@article{65098, author = {{Weber, Daniel and Lange, Jarren and Wallscheid, Oliver}}, issn = {{2687-9735}}, journal = {{IEEE Journal of Emerging and Selected Topics in Industrial Electronics}}, pages = {{1--12}}, publisher = {{Institute of Electrical and Electronics Engineers (IEEE)}}, title = {{{Reinforcement Learning-Based Control of Voltage-Forming Grid Inverters With Arbitrary Loads}}}, doi = {{10.1109/jestie.2026.3654784}}, year = {{2026}}, } @inproceedings{65178, abstract = {{Large intermediate results can cause join queries to run unexpectedly long. This problem is particularly common for analytical queries, which aggregate data over many tables to produce a comparatively small final output, and queries on graph data, where intermediate results blow up quickly. Recent work inspired by Yannakakis’ algorithm approaches this by modifying the query engine to avoid materializing unnecessary tuples. However, this requires significant changes to the core of the system, which is not feasible in many situations such as cloud environments or proprietary systems. In this work, we propose a flexible approach for optimizing long-running join queries from the outside of the DBMS. Rewriting-based realizations of Yannakakis’ algorithm suffer from inherent overhead due to the creation of intermediate tables. Thus, we present an approach for detecting and targeting queries which would benefit from a Yannakakis-style optimization. We introduce a new benchmark combining 5 standard benchmarks and augmenting them with additional instances, which provides a sufficient size and diversity for a machine learning based solution. On PostgreSQL, DuckDB and SparkSQL, slowdowns on queries where the rewriting is counterproductive are mostly avoided, as opposed to a naïve application of the rewriting, and we observe significant improvements in end-to-end runtimes over standard query execution and unconditional rewriting.}}, author = {{Böhm, Daniela and Gottlob, Georg and Lanzinger, Matthias and Longo, Davide Mario and Okulmus, Cem and Pichler, Reinhard and Selzer, Alexander}}, booktitle = {{Proceedings of the 28th International Workshop on Design, Optimization, Languages and Analytical Processing of Big Data (DOLAP 2026)}}, keywords = {{Join Queries, Acyclic Queries, Query Processing}}, title = {{{Selective Use of Yannakakis’ Algorithm for Consistent Performance Gains}}}, year = {{2026}}, } @article{57580, abstract = {{We investigate dispersive and Strichartz estimates for the Schrödinger equation involving the fractional Laplacian in real hyperbolic spaces and their discrete analogues, homogeneous trees. Due to the Knapp phenomenon, the Strichartz estimates on Euclidean spaces for the fractional Laplacian exhibit loss of derivatives. A similar phenomenon appears on real hyperbolic spaces. However, such a loss disappears on homogeneous trees, due to the triviality of the estimates for small times.}}, author = {{Palmirotta, Guendalina and Sire, Yannick and Anker, Jean-Philippe}}, journal = {{Journal of Differential Equations}}, keywords = {{Schrödinger equation, Fractional Laplacian, Dispersive estimates, Strichartz estimates, Real hyperbolic spaces, Homogeneous trees}}, publisher = {{Elsevier}}, title = {{{The Schrödinger equation with fractional Laplacian on hyperbolic spaces and homogeneous trees}}}, doi = {{10.1016/j.jde.2025.114065}}, year = {{2026}}, } @unpublished{65232, abstract = {{On finite regular graphs, we construct Patterson-Sullivan distributions associated with eigenfunctions of the discrete Laplace operator via their boundary values on the phase space. These distributions are closely related to Wigner distributions defined via a pseudo-differential calculus on graphs, which appear naturally in the study of quantum chaos. Using a pairing formula, we prove that Patterson-Sullivan distributions are also related to invariant Ruelle distributions arising from the transfer operator of the geodesic flow on the shift space. Both relationships provide discrete analogues of results for compact hyperbolic surfaces obtained by Anantharaman-Zelditch and by Guillarmou-Hilgert-Weich.