the impact of generative information retrieval on the reliability of synthesized answers, particularly focusing on how hallucination rates and semantic drift influence trust dynamics and information-seeking behavior. By evaluating the performance of different LLMs on fact-checking benchmarks, the study seeks to quantify the advantages of model scaling against the inherent risks of factual inaccuracy.
The study evaluates hallucination and user trust in LLM-augmented information retrieval systems using three fact-checking datasets. Three well-known semantic similarity metrics are employed to assess the alignment between LLM responses and ground-truth references. Furthermore, the hallucination rate and factual consistency are assessed by aligning model-generated responses with verified annotations in fact-checking datasets. We utilise bias detection measures to evaluate implicit stereotype reinforcement in LLM outputs. This study applies a comprehensive framework for evaluating and auditing hallucinations by combining quantitative performance metrics with user-level reliability insights. The work aims to establish a baseline for the transparency and reliability of LLMs in search and retrieval contexts.
Affordable RGB-D cameras have gained broad research attention in computer vision. This imaging modality provides registered 3-D and RGB data of the perceived environment in the point cloud data format. RGB-D cameras and point clouds allow for a seamless integration of object recognition in mobile robotics applications. Not only do point clouds pose a potential for novel algorithms, the provided geometric information also facilitates estimating adequate positions for mobile manipulation – an action that often follows the recognition process in mobile robotics. Research on vision algorithms based on 3-D data is far less advanced than on 2-D images. Further, RGB-D cameras also pose some technical challenges. Due to their inner workings, they provide erroneous measurements, have a limited range and a much lower resolution than RGB cameras.
In this thesis, I will focus on the object classification and detection parts of vision pipelines for mobile robotics. The main research question is whether point clouds from low-cost RGB-D cameras can be used to reliably solve these tasks. I decided to use a traditional algorithm for my investigations. Traditional algorithms usually demand less computational resources for training and require less data to build a model for inference. For these and other reasons some application scenarios might restrict or prohibit using methods based on deep neural networks (DNN).
The presented point cloud processing pipeline is a non-parametric approach to object classification and detection. It is inspired by the local Naive-Bayes Nearest Neighbor (NBNN) and the Implicit Shape Model (ISM) algorithms originally introduced for 2-D images. Local feature descriptors are used to construct a spatial code-book during the training stage. In the test stage this codebook is used in a Hough voting scheme to generate object hypotheses. I will carefully adapt ideas from the above methods and extend several pipeline steps by novel contributions. At all times, I will target fast processing with limited computational resources in mind.
In a first step, I will focus on isolated objects for classification to gain insights into using point clouds for vision. Subsequently, the presented pipeline will be extended to handle noisy data for object detection in cluttered environments. The contributions of this thesis include an efficient sampling method to find suitable locations for local descriptors and the creation of a descriptive codebook with ranked feature descriptors. The hypothesis generation is followed by an elaborate hypothesis verification step and an additional verification with global feature descriptors in an ensemble classifier. Further, I introduce modifications to two popular local descriptors and also extend them to the global scale.
The developed approaches are evaluated on publicly available datasets with simulated and real sensor data. Further, the mobile service robot Lisa was used for evaluation during several competitions and achieved excellent results. The results of this work enable fast and reliable shape classification of isolated objects, as well as object detection in cluttered environments. The complete pipeline is open-source and is published online in a software repository under a permissive license.
Along with the increased use of automation processes in every other task, job recommendations as well as hiring have also become partially automated. In the process of applying only to those jobs that are recommended by a system or choosing only from those candidates that are selected by an automated system, it becomes highly important to find out if the automated systems are trustworthy enough to provide fair decisions. There has been research on fairness in this sector, where the main focus has been on a single protected attribute, in most cases ’gender.’ That is why the aim of this research is to delve deeper into a deep learning transformer-based algorithm used for job matching with a text-based resume dataset containing several demographic attributes to investigate the fairness of the algorithm not only for gender but also for other demographic groups such as race, age group, and experience level. The fairness evaluation has been carried out using multiple fairness metrics, including demographic parity, conditional demographic parity, and equal opportunity. The transformer models that are pretrained language models are chosen for this study due to their ability to understand the meaning and context of words in resumes and job descriptions. The thesis further investigates bias using an alternative approach by working on a dataset containing varying protected attributes and then conducting a comparable analysis of several bias mitigation methods, including multiple layers of data resampling, along with sensitivity testing through data modification in the preprocessing step of the recommendation process. The research reveals that a system performing fairly when considering a single protected attribute can even hide intersectional unfairness and that the bias mitigation methods do not ensure a balanced improvement across each subgroup when considering multiple combinational demographic groups.