TY - THES A1 - Auer, Michael T1 - Improving Automated Android Test Generation N2 - Mobile apps are nowadays the preferred means to accomplish ubiquitous tasks like messaging, e-commerce and even playing games. Often, there exist multiple apps for the same purpose, and it is the choice of the end user to pick an appropriate app. Apps that behave unexpected, e.g., crash frequently, are sooner or later replaced, which isundesirable for the companies developing such apps. Thus, it is essential to tests apps properly before they are released onto the market. However, testing manually is often not only too cost-intensive but also too time-consuming in the short development phase, thus an automated solution is preferred. Testing mobile apps automatically received increased attention in the last decade from primarily people in academia, and several testing techniques evolved. One technique that yielded promising results, especially in different domains, is search-based software testing in which a metaheuristic, e.g., a genetic algorithm, is applied to solve an optimisation problem, e.g., test generation. A main objective of test generation is to produce tests that reveal as many faults as possible. This in turn requires the generation of tests that deeply explore the tested app. The core metric to quantify how much code tests cover is the measurement of code coverage, which can be computed at different levels of granularity ranging from determining the fraction of covered activities to a very fine-grained measurement that calculates the percentage of covered lines. This coverage information is then often used to guide the search of the employed metaheuristic. However, current automated test generation approaches produce tests with a rather low code coverage. Thus, a substantial part of tested apps remains unexplored, which in turn misses revealing deeply residing faults. We identified three core issues that are directly related to the generation of low-coverage tests. First, the applicability of current test generators is often limited. This comprises the fact that current state-of-the-art code coverage tools are incapable of instrumenting a substantial number of apps and consequently, test generators cannot utilise detailed coverage information during exploration. In addition, test generators are often only equipped with a primitive set of actions that are insufficient to simulate system events and complex user inputs. Second, the test execution is extremely time-consuming. This includes among other things the overhead associated with executing individual actions, intermediate restart operations as well as fitness evaluations. Since search-based algorithms require a substantial number of test executions to play out their strengths, the slow test execution impedes the effectiveness of the search. Third, the guidance offered by search-based algorithms is often hampered by applying inadequate fitness functions or by using non-representation-specific variation operators. In this thesis we address the problem of low-coverage tests in the Android domain by proposing several enhancements for the three identified core issues. Concerning the applicability problem, we provide the implementation of a robust code coverage tool that is capable of measuring coverage at different levels of granularity and requires no access to the source code. We also propose to include actions that can simulate system events as well as complex user inputs. Regarding the performance issue, we suggest the integration of a surrogate model that is capable of predicting the outcome of individual actions or complete tests over time in order to reduce the overall test execution costs. With respect to the lack of guidance offered by traditional search-based algorithms, we suggest alternative search strategies. In the case of a deceptive fitness landscape, we propose using novelty search algorithms. Alternatively, we suggest utilising estimation of distribution algorithms that require no crossover or mutation perators to sample new tests. While all those enhancements had a positive impact on the Android test generation process, the individual empirical studies highlighted that further research is necessary to unleash the full power of the proposed search-based algorithms. In particular, exploring complex user interfaces meaningfully requires more attention whether by introducing additional actions or by extracting valuable hints to infer reasonable text inputs. In addition, the guidance offered by fitness functions is often limited because they are either designed too coarse at all or do not accurately reflect the search objectives. Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-14955 ER - TY - JOUR A1 - Lengler, Johannes A1 - Opris, Andre A1 - Sudholt, Dirk T1 - Analysing Equilibrium States for Population Diversity JF - Algorithmica (ISSN: 1432-0541) N2 - Population diversity is crucial in evolutionary algorithms as it helps with global exploration and facilitates the use of crossover. Despite many runtime analyses showing advantages of population diversity, we have no clear picture of how diversity evolves over time. We study how the population diversity of (μ+1)algorithms, measured by the sum of pairwise Hamming distances, evolves in a fitness-neutral environment. We give an exact formula for the drift of population diversity and show that it is driven towards an equilibrium state. Moreover, we bound the expected time for getting close to the equilibrium state. We find that these dynamics, including the location of the equilibrium, are unaffected by surprisingly many algorithmic choices. All unbiased mutation operators with the same expected number of bit flips have the same effect on the expected diversity. Many crossover operators have no effect at all, including all binary unbiased, respectful operators. We review crossover operators from the literature and identify crossovers that are neutral towards the evolution of diversity and crossovers that are not. KW - Evolutionary algorithms KW - Runtime analysis KW - Diversity KW - Population dynamics Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2406282100292.409184699704 SN - 0178-4617 SN - 1432-0541 VL - 86 IS - 7 SP - 2317 EP - 2351 PB - Springer US CY - New York ER - TY - JOUR A1 - Schulte, Lukas A1 - Ledel, Benjamin A1 - Herbold, Steffen T1 - Studying the explanations for the automated prediction of bug and non-bug issues using LIME and SHAP JF - Empirical Software Engineering (ISSN: 1573-7616) N2 - Context The identification of bugs within issues reported to an issue tracking system is crucial for triage. Machine learning models have shown promising results for this task. However, we have only limited knowledge of how such models identify bugs. Explainable AI methods like LIME and SHAP can be used to increase this knowledge. Objective We want to understand if explainable AI provides explanations that are reasonable to us as humans and align with our assumptions about the model’s decision-making. We also want to know if the quality of predictions is correlated with the quality of explanations. Methods We conduct a study where we rate LIME and SHAP explanations based on their quality of explaining the outcome of an issue type prediction model. For this, we rate the quality of the explanations, i.e., if they align with our expectations and help us understand the underlying machine learning model. Results We found that both LIME and SHAP give reasonable explanations and that correct predictions are well explained. Further, we found that SHAP outperforms LIME due to a lower ambiguity and a higher contextuality that can be attributed to the ability of the deep SHAP variant to capture sentence fragments. Conclusion We conclude that the model finds explainable signals for both bugs and non-bugs. Also, we recommend that research dealing with the quality of explanations for classification tasks reports and investigates rater agreement, since the rating of explanations is highly subjective. KW - Explainable AI KW - LIME KW - SHAP KW - Issue type prediction Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2409232103207.812648424894 SN - 1382-3256 SN - 1573-7616 VL - 29 IS - 4 PB - Springer US CY - New York ER - TY - JOUR A1 - Aistleitner, Christoph A1 - Frühwirth, Lorenz A1 - Prochno, Joscha T1 - Diophantine conditions in the law of the iterated logarithm for lacunary systems JF - Probability Theory and Related Fields (ISSN: 1432-2064) N2 - It is a classical observation that lacunary function systems exhibit many properties which are typical for systems of independent random variables. However, it had already been observed by Erdős and Fortet in the 1950s that probability theory’s limit theorems may fail for lacunary sums (sum f(n_k x)) if the sequence ((n_k)_{k ge 1}) has a strong arithmetic “structure”. The presence of such structure can be assessed in terms of the number of solutions k, l of two-term linear Diophantine equations (an_k - bn_l = c). As the first author proved with Berkes in 2010, saving an (arbitrarily small) unbounded factor for the number of solutions of such equations compared to the trivial upper bound, rules out pathological situations as in the Erdős–Fortet example, and guarantees that (sum f(n_k x)) satisfies the central limit theorem (CLT) in a form which is in accordance with true independence. In contrast, as shown by the first author, for the law of the iterated logarithm (LIL) the Diophantine