}}, author = {{Arends, Christian and Palmirotta, Guendalina}}, booktitle = {{arXiv:2603.09779}}, pages = {{38}}, title = {{{Patterson-Sullivan distributions of finite regular graphs}}}, year = {{2026}}, } @unpublished{63530, abstract = {{The widespread deployment of 5G networks, together with the coexistence of 4G/LTE networks, provides mobile devices a diverse set of candidate cells to connect to. However, associating mobile devices to cells to maximize overall network performance, a.k.a. cell (re)selection, remains a key challenge for mobile operators. Today, cell (re)selection parameters are typically configured manually based on operator experience and rarely adapted to dynamic network conditions. In this work, we ask: Can an agent automatically learn and adapt cell (re)selection parameters to consistently improve network performance? We present a reinforcement learning (RL)-based framework called CellPilot that adaptively tunes cell (re)selection parameters by learning spatiotemporal patterns of mobile network dynamics. Our study with real-world data demonstrates that even a lightweight RL agent can outperform conventional heuristic reconfigurations by up to 167%, while generalizing effectively across different network scenarios. These results indicate that data-driven approaches can significantly improve cell (re)selection configurations and enhance mobile network performance.}}, author = {{Illian, Marvin and Khalili, Ramin and Rocha, Antonio A. de A. and Wang, Lin}}, booktitle = {{arXiv:2601.04083}}, title = {{{Cells on Autopilot: Adaptive Cell (Re)Selection via Reinforcement Learning}}}, year = {{2026}}, } @inproceedings{65249, author = {{Shaaban KabakiboKabakibo, Huzaifa and Trivedi, Animesh and Wang, Lin}}, booktitle = {{The 9th Annual Conference on Machine Learning and Systems (MLSys)}}, location = {{Bellevue, WA}}, title = {{{Breaking the Ice: Analyzing Cold Start Latency in vLLM}}}, year = {{2026}}, } @inproceedings{65250, author = {{Zohdi, Sepideh and Wang, Lin}}, booktitle = {{The 6th Workshop on Machine Learning and Systems (EuroMLSys)}}, location = {{Edinburg}}, title = {{{Before the First Token: Benchmarking Data Preprocessing in Vision-Language Models }}}, year = {{2026}}, } @article{65253, author = {{Abdelwanis, Ali Hassan Ali and Haucke-Korber, Barnabas and Jakobeit, Darius and Kirchgässner, Wilhelm and Meyer, Marvin and Schenke, Maximilian and Vater, Hendrik and Wallscheid, Oliver and Weber, Daniel}}, issn = {{2577-3569}}, journal = {{Journal of Open Source Education}}, number = {{97}}, publisher = {{The Open Journal}}, title = {{{Reinforcement Learning: A Comprehensive Open-Source Course}}}, doi = {{10.21105/jose.00306}}, volume = {{9}}, year = {{2026}}, } @article{65440, abstract = {{Abstract We present a novel algorithm for quantization and subsequent hexahedral mesh generation from seamless volumetric maps. Quantization is the process of choosing integers that represent the numbers of hexahedral elements to be placed in each region of the volume, and transforming the seamless map into an integer‐grid map matching that choice, inducing a hexahedral mesh. Previous work computes such quantizations under the restriction of a fixed predetermined singularity graph. Our novel approach allows for implicit modification and, in particular, simplification of the map's singularity structure wherever that benefits the chosen objective, such as matching target hexahedron sizes as closely as possible. It comes with two novel ingredients: A feature‐focused distortion measure guiding the quantization, and constraints ensuring map injectivity and structure preservation of geometric and topological features, both without relying on a fixed singularity structure. We demonstrate the benefit of the added flexibility offered by this approach: it allows for the generation of hexahedral meshes that more accurately match a desired resolution globally, as well as of meshes exhibiting a simpler block structure.}}, author = {{Brückler, Hendrik and Campen, Marcel}}, issn = {{0167-7055}}, journal = {{Computer Graphics Forum}}, publisher = {{Wiley}}, title = {{{Volume Quantization with Flexible Singularities for Hexahedral Meshing}}}, doi = {{10.1111/cgf.70349}}, year = {{2026}}, } @techreport{65426, abstract = {{In diesem Forschungsprojekt wurde ein Messverfahren zur Bestimmung akustischer Materialparameter von Polymeren im Ultraschallfrequenzbereich entwickelt. Das Verfahrens sollte, die üblichen standardisierten Prüfmethoden erweitern, die bislang primär im quasistatischen oder niederfrequenten Bereich eingesetzt wurden. Im Gegensatz zu bestehenden Verfahren wie dem Zeitstandversuch oder der Dynamisch Mechanischen Analyse (DMA) nach [DIN6721] sollte die neue Methode eine nicht-invasive Charakterisierung der (visko-)elastischen Materialparameter im Frequenzbereich von 0,75 MHz bis 2,5 MHz ermöglichen. Das entwickelte