Filtern
Erscheinungsjahr
Dokumenttyp
Sprache
- Englisch (518) (entfernen)
Schlagworte
- Stress (27)
- Modellierung (19)
- Fernerkundung (18)
- Optimierung (17)
- Deutschland (16)
- Hydrocortison (13)
- Satellitenfernerkundung (13)
- Cortisol (9)
- Europäische Union (9)
- Finanzierung (9)
Institut
- Raum- und Umweltwissenschaften (99)
- Psychologie (94)
- Fachbereich 4 (53)
- Mathematik (47)
- Fachbereich 6 (38)
- Wirtschaftswissenschaften (29)
- Fachbereich 1 (24)
- Informatik (19)
- Anglistik (15)
- Rechtswissenschaft (14)
Non-probability sampling is a topic of growing relevance, especially due to its occurrence in the context of new emerging data sources like web surveys and Big Data.
This thesis addresses statistical challenges arising from non-probability samples, where unknown or uncontrolled sampling mechanisms raise concerns in terms of data quality and representativity.
Various methods to quantify and reduce the potential selectivity and biases of non-probability samples in estimation and inference are discussed. The thesis introduces new forms of prediction and weighting methods, namely
a) semi-parametric artificial neural networks (ANNs) that integrate B-spline layers with optimal knot positioning in the general structure and fitting procedure of artificial neural networks, and
b) calibrated semi-parametric ANNs that determine weights for non-probability samples by integrating an ANN as response model with calibration constraints for totals, covariances and correlations.
Custom-made computational implementations are developed for fitting (calibrated) semi-parametric ANNs by means of stochastic gradient descent, BFGS and sequential quadratic programming algorithms.
The performance of all the discussed methods is evaluated and compared for a bandwidth of non-probability sampling scenarios in a Monte Carlo simulation study as well as an application to a real non-probability sample, the WageIndicator web survey.
Potentials and limitations of the different methods for dealing with the challenges of non-probability sampling under various circumstances are highlighted. It is shown that the best strategy for using non-probability samples heavily depends on the particular selection mechanism, research interest and available auxiliary information.
Nevertheless, the findings show that existing as well as newly proposed methods can be used to ease or even fully counterbalance the issues of non-probability samples and highlight the conditions under which this is possible.
Modern decision making in the digital age is highly driven by the massive amount of
data collected from different technologies and thus affects both individuals as well as
economic businesses. The benefit of using these data and turning them into knowledge
requires appropriate statistical models that describe the underlying observations well.
Imposing a certain parametric statistical model goes along with the need of finding
optimal parameters such that the model describes the data best. This often results in
challenging mathematical optimization problems with respect to the model’s parameters
which potentially involve covariance matrices. Positive definiteness of covariance matrices
is required for many advanced statistical models and these constraints must be imposed
for standard Euclidean nonlinear optimization methods which often results in a high
computational effort. As Riemannian optimization techniques proved efficient to handle
difficult matrix-valued geometric constraints, we consider optimization over the manifold
of positive definite matrices to estimate parameters of statistical models. The statistical
models treated in this thesis assume that the underlying data sets used for parameter
fitting have a clustering structure which results in complex optimization problems. This
motivates to use the intrinsic geometric structure of the parameter space. In this thesis,
we analyze the appropriateness of Riemannian optimization over the manifold of positive
definite matrices on two advanced statistical models. We establish important problem-
specific Riemannian characteristics of the two problems and demonstrate the importance
of exploiting the Riemannian geometry of covariance matrices based on numerical studies.
Even though proper research on Cauchy transforms has been done, there are still a lot of open questions. For example, in the case of representation theorems, i.e. the question when a function can be represented as a Cauchy transform, there is 'still no completely satisfactory answer' ([9], p. 84). There are characterizations for measures on the circle as presented in the monograph [7] and for general compactly supported measures on the complex plane as presented in [27]. However, there seems to exist no systematic treatise of the Cauchy transform as an operator on $L_p$ spaces and weighted $L_p$ spaces on the real axis.
