OPUS - Volltextserver Universität Passau
Not a member yet
1021 research outputs found
Sort by
Using Context for Credit Card Fraud Detection
No full textOnline payment fraud is one of the biggest challenges accompanying the ubiquitous adoption of digital payment methods. The academic literature shows that providing data-driven models with additional context of a transaction results in significant improvements in fraud detection performance. However, the methods used to generate suitable context representations often rely on human expert knowledge, which is expensive and suffers from several limitations.
In this thesis, we propose different methods to automate this process by learning these context representations end-to-end on the fraud detection objective. Each of these methods is evaluated on millions of real-world transactions from Worldline, our industrial partner.
Central to this thesis is our proposal of the Neural Aggregate Generator (NAG), a neural network that learns context representations automatically. The architecture of the NAG is designed to resemble the structure of expert feature aggregates, while also addressing their limitations. Our evaluation of the NAG reveals that it outperforms both approaches that use expert aggregates and other end-to-end methods across several months of testing. A thorough evaluation shows that the NAG improves over other approaches on several key factors including model size and robustness to shorter sequences.
We propose several extensions to the NAG with the dual motive of improved alignment with expert aggregates and improved expressiveness. Our evaluation of these extensions shows comparable performance to the NAG with ancillary benefits in terms of prospective interpretability and model size.
We also introduce the novel paradigm of using \lq future' transactions as context. Our analysis of real-world data from Worldline shows that verification of transactions are often delayed by several days and that within this delay there are often several transactions booked on the card which can be used as additional context. We show that this future context improves the performance of sequence models. Moreover, we also show that a balance between past and future context yields the best results and that using future context allows the use of shorter sequences overall.
Beyond context-based fraud detection, we also provide an initial proposal of generating synthetic credit card data using Generative Adversarial Networks (GANs), showing that a Wasserstein GAN can be used to generated synthetic data similar to a popular publicly available credit card fraud dataset.
We also describe several possible directions for future work including the incorporation of a adapted self-attention mechanism to the NAG and the use of transformers for synthetic data generation.Betrug bei Online-Zahlungen ist eine der größten Herausforderungen bei der weitverbreiteten Einführung digitaler Zahlungsmethoden. Die akademische Literatur zeigt, dass die Bereitstellung von datengesteuerten Modellen mit zusätzlichem Transaktionskontext zu einer erheblichen Verbesserung der Betrugserkennungsleistung führt. Die Methoden zur Generierung geeigneter Kontextrepräsentationen beruhen jedoch häufig auf menschlichem Expertenwissen, was teuer ist und einige Einschränkungen mit sich bringt.
In dieser Arbeit schlagen wir verschiedene Methoden vor, um diesen Prozess zu automatisieren, indem diese Kontextrepräsentationen end-to-end auf das Ziel der Betrugserkennung hin erlernt werden. Jede dieser Methoden wird anhand von Millionen echter Transaktionen von Worldline, unserem industriellen Partner, bewertet.
Zentral in dieser Arbeit ist unser Vorschlag des Neural Aggregate Generator (NAG), ein neuronales Netzwerk, das Kontextrepräsentationen automatisch erlernt. Die Architektur des NAGs ist so konzipiert, dass sie der Struktur von Expertenaggregaten ähnelt und gleichzeitig deren Einschränkungen adressiert. Unsere Bewertung des NAGs zeigt, dass es sowohl Ansätze, die Expertenaggregate verwenden, als auch andere end-to-end Methoden in mehreren Monaten des Testens übertrifft. Eine tiefergehende Bewertung zeigt, dass das NAG in mehreren wichtigen Faktoren, einschließlich der Modellgröße und der Robustheit gegenüber kürzeren Sequenzen, bessere Ergebnisse erzielt als andere Ansätze.
Wir schlagen mehrere Erweiterungen des NAGs vor, die das doppelte Ziel verfolgen, eine bessere Übereinstimmung mit Expertenaggregaten und eine verbesserte Ausdruckskraft zu erreichen. Unsere Bewertung dieser Erweiterungen zeigt eine vergleichbare Leistung zum NAG mit zusätzlichen Vorteilen hinsichtlich der zukünftigen Interpretierbarkeit und der Modellgröße.