Ultraschallmesssystem arbeitet nach dem Puls Echo-Prinzip und kann eine räumlich segmentierte, ringförmige Anregung erzeugen. Die Bestimmung der frequenzabhängigen Materialparameter geschieht hierbei über ein inverses Verfahren. Die Ergebnisse des Projekts zeigen, dass die Segmentierung der Anregung, die Geometrie der Probe sowie das Puls-Echo-Messprinzip die Messergebnisse sowie die Sensitivität gegenüber Scherparametern wesentlich beeinflussen. Im Rahmen des Projektes wurde auch eine statistische Auswertung des Optimierungsverfahrens hinsichtlich transversal-isotroper Materialsymmetrie mit Rayleigh-Dämpfung durchgeführt. Die Ergebnisse zeigen, dass das entwickelte Verfahren gute Konvergenzeigenschaften aufweist und sich durch verbesserte Robustheit auszeichnet.}}, author = {{Dreiling, Dmitrij and Itner, Dominik and Birk, Carolin and Gravenkamp, Hauke and Henning, Bernd}}, keywords = {{Materialcharakterisierung, Polymer, Inverses Problem, Ultraschall, Optimierung}}, pages = {{12}}, publisher = {{Hannover : Technische Informationsbibliothek}}, title = {{{Vollständige Bestimmung der akustischen Materialparameter von Polymeren II}}}, doi = {{https://doi.org/10.34657/33602}}, year = {{2026}}, } @inproceedings{61922, abstract = {{We present an extremely simple polynomial-space exponential-time $(1-\varepsilon)$-approximation algorithm for MAX-k-SAT that is (slightly) faster than the previous known polynomial-space $(1-\varepsilon)$-approximation algorithms by Hirsch (Discrete Applied Mathematics, 2003) and Escoffier, Paschos and Tourniaire (Theoretical Computer Science, 2014). Our algorithm repeatedly samples an assignment uniformly at random until finding an assignment that satisfies a large enough fraction of clauses. Surprisingly, we can show the efficiency of this simpler approach by proving that in any instance of MAX-k-SAT (or more generally any instance of MAXCSP), an exponential number of assignments satisfy a fraction of clauses close to the optimal value.}}, author = {{Buhrman, Harry and Gharibian, Sevag and Landau, Zeph and Gall, François Le and Schuch, Norbert and Tamaki, Suguru}}, booktitle = {{SIAM Symposium on Simplicity in Algorithms (SOSA)}}, pages = {{247--253}}, title = {{{A Simpler Exponential-Time Approximation Algorithm for MAX-k-SAT}}}, year = {{2026}}, } @proceedings{64797, editor = {{Birk, Lisa and Loth, Gerrit and Jotzo, Luca and Binder, Karin and Frischemeier, Daniel}}, location = {{Münster}}, publisher = {{International Association for Statistics Education}}, title = {{{14th IASE Satellite Conference "Statistics and Data Science Education in STEAM"}}}, doi = {{10.52041/iase25.158}}, year = {{2026}}, } @inproceedings{63918, abstract = {{Many real-world datasets, such as citation networks, social networks, and molecular structures, are naturally represented as heterogeneous graphs, where nodes belong to different types and have additional features. For example, in a citation network, nodes representing "Paper" or "Author" may include attributes like keywords or affiliations. A critical machine learning task on these graphs is node classification, which is useful for applications such as fake news detection, corporate risk assessment, and molecular property prediction. Although Heterogeneous Graph Neural Networks (HGNNs) perform well in these contexts, their predictions remain opaque. Existing post-hoc explanation methods lack support for actual node features beyond one-hot encoding of node type and often fail to generate realistic, faithful explanations. To address these gaps, we propose DiGNNExplainer, a model-level explanation approach that synthesizes heterogeneous graphs with realistic node features via discrete denoising diffusion. In particular, we generate realistic discrete features (e.g., bag-of-words features) using diffusion models within a discrete space, whereas previous approaches are limited to continuous spaces. We evaluate our approach on multiple datasets and show that DiGNNExplainer produces explanations that are realistic and faithful to the model's decision-making, outperforming state-of-the-art methods.