This is the point where this thesis draws on and we are interested in developing several characterizations for the representability of a function by Cauchy transforms of $L_p$ functions. Moreover, we will attack the issue of integrability of Cauchy transforms of functions and measures, a topic which is only partly explored (see [43]). We will develop different approaches involving Fourier transforms and potential theory and investigate into sufficient conditions and characterizations.
For our purposes, we shall need some notation and the concept of Hardy spaces which will be part of the preliminary Chapter 1. Moreover, we introduce Fourier transforms and their complex analogue, namely Fourier-Laplace transforms. This will be of extraordinary usage due to the close connection of Cauchy and Fourier(-Laplace) transforms.
In the second chapter we shall begin our research with a discussion of the Cauchy transformation on the classical (unweighted) $L_p$ spaces. Therefore, we start with the boundary behavior of Cauchy transforms including an adapted version of the Sokhotski-Plemelj formula. This result will turn out helpful for the determination of the image of the Cauchy transformation under $L_p(\R)$ for $p\in(1,\infty).$ The cases $p=1$ and $p=\infty$ are playing special roles here which justifies a treatise in separate sections. For $p=1$ we will involve the real Hardy space $H_{1}(\R)$ whereas the case $p=\infty$ shall be attacked by an approach incorporating intersections of Hardy spaces and certain subspaces of $L_{\infty}(\R).$
The third chapter prepares ourselves for the study of the Cauchy transformation on subspaces of $L_{p}(\R).$ We shall give a short overview of the basic facts about Cauchy transforms of measures and then proceed to Cauchy transforms of functions with support in a closed set $X\subset\R.$ Our goal is to build up the main theory on which we can fall back in the subsequent chapters.
The fourth chapter deals with Cauchy transforms of functions and measures supported by an unbounded interval which is not the entire real axis. For convenience we restrict ourselves to the interval $[0,\infty).$ Bringing once again the Fourier-Laplace transform into play, we deduce complex characterizations for the Cauchy transforms of functions in $L_{2}(0,\infty).$ Moreover, we analyze the behavior of Cauchy transform on several half-planes and shall use these results for a fairly general geometric characterization. In the second section of this chapter, we focus on Cauchy transforms of measures with support in $[0,\infty).$ In this context, we shall derive a reconstruction formula for these Cauchy transforms holding under pretty general conditions as well as results on the behaviur on the left half-plane. We close this chapter by rather technical real-type conditions and characterizations for Cauchy transforms of functions in $L_p(0,\infty)$ basing on an approach in [82].
The most common case of Cauchy transforms, those of compactly supported functions or measures, is the subject of Chapter 5. After complex and geometric characterizations originating from similar ideas as in the fourth chapter, we adapt a functional-analytic approach in [27] to special measures, namely those with densities to a given complex measure $\mu.$ The chapter is closed with a study of the Cauchy transformation on weighted $L_p$ spaces. Here, we choose an ansatz through the finite Hilbert transform on $(-1,1).$
The sixth chapter is devoted to the issue of integrability of Cauchy transforms. Since this topic has no comprehensive treatise in literature yet, we start with an introduction of weighted Bergman spaces and general results on the interaction of the Cauchy transformation in these spaces. Afterwards, we combine the theory of Zen spaces with Cauchy transforms by using once again their connection with Fourier transforms. Here, we shall encounter general Paley-Wiener theorems of the recent past. Lastly, we attack the issue of integrability of Cauchy transforms by means of potential theory. Therefore, we derive a Fourier integral formula for the logarithmic energy in one and multiple dimensions and give applications to Fourier and hence Cauchy transforms.
Two appendices are annexed to this thesis. The first one covers important definitions and results from measure theory with a special focus on complex measures. The second appendix contains Cauchy transforms of frequently used measures and functions with detailed calculations.
