- informatik-vortraege - lists.rwth-aachen.de

Survey Lecture - Highlights of UnRAVeL
by Kassing, Jan-Christoph 09 Jul '25

09 Jul '25

Dear all, We invite you to the next UnRAVeL survey lecture talk Tomorrow 12:30 to 14:00, Computer Science Center, Building E2, ground floor, B-IT room 5053.2, given by Christopher Morris about „Exploring the Power of Graph Neural Networks in Solving Optimization Problems“! -------------------- Titel: Exploring the Power of Graph Neural Networks in Solving Optimization Problems Abstract: In recent years, there has been a growing trend toward using machine learning to enhance optimization algorithms by leveraging the structure and distribution of instance data. This lecture will explore recent advances in integrating graph neural networks with exact combinatorial solvers to tackle hard optimization problems. -------------------- Part of the programme of the research training group UnRAVeL is a series of lectures on the topics of UnRAVeL’s research thrusts algorithms and complexity, verification, logic and languages, and their application scenarios. Each lecture is given by one of the researchers involved in UnRAVeL. This years topic is "Highlights of UnRAVeL!". As UnRAVeL nears its conclusion, we take this opportunity to reflect on seven years of excellent research. Through nine talks, we invite you to experience the scientific highlights of our RTG, along with its numerous successes, developments, and collaborations. All interested doctoral researchers and master students are invited to attend the UnRAVeL lecture series 2025 and engage in discussions with researchers and doctoral students. The overall remaining schedule is as following: 10.07.2025 - Christopher Morris - Exploring the Power of Graph Neural Networks in Solving Optimization Problems 17.07.2025 - Christina Büsing - Dealing with Uncertainty Inspired by Health Care – a Robust Optimization Approach Kind regards, Jan-Christoph for the organisation committee

1 0

Einladung Informatik Oberseminar Lubna Ali
by Bombelka, Angelika 08 Jul '25

08 Jul '25

+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Dienstag, 30. Januar 2024, 14.00 Uhr Ort: 9222, E3, Ahornstr. 55 und hybrid via Zoom (https://rwth.zoom-x.de/j/64937773189?pwd=eGttNUMzSElnQUVkc3FrYzBqK2F4UT09) Referent: Lubna Ali M.Sc. RWTH Lehr- und Forschungsgebiet Informatik 9 (Lerntechnologien) Thema: convOERter: A Technical Assistance Tool to Support Semi-Automatic Conversion of Images in Educational Materials as OER Abstract: Open Educational Resources (OER) are seen as an important element in the process of digitizing higher education teaching and as essential building blocks for openness in education. They can be defined as teaching, learning, and research materials that have been made openly available, shareable, and modifiable. OER include different types of resources such as full courses, textbooks, videos, presentations, tests, and images, which are usually published under the open Creative Commons licences. OER can play an important role in improving education by facilitating access to high quality digital educational materials. Accordingly, there is a steady increase among higher education institutions to participate in the so-called "open movement" in general and in utilizing OER in particular. Nevertheless, there are many challenges that still face the deployment of OER in the educational context. One of the main challenges is the production of new OER materials and converting already existing materials into OER, which could be viable by qualifying educators through training courses and/or supporting them with specific tools. There are many platforms and tools that support the creation of new OER content. However, to our knowledge, there are no tools that perform fully- or semi-automatic conversion of already existing educational materials. This identified gap was the basis for the design and implementation of the OER conversion tool (convOERter). The tool supports the user by semi-automatically converting educational materials containing images into OER-compliant materials. The main functionality of the tool is based on reading a file, extracting all images as well as all possible metadata, and substituting the extracted images with OER elements in a semi-automated way. The retrieved OER images are referenced and licenced properly according to the known TASLL rule. Finally, the entire file is automatically licenced under Creative Commons excluding specific elements from the entire licence such as logos. In order to evaluate the effectiveness of the tool in promoting the use of OER, a comprehensive user study was conducted with educators and OER enthusiastic at different universities. The study was accomplished by offering a series of OER evaluation workshops to compare the conversion efficiency of the tool with manual conversion. The results show that using the conversion tool improves the conversion process in terms of speed, license quality, and total efficiency. These results highlight that the tool can be a valuable addition to the community, especially for users less experienced with OER. As a future work, it is intended to further develop the tool and improve its functionality. Additionally, a long-term study can be conducted to assess the impact of the tool in facilitating and enhancing the production of OER on a larger scale. Es laden ein: die Dozentinnen und Dozenten der Informatik