Wir führen auch das neuartige Paradigma der Nutzung zukünftiger Transaktionen als Kontext ein. Unsere Analyse der Daten von Worldline zeigt, dass die Verifizierung von Transaktionen oft um mehrere Tage verzögert ist und dass während dieser Verzögerung oft mehrere Transaktionen auf der Karte gebucht werden, die als zusätzlicher Kontext verwendet werden können. Wir zeigen, dass dieser „zukünftige“ Kontext die Leistung von Sequenzmodellen verbessert. Darüber hinaus zeigen wir, dass ein Gleichgewicht zwischen vergangenem und zukünftigem Kontext die besten Ergebnisse liefert und dass die Nutzung zukünftigen Kontexts insgesamt kürzere Sequenzen ermöglicht.
Über die kontextbasierte Betrugserkennung hinaus geben wir auch einen ersten Vorschlag zur Generierung synthetischer Kreditkartendaten mittels Generative Adversarial Networks (GANs) und zeigen, dass ein durch einen Wasserstein-GAN generierter Datensatz geeignet ist, um darauf Erkennungsmodelle zu trainieren.
Wir beschreiben auch mehrere mögliche Richtungen für zukünftige Arbeiten, einschließlich der Integration eines angepassten Self-Attention Mechanismus in das NAG und der Verwendung von Transformern zur Generierung synthetischer Daten
User Simulation in Interactive Information Retrieval : methods and frameworks for simulating complex search behavior
No full textModern information retrieval (IR) systems, including web search engines and digital libraries, face challenges in simulating realistic user search behavior. Evolving interaction patterns and the integration of AI-powered interfaces make these challenges even harder. Traditional evaluation methods struggle to capture the dynamic nature of user interactions, particularly in complex search tasks and multi-stage information-seeking processes. User simulation offers a promising solution, providing a controlled environment for experimentation and allowing customization to model specific user behaviors and task contexts.
This research develops advanced techniques for user simulation in IR, creating more realistic and dynamic models than were previously possible. Key contributions include new methods for representing query reformulation, modeling how information needs change, and measuring the impact of different search environments on simulated user behavior. Specifically, this work introduces contextual Markov models, cognitive state models, and embedding space alignment techniques to accurately represent interactive search behavior.
Beyond model development, new evaluation methods and metrics are proposed for assessing the quality of simulated search sessions. These include statistical comparisons of session characteristics and classification-based approaches to distinguish between simulated and real user behavior. Additionally, this work leverages emerging technologies, such as large language models (LLMs) and retrieval-augmented generation, to improve the realism of user search behavior simulation.
The practical outcome of this research is a modular and extensible simulation framework. This framework incorporates advanced techniques like user type-specific Markov models, advanced query generation using LLMs, and conversational user models
Protection of Civilians in Armed Conflict: Safe Areas and the Silencing of Mass Atrocities
No full textThis publication-based thesis approaches the topic of the protection of civilians in armed conflict by ‘zooming in‘ on two specific sub-topics: safe areas as well as the silencing of mass atrocities. The thesis consists of five publications (four of them published in double-blind peer reviewed journals) and of an introductory chapter that presents the overall argument and contextualises the publications. Two publications argue that mass atrocities are silenced, or at least not politicised, in the discourses and debates of Germany as an exemplary UN member state. It is argued that ‘silencing’ is a structural feature of an ‘identity-mismatch’ with the domestic ideational structure that inhibits debates and freezes the possibility space for foreign policy. Empirically, the first