The COVID-19 pandemic has affected schooling worldwide. In many places, schools closed for weeks or months, only part of the student body could be educated at any one time, or students were taught online. Previous research discloses the relevance of schooling for the development of cognitive abilities. We therefore compared the intelligence test performance of 424 German secondary school students in Grades 7 to 9 (42% female) tested after the first six months of the COVID-19 pandemic (i.e., 2020 sample) to the results of two highly comparable student samples tested in 2002 (n = 1506) and 2012 (n = 197). The results revealed substantially and significantly lower intelligence test scores in the 2020 sample than in both the 2002 and 2012 samples. We retested the 2020 sample after another full school year of COVID-19-affected schooling in 2021. We found mean-level changes of typical magnitude, with no signs of catching up to previous cohorts or further declines in cognitive performance. Perceived stress during the pandemic did not affect changes in intelligence test results between the two measurements.
COVID-19 was a harsh reminder that diseases are an aspect of human existence and mortality. It was also a live experiment in the formation and alteration of disease-related attitudes. Not only are these attitudes relevant to an individual’s self-protective behavior, but they also seem to be associated with social and political attitudes more broadly. One of these attitudes is Social Darwinism, which holds that a pandemic benefits society by enabling nature “to weed out the weak”. In two countries (N = 300, N = 533), we introduce and provide evidence for the reliability, validity, and usefulness of the Disease-Related Social Darwinism (DRSD) Short Scale measuring this concept. Results indicate that DRSD is meaningful related to other central political attitudes like Social Dominance Orientation, Authoritarianism and neoliberalism. Importantly, the scale significantly predicted people’s protective behavior during the Pandemic over and above general social Darwinism. Moreover, it significantly predicted conservative attitudes, even after controlling for Social Dominance Orientation.
Die Dissertation beschäftigt sich mit einer neuartigen Art von Branch-and-Bound Algorithmen, deren Unterschied zu klassischen Branch-and-Bound Algorithmen darin besteht, dass
das Branching durch die Addition von nicht-negativen Straftermen zur Zielfunktion erfolgt
anstatt durch das Hinzufügen weiterer Nebenbedingungen. Die Arbeit zeigt die theoretische Korrektheit des Algorithmusprinzips für verschiedene allgemeine Klassen von Problemen und evaluiert die Methode für verschiedene konkrete Problemklassen. Für diese Problemklassen, genauer Monotone und Nicht-Monotone Gemischtganzzahlige Lineare Komplementaritätsprobleme und Gemischtganzzahlige Lineare Probleme, präsentiert die Arbeit
verschiedene problemspezifische Verbesserungsmöglichkeiten und evaluiert diese numerisch.
Weiterhin vergleicht die Arbeit die neue Methode mit verschiedenen Benchmark-Methoden
mit größtenteils guten Ergebnissen und gibt einen Ausblick auf weitere Anwendungsgebiete
und zu beantwortende Forschungsfragen.
The forensic application of phonetics relies on individuality in speech. In the forensic domain, individual patterns of verbal and paraverbal behavior are of interest which are readily available, measurable, consistent, and robust to disguise and to telephone transmission. This contribution is written from the perspective of the forensic phonetic practitioner and seeks to establish a more comprehensive concept of disfluency than previous studies have. A taxonomy of possible variables forming part of what can be termed disfluency behavior is outlined. It includes the “classical” fillers, but extends well beyond these, covering, among others, additional types of fillers as well as prolongations, but also the way in which fillers are combined with pauses. In the empirical section, the materials collected for an earlier study are re-examined and subjected to two different statistical procedures in an attempt to approach the issue of individuality. Recordings consist of several minutes of spontaneous speech by eight speakers on three different occasions. Beyond the established set of hesitation markers, additional aspects of disfluency behavior which fulfill the criteria outlined above are included in the analysis. The proportion of various types of disfluency markers is determined. Both statistical approaches suggest that these speakers can be distinguished at a level far above chance using the disfluency data. At the same time, the results show that it is difficult to pin down a single measure which characterizes the disfluency behavior of an individual speaker. The forensic implications of these findings are discussed.