condition which suffices to ensure “truly independent” behavior requires saving this factor of logarithmic order. In the present paper we show that, rather surprisingly, saving such a logarithmic factor is actually the optimal condition in the LIL case. This result reveals the remarkable fact that the arithmetic condition required of ((n_k)_{k ge 1}) to ensure that (sum f(n_k x)) shows “truly random” behavior is a different one at the level of the CLT than it is at the level of the LIL: the LIL requires a stronger arithmetic condition than the CLT does. KW - Lacunary trigonometric sums KW - Law of the iterated logarithm KW - Diophantine equations Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2406190937284.886045166092 SN - 0178-8051 SN - 1432-2064 VL - 192 IS - 1 SP - 545 EP - 574 PB - Springer CY - Berlin/Heidelberg ER - TY - JOUR A1 - Beurskens, Michael A1 - Scherzinger, Stefanie T1 - Legal Perspectives on Research Data Storage JF - Datenbank-Spektrum (ISSN: 1618-2162) N2 - Responsibly managing research data has become increasingly important for researchers, especially within the database research community. Despite significant progress in best practices, the state-of-the-art in research data storage is lacking from a legal perspective. We introduce the stakeholders and the dynamic nature of their relationships (such as researchers changing affiliation) and observe that no existing infrastructure for research data storage fully meets their requirements. Therefore, we emphasize the need to design a comprehensive system architecture for research data storage that is aligned with legal considerations from the start, rather than as an afterthought. KW - Storage Contract KW - Copyright KW - Sui Generis Right KW - Privacy KW - Trade Secrets KW - Research Data Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2410012125227.647518861484 SN - 1618-2162 SN - 1610-1995 VL - 24 IS - 2 SP - 85 EP - 95 PB - Springer CY - Berlin/Heidelberg ER - TY - JOUR A1 - Franke, Jan A1 - Heinrich, Florian A1 - Reisch, Raven T. T1 - Vision based process monitoring in wire arc additive manufacturing (WAAM) JF - Journal of Intelligent Manufacturing (ISSN: 1572-8145) N2 - A stable welding process is crucial to obtain high quality parts in wire arc additive manufacturing. The complexity of the process makes it inherently unstable, which can cause various defects, resulting in poor geometric accuracy and material properties. This demands for in-process monitoring and control mechanisms to industrialize the technology. In this work, process monitoring algorithms based on welding camera image analysis are presented. A neural network for semantic segmentation of the welding wire is used to monitor the working distance as well as the horizontal position of the wire during welding and classic image processing techniques are applied to capture spatter formation. Using these algorithms, the process stability is evaluated in real time and the analysis results enable the direction independent closed-loop-control of the manufacturing process. This significantly improves geometric fidelity as well as mechanical properties of the fabricated part and allows the automated production of parts with complex deposition paths including weld bead crossings, curvatures and overhang structures. KW - Wire arc additive manufacturing KW - Vision based monitoring KW - Machine learning KW - Nozzle-to-work distance monitoring KW - Contact tube wear off detection KW - Spatter detection Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2405052141549.085870766506 SN - 0956-5515 SN - 1572-8145 VL - 36 IS - 3 SP - 1711 EP - 1721 PB - Springer US CY - New York ER - TY - JOUR A1 - Andraschko, Bernhard A1 - Danner, Julian A1 - Kreuzer, Martin T1 - SAT Solving Using XOR-OR-AND Normal Forms JF - Mathematics in Computer Science (ISSN: 1661-8289) N2 - This paper introduces the XOR-OR-AND normal form (XNF) for logical formulas. It is a generalization of the well-known Conjunctive Normal Form (CNF) where literals are replaced by XORs of literals. As a first theoretic result, we show that every CNF formula is equisatisfiable to a formula in 2-XNF, i.e., a formula in XNF where each clause involves at most two XORs of literals. Subsequently, we present an algorithm which converts Boolean polynomials efficiently from their Algebraic Normal Form (ANF) to formulas in 2-XNF. Experiments with the cipher ASCON-128 show that cryptographic problems, which by design are based strongly on XOR-operations, can be represented using far fewer variables and clauses in 2-XNF than in CNF. In order to take advantage of this compact representation, new SAT solvers based on input formulas in 2-XNF need to be designed. By taking inspiration from