publication assesses the rhetoric of the German chancellor, foreign ministers and parliamentary group leaders vis-à-vis the mass atrocities committed in Yemen, Myanmar and South Sudan. The second publication widens the scope and looks at German political, media and societal debates in twelve cases of mass atrocities between 1992 and 2019. The remaining three publications focus on the topic of safe areas. The first publication systematically collects and assesses the existing conceptual literature on safe areas. The second publication presents a comprehensive definition, a four-fold typology based on a distinction by size and the logic of protection as well as an extensive empirical dataset of safe areas. By analysing case studies from Iraq and South Sudan, the third publication argues that safe areas have a very limited potential to provide an alternative to flight, especially from the perspective of the protection-seeking civilians themselves
The antinomian framework of Western and non-Western democracies — from theory to application
No full textThis article argues in favour of broadening the classical paradigm of democracy that historically emerged in the West. Without disregarding the undiminished significance of the idea of liberal democracy and its deep commitment to universal human rights, the necessity of a new accentuation of the concept of democracy is accepted in view of the observable democratic processes in non-Western societies. This should help to avoid both the blind spots of Eurocentrism and the misperception of countries as democracies that are merely masking their authoritarian or despotic character. The result is a theory of popular sovereignty that seeks to grasp and combine Western and non-Western conceptions of democracy in a balanced way and is based on a genealogical and comparative perspective of the history of political ideas in the global North and South. In a second step, the theory is then tested using a few selected examples (Confucian democracy, Islamic democracy, African democracy)
Populäre Artikulationen - Artikulationen des Populären
No full textZuerst erschienen auf: https://www.kultursemiotik.com/forschung/publikationen/schriftenreihe-online/
Inhalt
Vorwort: Populäres und Artikulationen – ein Überblick
11. Jahrestagung der AG Populärkultur und Medien in der Gesellschaft für Medienwissenschaft (GfM)
André Rottgeri, Günter Koch
I. Populäre Artikulationen –
Artikulationen des Populären
Kulturwissenschaftliche Anmerkungen aus der Perspektive der Populären Kultur
Barbara Hornberger
Populäre Artikulationen – Artikulationen des Populären
Sprachwissenschaftliche Perspektiven
Rüdiger Harnisch
II. Musikalische Artikulationen – Popular Music Studies
Populäre Musik & Artikulationen
Entwurf einer Typologie am Beispiel von Joe Satriani
André Rottgeri
Between Salvador and the Recôncavo baiano
Value disputes stemming from the rock bands Cascadura and Escola Pública
Jorge Cardoso Filho
Emphatische Dehnungen und sprechende Pausen
Über rhythmische Artikulationsmuster deutschsprachiger Songtexte am Beispiel der Wise Guys
Johannes Odendahl
Von der Berliner Mauer nach Moskau und Lateinamerika
Wind of Change von den Scorpions auf Russisch und Spanisch
Christina Richter-Ibáñez
III. Text- und Bild-Artikulationen – Literaturwissenschaft und Mediensemiotik
Wertevermittlung in Online-Storytelling-Werbespots der Firma Edeka
Dennis Gräf
Dekonstruktion eines historischen Mythos in Sofia Coppolas MARIE ANTOINETTE (2006)
Oliver Seidel
Zwischen Pille & Minirock
Feminismus und Sexismus als populäre Konzepte der Zeitschrift konkret der 1960er Jahre
Alix Michell
Artikulation als Verrat?
Rotwelsch und die Verweigerung der Übersetzbarkeit in Zeena Parkins’ Mouth=Maul=Betrayer
Benedikt Wolf
In Likes We Trust oder die unmögliche Möglichkeit vom Like als Gabe zu sprechen
Christian Schulz
Papst und Populus im Hochmittelalter
Papsturkunden als Artikulationen des päpstlichen Universalprimats
Daniel Berger/André Rottgeri
IV. Sprachliche Artikulationen – Sprachwissenschaft
Verhörer in der Popmusik
Franz Josef Bauer
WER WIRD MILLIONÄR?