Redox-driven biogeochemical cycling of iron plays an integral role in the complex process network of ecosystems, such as carbon cycling, the fate of nutrients and greenhouse gas emissions. We investigate Fe-(hydr)oxide (trans)formation pathways from rhyolitic tephra in acidic topsoils of South Patagonian Andosols to evaluate the ecological relevance of terrestrial iron cycling for this sensitive fjord ecosystem. Using bulk geochemical analyses combined with micrometer-scale-measurements on individual soil aggregates and tephra pumice, we document biotic and abiotic pathways of Fe released from the glassy tephra matrix and titanomagnetite phenocrysts. During successive redox cycles that are controlled by frequent hydrological perturbations under hyper-humid climate, (trans)formations of ferrihydrite-organic matter coprecipitates, maghemite and hematite are closely linked to tephra weathering and organic matter turnover. These Fe-(hydr)oxides nucleate after glass dissolution and complexation with organic ligands, through maghemitization or dissolution-(re)crystallization processes from metastable precursors. Ultimately, hematite represents the most thermodynamically stable Fe-(hydr)oxide formed under these conditions and physically accumulates at redox interfaces, whereas the ferrihydrite coprecipitates represent a so far underappreciated terrestrial source of bio-available iron for fjord bioproductivity. The insights into Fe-(hydr)oxide (trans)formation in Andosols have implications for a better understanding of biogeochemical cycling of iron in this unique Patagonian fjord ecosystem.
We use a novel sea-ice lead climatology for the winters of 2002/03 to 2020/21 based on satellite observations with 1 km2 spatial resolution to identify predominant patterns in Arctic wintertime sea-ice leads. The causes for the observed spatial and temporal variabilities are investigated using ocean surface current velocities and eddy kinetic energies from an ocean model (Finite Element Sea Ice–Ice-Shelf–Ocean Model, FESOM) and winds from a regional climate model (CCLM) and ERA5 reanalysis, respectively. The presented investigation provides evidence for an influence of ocean bathymetry and associated currents on the mechanic weakening of sea ice and the accompanying occurrence of sea-ice leads with their characteristic spatial patterns. While the driving mechanisms for this observation are not yet understood in detail, the presented results can contribute to opening new hypotheses on ocean–sea-ice interactions. The individual contribution of ocean and atmosphere to regional lead dynamics is complex, and a deeper insight requires detailed mechanistic investigations in combination with considerations of coastal geometries. While the ocean influence on lead dynamics seems to act on a rather long-term scale (seasonal to interannual), the influence of wind appears to trigger sea-ice lead dynamics on shorter timescales of weeks to months and is largely controlled by individual events causing increased divergence. No significant pan-Arctic trends in wintertime leads can be observed.
Regional climate models are a valuable tool for the study of the climate processes and climate change in polar regions, but the performance of the models has to be evaluated using experimental data. The regional climate model CCLM was used for simulations for the MOSAiC period with a horizontal resolution of 14 km (whole Arctic). CCLM was used in a forecast mode (nested in ERA5) and used a thermodynamic sea ice model. Sea ice concentration was taken from AMSR2 data (C15 run) and from a high-resolution data set (1 km) derived from MODIS data (C15MOD0 run). The model was evaluated using radiosonde data and data of different profiling systems with a focus on the winter period (November–April). The comparison with radiosonde data showed very good agreement for temperature, humidity, and wind. A cold bias was present in the ABL for November and December, which was smaller for the C15MOD0 run. In contrast, there was a warm bias for lower levels in March and April, which was smaller for the C15 run. The effects of different sea ice parameterizations were limited to heights below 300 m. High-resolution lidar and radar wind profiles as well as temperature and integrated water vapor (IWV) data from microwave radiometers were used for the comparison with CCLM for case studies, which included low-level jets. LIDAR wind profiles have many gaps, but represent a valuable data set for model evaluation. Comparisons with IWV and temperature data of microwave radiometers show very good agreement.