graph-based 2-CNF SAT solving, we devise a new DPLL-based SAT solver for formulas in 2-XNF. Among others, we present advanced pre- and in-processing techniques. Finally, we give timings for random 2-XNF instances and instances related to key recovery attacks on round reduced ASCON-128, where our solver outperforms state-of-the-art alternative solving approaches. KW - - KW - SAT solving KW - XOR constraint KW - Algebraic normal form KW - Implication graph KW - Cryptographic attack KW - 03B70 KW - 13P15 KW - 05C90 KW - 94A60 Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2502242130589.859216082138 SN - 1661-8270 SN - 1661-8289 VL - 18 IS - 4 PB - Springer International Publishing CY - Cham ER - TY - JOUR A1 - Kosch, Harald A1 - Brunie, Lionel A1 - Mayer, Tobias A1 - Hasan, Omar A1 - Schiedermeier, Maximilian T1 - Anonymous voting using distributed ledger-assisted secure multi-party computation JF - Applied Network Science N2 - High voter turnout in elections and referendums is desirable to ensure a robust democracy. Secure electronic voting is a vision for the future of elections and referendums. Such a system can counteract factors hindering strong voter turnout such as the requirement of physical presence during limited hours at polling stations. However, this vision brings transparency and confidentiality requirements that render the design of such solutions challenging. Specifically, the counting implementation must support reproducibility, and the choice of individual voters must remain confidential. In this paper, we propose and evaluate a novel referendum protocol that ensures transparency, confidentiality, and integrity, in trustless networks. The protocol is built by combining secure multi-party computation and distributed ledger technology, e.g., a Blockchain. The persistence and immutability of the protocol communication allow verifiability of the referendum outcome by any participant. Voters therefore do not need to trust third parties. We provide a formal description and conduct a thorough security evaluation of our proposal. KW - E-voting KW - Distributed ledger KW - Trustless networks KW - Transparency Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15176 VL - 2024 IS - 9 PB - Springer International Publishing CY - Cham (Schweiz) ER - TY - JOUR A1 - Ghodselahi, Abdolhamid A1 - Kuhn, Fabian T1 - Toward Online Mobile Facility Location on General Metrics JF - Theory of Computing Systems N2 - We introduce an online variant of mobile facility location (MFL) (introduced by Demaine et al. (SODA 258–267 2007)). We call this new problem online mobile facility location (OMFL). In the OMFL problem, initially, we are given a set of k mobile facilities with their starting locations. One by one, requests are added. After each request arrives, one can make some changes to the facility locations before the subsequent request arrives. Each request is always assigned to the nearest facility. The cost of this assignment is the distance from the request to the facility. The objective is to minimize the total cost, which consists of the relocation cost of facilities and the distance cost of requests to their nearest facilities. We provide a lower bound for the OMFL problem that even holds on uniform metrics. A natural approach to solve the OMFL problem for general metric spaces is to utilize hierarchically well-separated trees (HSTs) and directly solve the OMFL problem on HSTs. In this paper, we provide the first step in this direction by solving a generalized variant of the OMFL problem on uniform metrics that we call G-OMFL. We devise a simple deterministic online algorithm and provide a tight analysis for the algorithm. The second step remains an open question. Inspired by the k-server problem, we introduce a new variant of the OMFL problem that focuses solely on minimizing movement cost. We refer to this variant as M-OMFL. Additionally, we provide a lower bound for M-OMFL that is applicable even on uniform metrics. KW - Mobile resources KW - Online requests KW - Competitive analysis KW - General cost function KW - Movement minimization Y1 - 2023 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2024022609363705445898 VL - 67 IS - 6 SP - 1268 EP - 1306 PB - Springer Nature CY - Berlin ER - TY - JOUR A1 - von der Heyde, Markus A1 - Gerl, Armin T1 - Entwicklungsstand der CIO-Funktion und hochschulübergreifenden IT-Governance im Kontext der Digitalen Transformation an Hochschulen in Bayern JF - HMD Praxis der Wirtschaftsinformatik N2 - Die Hochschulen befinden sich durch vielfältige Veränderungs-prozesse in Verbindung mit dem Einsatz von Informationstechnologien (IT) aufdem Weg der Digitalen Transformation. Diese Digitale Transformation der Hoch-schulen umfasst intensive