}}, author = {{Das, Pallabee and Heindorf, Stefan}}, booktitle = {{Proceedings of the ACM Web Conference 2026 (WWW ’26)}}, location = {{Dubai, United Arab Emirates}}, publisher = {{ACM}}, title = {{{Discrete Diffusion-Based Model-Level Explanation of Heterogeneous GNNs with Node Features}}}, year = {{2026}}, } @inproceedings{65489, author = {{Okulmus, Cem and Ahmetaj, Shqiponja and Boneva, Iovka and Hidders, Jan and Jakubowski, Maxime and Labra Gayo, José Emilio and Martens, Wim and Mogavero, Fabio and Murlak, Filip and Savković, Ognjen and Šimkus, Mantas and Tomaszuk, Dominik }}, booktitle = {{Proceedings of the 23rd International Conference on Principles of Knowledge Representation and Reasoning (KR 2026)}}, location = {{Lisbon, Portugal}}, title = {{{Common Foundations for Recursive Shape Languages}}}, year = {{2026}}, } @article{63135, abstract = {{We propose a definition of Coxeter-Dynkin algebras of canonical type generalising the definition as a path algebra of a quiver. Moreover, we construct two tilting objects over the squid algebra - one via generalised APR-tilting and one via one-point-extensions and reflection functors - and identify their endomorphism algebras with the Coxeter-Dynkin algebra. This shows that our definition gives another representative in the derived equivalence class of the squid algebra, and hence of the corresponding canonical algebra. Finally, we have a closer look at the Grothendieck group and the Euler form which illustrates the connection to Saito's classification of marked extended affine root systems. On the other hand, this enables us to prove that in the domestic case Coxeter-Dynkin algebras are of finite representation type.}}, author = {{Perniok, Daniel}}, journal = {{Journal of Pure and Applied Algebra}}, number = {{5}}, title = {{{Coxeter-Dynkin algebras of canonical type}}}, doi = {{10.1016/j.jpaa.2026.108250}}, volume = {{230}}, year = {{2026}}, } @article{61152, abstract = {{While neural network quantization effectively reduces the cost of matrix multiplications, aggressive quantization can expose non-matrix-multiply operations as significant performance and resource bottlenecks on embedded systems. Addressing such bottlenecks requires a comprehensive approach to tailoring the precision across operations in the inference computation. To this end, we introduce scaled-integer range analysis (SIRA), a static analysis technique employing interval arithmetic to determine the range, scale, and bias for tensors in quantized neural networks. We show how this information can be exploited to reduce the resource footprint of FPGA dataflow neural network accelerators via tailored bitwidth adaptation for accumulators and downstream operations, aggregation of scales and biases, and conversion of consecutive elementwise operations to thresholding operations. We integrate SIRA-driven optimizations into the open-source FINN framework, then evaluate their effectiveness across a range of quantized neural network workloads and compare implementation alternatives for non-matrix-multiply operations. We demonstrate an average reduction of 17\% for LUTs, 66\% for DSPs, and 22\% for accumulator bitwidths with SIRA optimizations, providing detailed benchmark analysis and analytical models to guide the implementation style for non-matrix layers. Finally, we open-source SIRA to facilitate community exploration of its benefits across various applications and hardware platforms.}}, author = {{Umuroglu, Yaman and Berganski, Christoph and Jentzsch, Felix and Danilowicz, Michal and Kryjak, Tomasz and Bezaitis, Charalampos and Sjalander, Magnus and Colbert, Ian and Preusser, Thomas and Petri-Koenig, Jakoba and Blott, Michaela}}, issn = {{1936-7406}}, journal = {{ACM Transactions on Reconfigurable Technology and Systems}}, title = {{{SIRA: Scaled-Integer Range Analysis for Optimizing FPGA Dataflow Neural Network Accelerators}}}, doi = {{10.1145/3807510}}, year = {{2026}}, } @inproceedings{65501, author = {{Stasytis, Lukas and Jentzsch, Felix and Preusser, Thomas and Umuroglu, Yaman and Petri-Koenig, Jakoba and István, Zsolt}}, booktitle = {{2025 International Conference on Field Programmable Technology (ICFPT)}}, publisher = {{IEEE}}, title = {{{Heuristic & Expert-Guided Buffer Sizing for Neural Network Inference Applications on FPGAs}}}, doi = {{10.1109/icfpt67023.2025.00032}}, year = {{2026}}, } @inproceedings{65500, author = {{Jentzsch, Felix and Platzner, Marco}}, booktitle = {{2025 International Conference on Field Programmable Technology (ICFPT)}}, publisher = {{IEEE}}, title = {{{Empirical QoR Estimation Flow for Fast Design Space Exploration of DNN Dataflow Accelerators}}}, doi = {{10.1109/icfpt67023.2025.00044}}, year = {{2026}}, } @article{65512, abstract = {{Zusammenfassung Risikokompetenz beinhaltet auch die Fähigkeit, stochastische Informationen, wie beispielsweise Anteile und Wahrscheinlichkeiten, richtig