The benefits of prosocial power motivation in leadership: Action orientation fosters a win-win
(2023)
Power motivation is considered a key component of successful leadership. Based on its dualistic nature, the need for power (nPower) can be expressed in a dominant or a prosocial manner. Whereas dominant motivation is associated with antisocial behaviors, prosocial motivation is characterized by more benevolent actions (e.g., helping, guiding). Prosocial enactment of the power motive has been linked to a wide range of beneficial outcomes, yet less has been investigated what determines a prosocial enactment of the power motive. According to Personality Systems Interactions (PSI) theory, action orientation (i.e., the ability to self-regulate affect) promotes prosocial enactment of the implicit power motive and initial findings within student samples verify this assumption. In the present study, we verified the role of action orientation as an antecedent for prosocial power enactment in a leadership sample (N = 383). Additionally, we found that leaders personally benefited from a prosocial enactment strategy. Results show that action orientation through prosocial power motivation leads to reduced power-related anxiety and, in turn, to greater leader well-being. The integration of motivation and self-regulation research reveals why leaders enact their power motive in a certain way and helps to understand how to establish a win-win situation for both followers and leaders.
Repeatedly encountering a stimulus biases the observer’s affective response and evaluation of the stimuli. Here we provide evidence for a causal link between mere exposure to fictitious news reports and subsequent voting behavior. In four pre-registered online experiments, participants browsed through newspaper webpages and were tacitly exposed to names of fictitious politicians. Exposure predicted voting behavior in a subsequent mock election, with a consistent preference for frequent over infrequent names, except when news items were decidedly negative. Follow-up analyses indicated that mere media presence fuels implicit personality theories regarding a candidate’s vigor in political contexts. News outlets should therefore be mindful to cover political candidates as evenly as possible.
Addition of Phosphogypsum to Fire-Resistant Plaster Panels:
A Physic–Mechanical Investigation
(2023)
Gypsum (GPS) has great potential for structural fire protection and is increasingly used in construction due to its high-water retention and purity. However, many researchers aim to improve its physical and mechanical properties by adding other organic or inorganic materials such as fibers, recycled GPS, and waste residues. This study used a novel method to add non-natural GPS from factory waste (phosphogypsum (PG)) as a secondary material for GPS. This paper proposes to mix these two materials to properly study the effect of PG on the physico-mechanical properties and fire performance of two Tunisian GPSs (GPS1 and GPS2). PG initially replaced GPS at 10, 20, 30, 40, and 50% weight percentage (mixing plan A). The PGs were then washed with distilled water several times. Two more mixing plans were run when the pH of the PG was equal to 2.4 (mixing plan B), and the pH was equal to 5 (mixing plan C). Finally, a comparative study was conducted on the compressive strength, flexural strength, density, water retention, and mass loss levels after 90 days of drying, before/after incineration of samples at 15, 30, 45, and 60 min. The results show that the mixture of GPS1 and 30% PG (mixing plan B) obtained the highest compressive strength (41.31%) and flexural strength (35.03%) compared to the reference sample. The addition of 10% PG to GPS1 (mixing plan A) improved fire resistance (33.33%) and the mass loss (17.10%) of the samples exposed to flame for 60 min compared to GPS2. Therefore, PG can be considered an excellent insulating material, which can increase physico-mechanical properties and fire resistance time of plaster under certain conditions.
People are increasingly concerned about how meat affects the environment, human health, and animal welfare, yet eating and enjoying meat remains a norm. Unsurprisingly, many people are ambivalent about meat—evaluating it as both positive and negative. Here, we propose that meat-related conflict is multidimensional and depends on people’s dietary group: Omnivores’ felt ambivalence relates to multiple negative associations that oppose a predominantly positive attitude towards meat, and veg*ans’ ambivalence relates to various positive associations that oppose a predominantly negative attitude. A qualitative study (N = 235; German) revealed that omnivores and veg*ans experience meat-related ambivalence due to associations with animals, sociability, sustainability, health, and sensory experiences. To quantify felt ambivalence in these domains, we developed the Meat Ambivalence Questionnaire (MAQ). We validated the MAQ in four pre-registered studies using self-report and behavioral data (N = 3,485; German, UK, representative US). Both omnivores and veg*ans reported meat-related ambivalence, but with differences across domains and their consequences for meat consumption. Specifically, ambivalence was associated with less meat consumption in omnivores (especially sensory-/animal-based ambivalence) and more meat consumption in veg*ans (especially sensory-/socially-based ambivalence). Network analyses shed further light on the nomological net of the MAQ while controlling for a comprehensive set of determinants of meat consumption. By introducing the MAQ, we hope to provide researchers with a tool to better understand how ambivalence accompanies behavior change and maintenance.