Veränderungsprozesse in der gesamten Hochschulkulturin Lehre, Forschung und Verwaltung in übergreifender und strukturierter Weise.Seit vielen Jahren werden vielfältige Digitalisierungsvorhaben zur Modernisierungvon einzelnen Prozessen an den Hochschulen umgesetzt. Die Leitungen der Re-chenzentren leisten mit der Umsetzung von IT-Projekten einen zentralen Beitrag zudiesem Wandel. Mit der Einführung der CIO-Funktion in den Hochschulleitungenund der hochschulübergreifenden Kooperationen hat sich die IT-Governance wei-terentwickelt. Insbesondere für die Digitale Transformation werden Strukturen zurKoordination der übergreifenden Vorhaben benötigt, wobei zusätzlich zur IT-Lei-tung eine Vielzahl von Funktionsträgern mit fachlichen Aufgaben aus Forschung,Lehre und Verwaltung involviert ist. Es stellt sich die Frage, wie die Digitale Trans-formation an Hochschulen gesteuert werden kann und in welcher organisatorischenForm sich die Aufgaben und Verantwortlichkeiten im Hochschulkontext realisierenlassen. An der Weiterentwicklung der IT-Governance an bayerischen Hochschulenwird beispielhaft erläutert, welche übergreifenden Aufgaben der Koordination vonBedarf und Versorgung mit IT-Services zwischen und innerhalb der Hochschulenbestehen. Die CIO-Funktion wird durch die Verankerung in der Leitungsebene derFunktion des Chief Digital Officers (CDO) aus der Wirtschaft ähnlicher, auch wennin Hochschulen aufgrund der klassischen Ressort-Einteilung die Rolle oft als Vize-präsident:in für Digitalisierung bezeichnet wird. KW - Digitale Transformation KW - IT-Governance KW - Hochschulen KW - Bildung KW - CIO KW - CDO Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2022072222053919381076 VL - 2022 IS - 59 SP - 881 EP - 895 PB - Springer Nature CY - Berlin ER - TY - JOUR A1 - Hevia Fajardo, Mario Alejandro A1 - Sudholt, Dirk T1 - Self-adjusting Population Sizes for Non-elitist Evolutionary Algorithms: why Success Rates Matter JF - Algorithmica N2 - Evolutionary algorithms (EAs) are general-purpose optimisers that come with several parameters like the sizes of parent and offspring populations or the mutation rate. It is well known that the performance of EAs may depend drastically on these parameters. Recent theoretical studies have shown that self-adjusting parameter control mechanisms that tune parameters during the algorithm run can provably outperform the best static parameters in EAs on discrete problems. However, the majority of these studies concerned elitist EAs and we do not have a clear answer on whether the same mechanisms can be applied for non-elitist EAs. We study one of the best-known parameter control mechanisms, the one-fifth success rule, to control the offspring population size λ in the non-elitist (1, λ) EA. It is known that the (1, λ) EA has a sharp threshold with respect to the choice of λ where the expected runtime on the benchmark function OneMax changes from polynomial to exponential time. Hence, it is not clear whether parameter control mechanisms are able to find and maintain suitable values of λ. For OneMax we show that the answer crucially depends on the success rates (i. e. a one-(s + 1)-th success rule). We prove that, if the success rate is appropriately small, the self-adjusting (1, λ) EA optimises OneMax in O(n) expected generations and O(n log n) expected evaluations, the best possible runtime for any unary unbiased black-box algorithm. A small success rate is crucial: we also show that if the success rate is too large, the algorithm has an exponential runtime on OneMax and other functions with similar characteristics. KW - Evolutionary algorithms KW - Parameter control KW - Theory KW - Runtime analysis KW - Non-elitism Y1 - 2023 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2023102616463077837436 VL - 86 IS - 2 SP - 526 EP - 565 PB - Springer Nature CY - Berlin ER - TY - JOUR A1 - Gagin, Stepan A1 - Bettermann, Michael A1 - de Meer, Hermann T1 - Multi-vector optimization scheme for distributed components in energy islands JF - e+i Elektrotechnik und Informationstechnik N2 - Recent advancements in energy systems, such as the emergence of prosumers and sector coupling approaches, introduce additional flexibilities over multiple energy sectors, such as heating, electricity, and mobility. Due to the complexity of such distributed systems, the optimization of energy allocation is a non-trivial task, especially considering constraints and limitations introduced by distributed devices or sub-systems. Additionally, the variety of devices forces approaches to be highly situational and not universally applicable. In this paper, a two-level optimization scheme is proposed, which aims at reducing the optimization complexity of sector-coupled systems. The