zu versprachlichen. Aus der Forschung zu bedingten Wahrscheinlichkeiten und Bayesianischen Aufgaben ist bekannt, dass die Nutzung von Visualisierungen und sogenannten „natürlichen Häufigkeiten“ (z. B. „80 von 100 Personen“) statt Wahrscheinlichkeiten in Prozent Verwechslungen beim Bestimmen von Wahrscheinlichkeiten eindämmen kann. Über den umgekehrten Prozess – das Versprachlichen von in Visualisierungen dargestellten Informationen – ist bisher jedoch wenig bekannt, obwohl diese Versprachlichungen auch für den Aufbau konzeptuellen Wissen als essentiell angesehen werden. In der vorliegenden Studie wurde daher untersucht, wie gut Schüler:innen die Versprachlichung von in Visualisierungen dargestellten Anteilen und natürlichen Häufigkeiten gelingt. Dazu wurde mit 138 Realschüler:innen aus der 9. Jahrgangsstufe ein Papier-und-Bleistift-Test durchgeführt, bei dem den Teilnehmenden nacheinander je ein (vollständig ausgefülltes) Baumdiagramm und ein Netzdiagramm präsentiert wurde. Die Schüler:innen sollten die inhaltliche Bedeutung der in der Visualisierung dargestellten stochastischen Informationen möglichst genau versprachlichen. Fokus der Studie ist der Einfluss der Visualisierung (Baumdiagramm vs. Netzdiagramm) und des Informationsformats der Visualisierung (Anteile in Prozent vs. natürliche Häufigkeiten) auf die richtige Versprachlichung von verschiedenen Relationstypen (Schnittinformationen vs. bedingte Informationen). Die Ergebnisse zeigen unter anderem, dass Informationen in natürlichen Häufigkeiten deutlich besser versprachlicht werden als in Prozenten und dass Schnittinformationen in Prozent besser anhand von Netzdiagrammen als von Baumdiagrammen versprachlicht werden. Die gewonnenen Erkenntnisse geben überdies Einblicke in typische (fehlerhafte) Versprachlichungen von Schüler:innen und könnten als Basis für die Entwicklung eines sprachsensiblen Unterrichts zu Anteilen und Wahrscheinlichkeiten im schulischen Stochastikunterricht dienen.}}, author = {{Rößner, Michael and Binder, Karin and Albrecht, Julian}}, issn = {{0173-5322}}, journal = {{Journal für Mathematik-Didaktik}}, number = {{1}}, publisher = {{Springer Science and Business Media LLC}}, title = {{{Versprachlichung von Anteilen und natürlichen Häufigkeiten anhand von Baum- und Netzdiagrammen Verbalization of proportions and natural frequencies based on tree diagrams and net diagrams}}}, doi = {{10.1007/s13138-026-00267-1}}, volume = {{47}}, year = {{2026}}, } @inbook{65515, abstract = {{Abstract This study explores the usability and onboarding process of a Mixed Reality (MR) application called PEARL, designed to prepare students for laboratory work. Originally developed for mobile Augmented Reality (mAR), PEARL was adapted for MR to offer a more immersive and intuitive experience through hand and gesture controls. Since many students lack experience with MR devices, a user-friendly onboarding system is essential. The study aims to redesign PEARL’s user interface and onboarding experience, evaluating how intuitive interaction elements impact usability. First, a literature review will identify existing usability guidelines for MR applications, which will guide the redesign of the interface. This new version will then be tested with students through a user study. Feedback will be collected via an online survey to assess the onboarding and user experience, and the findings will be used to refine the design further. The expected outcome is an improved onboarding process and interface, making PEARL accessible even for MR novices, enhancing their ability to interact with 3D objects in a real-world setting. Ultimately, the study aims to provide best practices for developing intuitive MR interfaces and effective onboarding experiences, especially in educational contexts.}}, author = {{Alptekin, Mesut and Münstermann, Daniel and Temmen, Katrin}}, booktitle = {{Lecture Notes in Networks and Systems}}, isbn = {{9783032073181}}, issn = {{2367-3370}}, keywords = {{Meta Quest 3, Augmented Reality, Mixed Reality, PEARL, Electrical Engineering, Laboratory Training, Onboarding, User Experience, User Interface, Heuristics}}, location = {{Santiago, Chile}}, pages = {{199--211}}, publisher = {{Springer Nature Switzerland}}, title = {{{Designing and Evaluating the Usability and Onboarding for a Mixed-Reality Application: A Case Study with PEARL (Paderborn Electrical Engineering AR Laboratory) and Meta Quest 3}}}, doi = {{10.1007/978-3-032-07319-8_19}}, volume = {{2}}, year = {{2026}}, }