Properties Evaluation of Composite Materials Based on Gypsum Plaster and Posidonia Oceanica Fibers
(2023)
Estimating the amount of material without significant losses at the end of hybrid casting is a problem addressed in this study. To minimize manufacturing costs and improve the accuracy of results, a correction factor (CF) was used in the formula to estimate the volume percent of the material in order to reduce material losses during the sample manufacturing stage, allowing for greater confidence between the approved blending plan and the results obtained. In this context, three material mixing schemes of different sizes and shapes (gypsum plaster, sand (0/2), gravel (2/4), and Posidonia oceanica fibers (PO)) were created to verify the efficiency of CF and more precisely study the physico-mechanical effects on the samples. The results show that the use of a CF can reduce mixing loss to almost 0%. The optimal compressive strength of the sample (S1B) with the lowest mixing loss was 7.50 MPa. Under optimal conditions, the addition of PO improves mix volume percent correction (negligible), flexural strength (5.45%), density (18%), and porosity (3.70%) compared with S1B. On the other hand, the addition of PO thermo-chemical treatment by NaOH increases the compressive strength (3.97%) compared with PO due to the removal of impurities on the fiber surface, as shown by scanning electron microscopy. We then determined the optimal mixture ratio (PO divided by a mixture of plaster, sand, and gravel), which equals 0.0321 because Tunisian gypsum contains small amounts of bassanite and calcite, as shown by the X-ray diffraction results.
The publication of statistical databases is subject to legal regulations, e.g. national statistical offices are only allowed to publish data if the data cannot be attributed to individuals. Achieving this privacy standard requires anonymizing the data prior to publication. However, data anonymization inevitably leads to a loss of information, which should be kept minimal. In this thesis, we analyze the anonymization method SAFE used in the German census in 2011 and we propose a novel integer programming-based anonymization method for nominal data.
In the first part of this thesis, we prove that a fundamental variant of the underlying SAFE optimization problem is NP-hard. This justifies the use of heuristic approaches for large data sets. In the second part, we propose a new anonymization method belonging to microaggregation methods, specifically designed for nominal data. This microaggregation method replaces rows in a microdata set with representative values to achieve k-anonymity, ensuring each data row is identical to at least k − 1 other rows. In addition to the overall dissimilarities of the data rows, the method accounts for errors in resulting frequency tables, which are of high interest for nominal data in practice. The method employs a typical two-step structure: initially partitioning the data set into clusters and subsequently replacing all cluster elements with representative values to achieve k-anonymity. For the partitioning step, we propose a column generation scheme followed by a heuristic to obtain an integer solution, which is based on the dual information. For the aggregation step, we present a mixed-integer problem formulation to find cluster representatives. To this end, we take errors in a subset of frequency tables into account. Furthermore, we show a reformulation of the problem to a minimum edge-weighted maximal clique problem in a multipartite graph, which allows for a different perspective on the problem. Moreover, we formulate a mixed-integer program, which combines the partitioning and the aggregation step and aims to minimize the sum of chi-squared errors in frequency tables.
Finally, an experimental study comparing the methods covered or developed in this work shows particularly strong results for the proposed method with respect to relative criteria, while SAFE shows its strength with respect to the maximum absolute error in frequency tables. We conclude that the inclusion of integer programming in the context of data anonymization is a promising direction to reduce the inevitable information loss inherent in anonymization, particularly for nominal data.