Das Sprachspiel als populärer Einstieg in den Wissenswettbewerb
Günter Koch
Die Autorinnen und Autore
Reisebegegnungen in Marokko
No full textDie Arbeit beleuchtet Marokko als Reiseziel für Individualreisende aus landeskundlicher und tourismusgeographischer Perspektive. Darauf aufbauend erfolgt eine Analyse der touristischen Praxis anhand von Begegnungsrepräsentationen in verschiedenen medialen Formaten von Reiseberichten, mit Fokus auf interkulturelle Begegnungen. Diese zeigen sich in ihrer Darstellung als vielfältig und differenziert. Durch eine multiperspektivische Betrachtung des Phänomens werden komplexe Spielarten interkultureller Begegnung in Bezug auf Darstellung, Pragmatik und Funktionalität herausgearbeitet
„Insgesamt sah ich wirklich die Wellen an mir vorbeirauschen, mich darin befindend“ - Die Folgen negativer Reaktionen gegen Wissenschaftler*innen in den sozialen Medien
No full textWissenschaftler*innen werden immer häufiger zum Ziel von Kritik, Anfeindungen und Ähnlichem in den sozialen Medien. Die Studie nimmt die reziproken Effekte dieser negativen Reaktionen in den Blick. Konkret wird untersucht, mit welchen Arten negativer Reaktionen Wissenschaftler*innen konfrontiert werden, welche mentalen Verarbeitungs- und Bewertungsprozesse in diesem Kontext relevant sind und wie die Betroffenen mit den negativen Reaktionen anschließend umgehen, um die entstandene Belastungssituation zu bewältigen. Neben der theoretischen Aufarbeitung der Thematik wird dafür auf qualitative Leitfadeninterviews mit betroffenen Wissenschaftler*innen zurückgegriffen. Die Erkenntnisse der Studie werden in einem „Modell der Konfrontation mit negativen Reaktionen in den sozialen Medien“ zusammengeführt. Das Modell ermöglicht zum einen die strukturierte Beschreibung entsprechender Vorfälle und kann zum anderen als theoretisch-konzeptionelle Grundlage für weitere empirische Untersuchungen dienen
Towards Data Efficiency and Controllable Representations for Deep Learning in Resource-Constrained Domains
No full textThe deployment of artificial intelligence (AI) in specialised domains such as planetary science and healthcare, as well as in low-resource NLP settings, faces two fundamental challenges: label scarcity and data scarcity. Label scarcity stems from the high cost of expert annotation, the scarcity of domain experts, and the infeasibility of crowdsourcing, particularly in complex tasks requiring specialised knowledge. In parallel, data scarcity stems from the inherent difficulty of acquiring sufficient raw data, whether due to limited observational opportunities, environmental and technical barriers, or stringent privacy constraints.
Together, these limitations impede the broader adoption of AI in these fields. Many existing approaches to label efficiency, such as active learning, rely on problem-specific heuristics and often, as a design choice, employ naive uncertainty estimations—typically at the instance level.
However, such methods can lead to redundant or suboptimal sample selection by ignoring structural data properties and failing to account for representational diversity. In practice, they often perform no better than random sampling. For data synthesis, generative models face their own set of challenges. Despite their promise for synthetic data generation, these models frequently lack mechanisms to disentangle generative factors at the representation level, limiting their controllability. Additionally, standardised evaluation metrics to assess the quality of disentanglement remain underdeveloped, limiting their practical utility. These limitations highlight the need for advancements in data-efficient machine learning and controllable generative modelling, focusing on domain-specific validity and rigorous evaluation.
This thesis contributes to addressing these challenges by proposing tailored solutions in two key directions.
First, for data-efficient learning, a deep active learning (DAL) framework is introduced to enhance label efficiency by prioritising the most informative samples for annotation. Unlike traditional per-sample approaches, this framework aggregates uncertainty across larger data segments—such as orbital intervals in planetary science—allowing it to capture contextual variations. This method reduces labelled data requirements by up to 90% in the case of boundary crossing detection at Mercury’s magnetosphere. To further improve sampling diversity, a GAN-based concept drift detection method is integrated into the DAL framework, leveraging uncertainty and diversity together to offer a sampling method that outperforms random sampling. Additionally, foundation models such as the Segment Anything Model (SAM) are employed for zero-shot annotation to generate high-quality pseudo-labels, which are subsequently used to train a domain-specific model via knowledge distillation. This approach significantly enhances data efficiency, reducing the need for annotated samples several times over in the tested scenario of image segmentation for geological mapping. Furthermore, large language models (LLMs) are explored as active annotators for linguistic tasks in low-resource languages, achieving near-baseline performance while reducing annotation costs by up to 40x.