3 5

Einladung: Informatik-Oberseminar Dan Jia
by Jia, Dan 08 Jul '25

08 Jul '25

+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Freitag, 11. Juli 2025, 15.30 Uhr Ort: Raum 025, Mies-van-der-Rohe Str. 15 (UMIC Gebäude) Referent: Dan Jia, M.Sc. Lehrstuhl Informatik 13 Thema: Efficient Person Detection with LiDAR Sensors for Mobile Robots Abstract: Person detection is a fundamental capability for mobile robots operating in unstructured and dynamic environments, including public spaces, healthcare facilities, and households. While RGB(-D) cameras combined with deep learning-based object detectors have become standard solutions, they face notable limitations such as restricted field of view, reduced depth accuracy at long range, and increased computational complexity in multi-camera setups. These limitations are particularly critical in real-time robotic applications. In contrast, 2D LiDAR sensors—widely deployed for mapping and navigation—offer accurate range measurements, wide field of view, and high frame rates, while also being less intrusive in terms of privacy. This talk introduces DR-SPAAM, a novel and efficient deep neural network designed specifically for detecting persons using sparse 2D LiDAR data. At the core of DR-SPAAM is a Spatial Attention and Auto-Regressive Module that aggregates temporal information across consecutive scans. This architecture addresses the inherent sparsity of 2D LiDAR data while maintaining minimal computational overhead. DR-SPAAM achieves state-of-the-art performance on the DROW benchmark dataset, reaching 70.3% average precision, while operating in real-time on both high-end and embedded GPUs. In addition, DR-SPAAM demonstrates robustness to changes in sensor specifications such as resolution and scan rate, making it adaptable to a wide range of robotic platforms without the need for re-training. To address the scarcity of annotated 2D LiDAR training data, an approach is introduced that automatically generates pseudo-labels using image-based detectors from calibrated cameras. These pseudo-labels enable training or fine-tuning of LiDAR-based detectors during deployment without the need for manual annotation, and performance can be further enhanced through techniques such as mixup regularization and robust loss functions. Finally, the talk provides a comparative analysis of 2D and 3D LiDAR sensors for person detection in terms of accuracy, inference speed, localization performance, and occlusion handling. The results reveal that 2D LiDAR can match 3D LiDAR in detecting nearby, visible persons at a fraction of the cost and computational demand. However, for environments with frequent occlusions or the need for long-range detection, 3D LiDAR remains the preferred choice. Taken together, these findings offer practical guidance for selecting sensor modalities and designing robust, efficient person detection systems for a wide spectrum of robotic applications. Es laden ein: die Dozentinnen und Dozenten der Informatik

1 0

Einladung: Informatik-Oberseminar Andreas Klinger
by Andreas Klinger 04 Jul '25

04 Jul '25

+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Freitag, 18. Juli 2025, 10:00 Uhr Ort: UMIC_025 (2165|025), Mies-van-der-Rohe-Str. 15, EG Referent: Andreas Klinger M.Sc. Lehr- und Forschungsgebiet IT-Sicherheit Thema: Variations of Secure Multi-Party Computation Abstract: Secure multi-party computation (SMPC) enables multiple parties to evaluate functions over their private inputs such that none of the parties can learn anything about the other parties’ private inputs and outputs but what can be deduced from their own input and output of the function evaluation. In classical SMPC it is assumed that the set of parties evaluating the function is fixed and a priori known. However, there are also online functionalities, in which parties arrive and leave over time, e. g., when providing possible dating recommendations, or finding other people for ride-sharing on the fly, where the personal preferences or start and destination should be kept private. Furthermore, in some applications participants may not only want to hide their private inputs and outputs, but may also want to hide the fact that they are participating in a given function evaluation in the first place, e. g., when checking the (private) DNA for a disease, then the participation in the check itself can already leak sensitive information. On the practical side of SMPC, the performance and applicability of SMPC protocols is typically evaluated with extensive benchmarking in varying settings, e. g., different number of parties, problem sizes, and network settings. However, evaluating the performance of SMPC protocols can be very time and resource consuming, as SMPC protocols typically introduce a considerable performance overhead compared to non-privacy-preserving computations. In this work we will extend the classical notions of SMPC to allow the secure evaluation of online functionalities in a distributed fashion. We present the notion of an online trusted third party that allows to prove the security of SMPC protocols implementing online functionalities. We also show in a general feasibility result that online functionalities can be implemented in a privacy-preserving fashion. In addition, we present two privacy-preserving online SMPC protocols for the online problem “fully online matching with deadlines”. In this problem an (a priori unknown) set of parties arrives with their inputs over time and can then be matched with other parties until they leave when their individual deadline is reached. We explore the performance of our protocols in an extensive evaluation. Furthermore, we present an anonymous system for SMPC that allows parties to anonymously evaluate an (offline) functionality in a fully distributed and secure fashion. The overall system provides robustness with penalty against a dishonest majority in the presence of a malicious adversary. Finally, we present an approach to estimating the runtime and global network traffic of SMPC protocols. We implemented this approach in the tool XENON, and show its estimation capabilities in an extensive evaluation. In addition, we present a second tool called NEON that simplifies the benchmarking of SMPC protocols, which might be of independent interest. Es laden ein: die Dozentinnen und Dozenten der Informatik