multi-vector optimization embedded in the two-level optimization scheme is formulated as a mixed-integer linear problem, optimizing the energy flow between domains, which are modeled as an abstraction of a sector. Distributed devices are modeled as components that represent an abstraction of devices connected to an energy domain. The optimization process is evaluated based on the data from a residential complex in Ghent, Belgium. It shows that the approach is capable of minimizing costs, CO2 emissions, and dependency on external resources. KW - Energy optimization KW - Multi-energy systems KW - Sector coupling KW - Energy Management KW - Systems Y1 - 2023 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2023110413071966670578 VL - 140 IS - 5 SP - 460 EP - 470 PB - Springer Nature CY - Berlin ER - TY - JOUR A1 - Fink, Thomas A1 - Forster, Brigitte A1 - Heinrich, Florian T1 - Gabor’s “complex signal” revisited: Complexifying frames and bases JF - PAMM (Proceedings in Applied Mathematics and Mechanics) N2 - In 1946, Dennis Gabor introduced the analytic signal 𝑓 + 𝑖𝐻𝑓 for real-valued signals 𝑓. Here, 𝐻 is the Hilbert transform. This complexification of functions allows for an analysis of their amplitude and phase information and has ever since given well-interpretable insight into the properties of the signals over time. The idea of complexification has been reconsidered with regard to many aspects: examples are the dual tree complex wavelet transform, or via the Riesz transform and the monogenic signal, that is, a multi-dimensional version of the Hilbert transform, which in combination with multi-resolution approaches leads to Riesz wavelets, and others. In this context, we ask two questions: - Which pairs of real orthonormal bases (ONBs), Riesz bases, frames and Parseval frames {𝑓 𝑛 } 𝑛∈ℕ and {𝑔 𝑛 } 𝑛∈ℕ can be “rebricked” to complex-valued ones {𝑓𝑛 + 𝑖𝑔 𝑛 } 𝑛∈ℕ? - And which real operators A allow for rebricking via the ansatz {𝑓𝑛 + 𝑖𝐴𝑓𝑛 } 𝑛∈ℕ? In this short note, we give answers to these questions with regard to a characterization which linear operators A are suitable for rebricking while maintaining the structure of the original real valued family. Surprisingly, the Hilbert transform is not among them. Y1 - 2023 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:101:1-2023091815015485698899 VL - 23 IS - 3 PB - Wiley CY - Hoboken ER - TY - THES A1 - Stumpf, Peter Frederik T1 - Partial Representation Extension and Simultaneous Representation of Intersection Graphs N2 - Many real world problems can be modeled with geometric intersection graphs. A (geometric) intersection representation of a graph G=(V,E) is a family {R_v}_{v\in V} of geometric objects such that two geometric objects R_u, R_v intersect if and only if the corresponding vertices u, v are adjacent in G. The most prominent class of intersection graphs are interval graphs, which have representations consisting only of intervals on the real line. Interval graphs have applications in genetics, scheduling, archaeology and many more fields. The recognition problem asks the question whether a given graph belongs to a certain graph class. Two natural generalizations of the recognition problem are the partial representation extension problem and the simultaneous representation problem. In the partial representation extension problem one is given a graph G and a partial representation, i.e., a representation of a subgraph of G. The question then is whether the partial representation can be extended to the whole graph G without changing the given partial representation. In the simultaneous representation problem one is given multiple graphs G_1,...,G_k that can have shared parts, and the question is whether there are representations of all input graphs such that shared vertices are represented by the same geometric objects. Often the sunflower case is considered, where the shared part of any two input graphs is the same. We determine the complexity of the partial representation extension problem and the simultaneous representation problem, especially in the sunflower case for various intersection graph classes. We also improve the running time for various intersection graph classes. In particular, we show that the partial representation extension problem for circular-arc graphs is NP-complete and that the simultaneous representation problem for interval graphs can be solved in linear time in the sunflower case, answering open questions from 2014 and 2010. KW - partial representation KW - simultaneous representation KW - sunflower representation Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15201 ER - TY - THES A1 - Ghosh Dastidar, Kanishka T1 - Using Context for Credit Card Fraud Detection N2 - Online 