Second, the thesis investigates methods to induce controllability in generative models, enabling the production of high-fidelity, controllable synthetic data. Conditional generative adversarial networks (CGANs) and disentangled representation learning techniques (DRL) are explored, particularly in the context of pedestrian trajectory prediction in the mobility domain, where controlled synthesis of diverse motion patterns is critical. Additionally, the work examines existing metrics for evaluating disentanglement and identifies critical limitations in them. A novel metric, the Exclusivity Disentanglement Index (EDI), is proposed as an improved standardised measure. Based on the principle of exclusivity in factor-code relationships, this metric offers advantages over existing alternatives in terms of efficiency and robustness.
By advancing data-efficient learning and controllable generation strategies, this thesis aims to bridge the gap between AI’s vast potential and its practical adoption in resource-constrained environments. These contributions pave the way for transformative applications in planetary science, healthcare, and beyond, where label and data scarcity have long been barriers to progress
Lehramtsstudierende auf ihrem Weg der Professionalisierung
No full textDie vorliegende Studie untersucht subjektive Vorstellungen von Professionalisierung aus Sicht Studierender des Lehramts Grundschule an der Universität Passau im Rahmen unterschiedlicher Praktikumsformen: Pädagogisch-didaktisches Praktikum (PDP), Exercitium Paedagogicum (ExPaed) und Modellcurriculum (MC). Aus der Theorie heraus wurden zwei Hauptkategorien der Studie abgeleitet: Kategorie I - Inwiefern unterscheiden sich die subjektiven Sichtweisen Studierender unterschiedlicher Praktika (PDP, ExPead, MC) auf Professionalisierung? Und Kategorie II - Inwiefern unterscheidet sich die systematische Reflexion der individuellen Professionalisierung von Studierenden unterschiedlicher Praktika (PDP, ExPaed, MC)? Durch diese drei Gruppierungen ist es möglich, verschiedene Perspektiven zu berücksichtigen und Unterschiede herauszuarbeiten. Die leitfaden-gestützten ExpertInneninterviews (n = 27: 8 PDP, 10 ExPaed, 9 MC) wurden mit dem Programm f4 transkribiert und mit Hilfe der Datenanalyse-Software MAXQDA (2020/2022) inhaltsanalytisch nach Mayring (2015) ausgewertet.
Ergebnisse der Studie zeigen Abweichungen der einzelnen Studierendengruppen bezüglich biografischer Ausgangslagen für Professionalisierung. StudentInnen der einzelnen Gruppierungen nehmen Professionalisierung im systemisch-strukturellen Rahmen in Form von Praktika anders war, zeigen Unterschiede im professionellen Selbstverständnis und in den subjektiven Einflussfaktoren für Professionalisierung. Auch in der systematischen Reflexion der eigenen Professionalisierung können Diskrepanzen eruiert werden
Multi-Leader Congestion Games with an Adversary
No full textIn this thesis, we introduced a congestion game with multiple leaders and a single follower (adversary) which is motivated by security applications with congestion effects. Our objective was to understand the result and the impact of selfish acting individuals in these games. In this regard, we analyzed the existence, the computation and the quality of (approximate) pure Nash equilibria.
First, we observed that an exact pure Nash equilibrium always exists in the resulting strategic game among the leaders if the resource cost coefficients are identical and the underlying congestion game is a matroid congestion game. If one of these two conditions is not fulfilled, the existence of PNE is not ensured anymore in general. Consequently, we focused on approximate equilibria. For the case of symmetric singleton strategies, one of our main result established that K ≈ 1.1974, the unique solution of a cubic polynomial equation, is the smallest possible factor such that the existence of a K-approximate equilibrium is guaranteed for all instances of the game. To this end, we presented an efficient algorithm which computes a K-approximate PNE. Furthermore, we showed that the factor K is tight by providing an instance where no α-approximate PNE with α < K exists. However, for a specific symmetric singleton instance there might be a better α-approximate
PNE, i.e., with α < K. A given instance could even admit an exact PNE. We
provided therefore a polynomial time procedure that computes a best approximate PNE of a given instance. In particular, this procedure can verify the existence of an exact PNE in a given instance efficiently and, if it exists, can also determine the corresponding load vector. Finally, for symmetric singleton instances with two resources, we compared the total cost of a best (cheapest) and worst (most expensive) PNE to the total cost of an optimal outcome, termed by the price of stability and the price of anarchy, respectively. In particular, we verified that the PoS and the PoA are 4/3