The microbial enzyme alkaline phosphatase contributes to the removal of organic phosphorus compounds from wastewaters. To cope with regulatory threshold values for permitted maximum phosphor concentrations in treated wastewaters, a high activity of this enzyme in the biological treatment stage, e.g., the activated sludge process, is required. To investigate the reaction dynamics of this enzyme, to analyze substrate selectivities, and to identify potential inhibitors, the determination of enzyme kinetics is necessary. A method based on the application of the synthetic fluorogenic substrate 4-methylumbelliferyl phosphate is proven for soils, but not for activated sludges. Here, we adapt this procedure to the latter. The adapted method offers the additional benefit to determine inhibition kinetics. In contrast to conventional photometric assays, no particle removal, e.g., of sludge pellets, is required enabling the analysis of the whole sludge suspension as well as of specific sludge fractions. The high sensitivity of fluorescence detection allows the selection of a wide substrate concentration range for sound modeling of kinetic functions.
- Fluorescence array technique for fast and sensitive analysis of high sample numbers
- No need for particle separation – analysis of the whole (diluted) sludge suspension
- Simultaneous determination of standard and inhibition kinetics
Startups are essential agents for the evolution of economies and the creative destruction of established market conditions for the benefit of a more effective and efficient economy. Their significance is manifested in their drive for innovation and technological advancements, their creation of new jobs, their contribution to economic growth, and their impact on increased competition and increased market efficiency. By reason of their attributes of newness and smallness, startups often experience a limitation in accessing external financial resources. Extant research on entrepreneurial finance examines the capital structure of startups, various funding tools, financing environments in certain regions, and investor selection criteria among other topics. My dissertation contributes to this research area by examining the becoming increasingly important funding instrument of venture debt. Prior research on venture debt only investigated the business model of venture debt, the concept of venture debt, the selection criteria of venture debt providers, and the role of patents in the venture debt provider’s selection process. Based on qualitative and quantitative methods, the dissertation outlines the emergence of venture debt in Europe as well as the impact of venture debt on startups to open up a better understanding of venture debt.
The results of the qualitative studies indicate that venture debt was formed based on a ‘Kirznerian’ entrepreneurial opportunity and venture debt impacts startups positive and negative in their development via different impact mechanisms.
Based on these results, the dissertation analyzes the empirical impact of venture debt on a startup’s ability to acquire additional financial resources as well as the role of the reputation of venture debt providers. The results suggest that venture debt increases the likelihood of acquiring additional financial resources via subsequent funding rounds and trade sales. In addition, a higher venture debt provider reputation increases the likelihood of acquiring additional financial resources via IPOs.
Diese Dissertation beschäftigt sich mit der Fragestellung, ob und wie Intersektionalität als analytische Perspektive für literarische Texte eine nützliche Ergänzung für ethnisch geordnete Literaturfelder darstellt. Diese Fragestellung wird anhand der Analyse dreier zeitgenössischer chinesisch-kanadischer Romane untersucht.