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. N2 - 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. KW - Credit Card Fraud Detection KW - Deep Learning KW - Neural Networks Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15561 ER - TY - THES A1 - Soller, Sebastian T1 - Anomaly Detection and Forecasting Techniques and their Applications Scenarios, Challenges and Limits in Industrial Production Settings N2 - What needs to be done to get machine learning and artificial intelligence from the lab to the shop floor? This work and its affiliated publications focus on challenges and solutions to apply machine learning applications inside industrial setups and what steps are needed to improve those setups. In industrial setups it is easy to run into a "hen and egg" problem. To gather data, the information which data to gather is ideally given beforehand and these information are not available when studying new setups and machines. In this work setups and concepts are created to dynamically connect to a network and start gathering data from available endpoints inside a manufacturing setup. The data streams of these endpoints are further analyzed to give an initial analysis of the data and advise further processing. To further analyze these data streams with the current advances in the machine learning field and AI, plug and play solutions are presented by manufacturers and scientific research. Limits are determined for this plug and play capability and solutions are provided to further improve upon the base solutions. The capability to apply commonly applied methods was analyzed and initially provided non-sufficient results. In the sub-fields of anomaly detection, regression analysis, forecasting and classification the addition of context information, such as production specific information and time dependent analysis were used to improve the results. Context information, especially periodic information, were further conceptualized and integrated into the initial data analysis. Difficulties with correct labeling of ground truth due to differing biases of participants were encountered, and counter measurements were proposed. Results of the classification, regression, forecast and context information extraction were investigated for their influence on the human operator. A significant change could be measured in multiple cases, just by providing information about underlying problems and errors. The Aforementioned machine learning methods further improved the performance of machine and operator. KW - Anomalieerkennung KW - Künstliche Intelligenz KW - Maschinelles Learning KW - Plug-and-Play Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15659 ER - TY - THES A1 - Frühwirth, Lorenz T1 - The Asymptotic Behavior of Birkhoff- and Lacunary Sums N2 - This doctoral thesis consists of three independently published research articles on the asymptoic behaviour of Lacunary- and Birkhoff sums. The former are sums formed by periodic functions and exponentially growing sequences of natural numbers. The corresponding summands often exhibit behavior typical of independent and identically distributed random variables. The methods used are of an analytical and probabilistic nature. The Birkhoff sums considered in this work are generated by the Kronecker sequence and by discontinuous functions. The methods employed are from the field of metric number theory, specifically classical results from continued fraction theory are utilized. Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15677 ER - TY - GEN A1 - Mexis, Nico A1 - Lill, Bjarne A1 - Doleh, Yousef A1 - Katzenbeisser, Stefan T1 - Supplementary Material for the Work "Exposing the Gaps: The State of Supply Chain Coverage in Current Security Standards" N2 - This document contains the supplementary material for our work "Exposing the Gaps: The State of Supply Chain Coverage in Current Security Standards", which should be consulted for further information and details. While this supplement has not been peer-reviewed, the above article has. The reviewers also had access to all the data presented in this document during the review phase. The article is available from 26 July 2025 as part of the "IFIP Advances in Information and Communication Technology", volume 742: 10.1007/978-3-031-94924-1_14 KW - Supply Chain Management KW - Informationssicherheit KW - Supply Chain KW - Risk Management KW - Coverage KW - Cybersecurity KW - Standard Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15633 ER - TY - THES A1 - Raich, Krispin T1 - Multimodal Data Space for Cooperative Intelligent Transport Systems N2 - Modern Cooperative Intelligent Transport Systems (C-ITSs) are comprehensive applications that must cope with a multitude of challenges while meeting strict service and security standards. One of these challenges is a fast, secure, reliable, and universal way to store and exchange data in such a traffic system. Furthermore, multimodal scenarios where different types of vehicles (e.g., cars and Unmanned Aerial System) interact with each other, are increasingly emerging. To overcome these challenges, this thesis presents a set of key innovations to establish a multimodal capable data space for transport application. Therefore, a multimodal optimized geographic model is presented, called SpatialJSON, that is capable of depicting two- and three-dimensional geometries. To accomplish this feat, SpatialJSON extends the popular GeoJSON format with two new data types: area and corridor. Exchanging, managing, and storing data is handled in a novel data-centric middleware, called Large Scale Multimodal Data Processing Middleware for Intelligent Transport Systems (LDPM). This LDPM uses cryptographic- and trust-based schemas to allow secure data exchange and provide data quality assessment. Furthermore, a service architecture is introduced, that fulfils modern service requirements. Trust management is also another essential part of a C-ITS. Hence, a novel scheme to describe traffic related evidence in a multimodal environment is introduced. This schema allows assessing arbitrary traffic related data. This information is then processed in a specialized and modified Bayesian Inference (BI) function. Subsequently, a comprehensive data centric trust management method is introduced. Finally, a use case is presented that relies on the aforementioned technologies to collect data in a hazardous environmental. This data is then distributed and managed via the LDPM, and finally visualized. Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15496 ER - TY - THES A1 - Julka, Sahib T1 - Towards Data Efficiency and Controllable Representations for Deep Learning in Resource-Constrained Domains N2 - The 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. KW - artificial intelligence KW - deep active learning KW - label scarcity KW - data scarcity Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-16030 ER - TY - THES A1 - Wilhelm, Sebastian T1 - Emergency Detection in Private Households Utilizing Existing Data Sources for Human Activity Event Recognition N2 - In an aging society, the need for efficient emergency detection systems in smart homes is becoming increasingly important. Over 30% of those aged 65 and older experience at least one fall per year, often resulting in the inability to rise without assistance, leading to ‘long lies’ lasting hours or even days. Systems for detecting such emergency events usually rely on wearable sensors or specific installations of ambient sensors, which can be intrusive and complex, hindering acceptance. This thesis proposes a novel approach that utilizes existing digital data sources within the residential infrastructure to detect human activities and identify potential emergencies. A survey identifies 44 potential data sources in private households for recognizing human activity. However, extracting activity information often requires complex preprocessing. In this thesis, methodologies are developed for three of these data sources to highlight practical applications: Smart Power Meters, Smart Water Meters, and Home Weather Stations. It is shown that detecting human activities using these sources is feasible in a practical environment, although accuracy and reliability vary. Notably, Smart Water Meters demonstrate high reliability, with a precision of 0.86 and a recall of 1.00, making them particularly suitable for emergency detection. Existing emergency detection methods are not designed to handle uncertain activity data. This thesis introduces a novel approach based on probabilistic activity information, employing an Inactivity Score that provides a probabilistic weighting of inactivity periods based on the reliability of sensor measurements. By analyzing historical Inactivity Scores, anomalies that potentially represent an emergency can be identified. Evaluations across seven datasets show this approach outperforms existing methods, achieving a mean time to detect emergencies of approximately 05:23:28 hours and producing 0.09 false positives per day under noise-free conditions. Moreover, unlike related approaches, the proposed method remains effective with noisy data. This thesis demonstrates that emergencies in private households can be detected using existing data sources from the home infrastructure, offering a cost-effective and non-intrusive solution to enhance the safety and autonomy of the elderly at home. Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:739-opus4-15992 ER -