In der Einleitung wird die Relevanz der Themenbereiche Intersektionalität und asiatisch-kanadische Literatur erörtert. Das darauffolgende Kapitel bietet einen historischen Überblick über die chinesisch-kanadische Einwanderung und geht detailliert auf die literarischen Produktionen ein. Es wird aufgezeigt, dass, obwohl kulturelle Güter auch zur Artikulation von Ungleichheitsverhältnissen aufgrund von zugeschriebener ethnischer Zugehörigkeit entstehen, ein Diversifizierungsbestreben innerhalb der literarischen Gemeinschaft von chinesisch-kanadischen Autor:innen identifiziert werden kann. Das dritte Kapitel widmet sich dem Begriff „Intersektionalität“ und stellt, nach einer historischen Einordnung des Konzeptes mit seinen Ursprüngen im Black Feminism, Intersektionalität als bindendes Element zwischen Postkolonialismus, Diversität und Empowerment dar – Konzepte, die für die Analyse (kanadischer) Literatur in dieser Dissertation von besonderer Relevanz sind. Anschließend wird die Rolle von Intersektionalität in der Literaturwissenschaft aufgegriffen. Die darauffolgenden exemplarischen Analysen von Kim Fus For Today I Am a Boy, Wayson Choys The Jade Peony und Yan Lis Lily in the Snow veranschaulichen die vorangegangen methodischen Überlegungen. Allen drei Romanen vorangestellt ist die Kontextualisierung des jeweiligen Werkes als chinesisch-kanadisch, aber auch bisher vorgenommene Überlegungen, die diese Einordnung infrage stellen. Nach einer Zusammenfassung des Inhalts folgt eine intersektionale Analyse auf der inhaltlichen Ebene, die in den familiären und weiteren sozialen Bereich unterteilt ist, da sich die Hierarchiemechanismen innerhalb dieser Bereiche unterscheiden oder gegenseitig verstärken, wie aus den Analysen hervorgeht. Anschließend wird die formale Analyse mit einem intersektionalen Schwerpunkt in einem separaten Unterkapitel näher beleuchtet. Ein drittes Unterkapitel widmet sich einem dem jeweiligen Roman spezifischen Aspekt, der im Zusammenhang mit einer intersektionalen Analyse von besonderer Relevanz ist. Die Arbeit schließt mit einem übergreifenden Fazit, welches die wichtigsten Ergebnisse aus der Analyse zusammenfasst und mit weiteren Überlegungen zu den Implikationen dieser Dissertation, vor allem im Hinblick auf sogenannte kanadische „master narratives“, die eine weitreichende, kontextuelle Relevanz für das Arbeiten mit literarischen Texten aufweisen und durch einen intersektionalen literarischen Ansatz in Zukunft gegebenenfalls gewinnbringend ergänzt werden können.
While humans find it easy to process visual information from the real world, machines struggle with this task due to the unstructured and complex nature of the information. Computer vision (CV) is the approach of artificial intelligence that attempts to automatically analyze, interpret, and extract such information. Recent CV approaches mainly use deep learning (DL) due to its very high accuracy. DL extracts useful features from unstructured images in a training dataset to use them for specific real-world tasks. However, DL requires a large number of parameters, computational power, and meaningful training data, which can be noisy, sparse, and incomplete for specific domains. Furthermore, DL tends to learn correlations from the training data that do not occur in reality, making DNNs poorly generalizable and error-prone.
Therefore, the field of visual transfer learning is seeking methods that are less dependent on training data and are thus more applicable in the constantly changing world. One idea is to enrich DL with prior knowledge. Knowledge graphs (KG) serve as a powerful tool for this purpose because they can formalize and organize prior knowledge based on an underlying ontological schema. They contain symbolic operations such as logic, rules, and reasoning, and can be created, adapted, and interpreted by domain experts. Due to the abstraction potential of symbols, KGs provide good prerequisites for generalizing their knowledge. To take advantage of the generalization properties of KG and the ability of DL to learn from large-scale unstructured data, attempts have long been made to combine explicit graph and implicit vector representations. However, with the recent development of knowledge graph embedding methods, where a graph is transferred into a vector space, new perspectives for a combination in vector space are opening up.
In this work, we attempt to combine prior knowledge from a KG with DL to improve visual transfer learning using the following steps: First, we explore the potential benefits of using prior knowledge encoded in a KG for DL-based visual transfer learning. Second, we investigate approaches that already combine KG and DL and create a categorization based on their general idea of knowledge integration. Third, we propose a novel method for the specific category of using the knowledge graph as a trainer, where a DNN is trained to adapt to a representation given by prior knowledge of a KG. Fourth, we extend the proposed method by extracting relevant context in the form of a subgraph of the KG to investigate the relationship between prior knowledge and performance on a specific CV task. In summary, this work provides deep insights into the combination of KG and DL, with the goal of making DL approaches more generalizable, more efficient, and more interpretable through prior knowledge.