1 0

Survey Lecture - Highlights of UnRAVeL
by Kassing, Jan-Christoph 03 Jul '25

03 Jul '25

Dear all, We invite you to the next UnRAVeL survey lecture talk today (03.07.) 12:30 to 14:00, Computer Science Center, Building E2, ground floor, B-IT room 5053.2, given by Sebastian Trimpe about "Uncertainty in Learning and Control“! -------------------- Titel: Uncertainty in Learning and Control Abstract: In the research at DSME, we combine feedback control and machine learning to develop algorithms for physical systems such as robots, autonomous vehicles, or production. A central theme across our research is uncertainty. In feedback control, uncertainty is the prime reason for using feedback in the first place. When we add learning to the loop, we make decisions based on components learned from data, which are inherently imperfect. This makes uncertainty quantification and management even more critical. In this talk, I will discuss several of our approaches in learning-based control, emphasizing the role of uncertainty in each. In particular, I will focus on methods in deep reinforcement learning, imitation learning from optimal control, and auto-tuning via Bayesian optimization. I will highlight recent results and describe different mechanisms we use to address and leverage uncertainty. Throughout, I will connect these methods to real-world applications, including hardware experiments with unicycle robots, quadcopters, and cars. -------------------- Part of the programme of the research training group UnRAVeL is a series of lectures on the topics of UnRAVeL’s research thrusts algorithms and complexity, verification, logic and languages, and their application scenarios. Each lecture is given by one of the researchers involved in UnRAVeL. This years topic is "Highlights of UnRAVeL!". As UnRAVeL nears its conclusion, we take this opportunity to reflect on seven years of excellent research. Through nine talks, we invite you to experience the scientific highlights of our RTG, along with its numerous successes, developments, and collaborations. All interested doctoral researchers and master students are invited to attend the UnRAVeL lecture series 2025 and engage in discussions with researchers and doctoral students. The overall remaining schedule is as following: 03.07.2025 - Sebastian Trimpe - Uncertainty in Learning and Control 10.07.2025 - Christopher Morris - Exploring the Power of Graph Neural Networks in Solving Optimization Problems 17.07.2025 - Christina Büsing - Dealing with Uncertainty Inspired by Health Care – a Robust Optimization Approach Kind regards, Jan-Christoph for the organisation committee

1 0

Survey Lecture - Highlights of UnRAVeL
by Kassing, Jan-Christoph 25 Jun '25

25 Jun '25

Dear all, We invite you to the next UnRAVeL survey lecture talk Tomorrow (26.06.) 12:30 to 14:00, Computer Science Center, Building E2, ground floor, B-IT room 5053.2, given by Michael Schaub about "Oversmoothing in graph neural networks“! -------------------- Titel: Oversmoothing in graph neural networks Abstract: Graph neural networks (GNNs) have emerged as powerful tools for processing relational data in applications. However, GNNs suffer from the problem of oversmoothing, the property that the features of all nodes exponentially converge to the same vector over layers, prohibiting the design of deep GNNs. In this work we study oversmoothing in graph convolutional networks (GCNs) from different perspectives and demonstrate that, contrary to what is often claimed in the literature, GNNs do not have to oversmooth. -------------------- Part of the programme of the research training group UnRAVeL is a series of lectures on the topics of UnRAVeL’s research thrusts algorithms and complexity, verification, logic and languages, and their application scenarios. Each lecture is given by one of the researchers involved in UnRAVeL. This years topic is "Highlights of UnRAVeL!". As UnRAVeL nears its conclusion, we take this opportunity to reflect on seven years of excellent research. Through nine talks, we invite you to experience the scientific highlights of our RTG, along with its numerous successes, developments, and collaborations. All interested doctoral researchers and master students are invited to attend the UnRAVeL lecture series 2025 and engage in discussions with researchers and doctoral students. The overall remaining schedule is as following: 26.06.2025 - Michael Schaub - Oversmoothing in Graph Neural Networks 03.07.2025 - Sebastian Trimpe - Uncertainty in Learning and Control 10.07.2025 - Christopher Morris - Exploring the Power of Graph Neural Networks in Solving Optimization Problems 17.07.2025 - Christina Büsing - Dealing with Uncertainty Inspired by Health Care – a Robust Optimization Approach Kind regards, Jan-Christoph for the organisation committee

1 0

Einladung: Informatik-Oberseminar Markus Dahlmanns
by Dahlmanns, Markus 24 Jun '25

24 Jun '25

+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Mittwoch, 2. Juli 2025, 10.00 Uhr Ort: Raum 9007, Gebäude E3, Ahornstr. 55 (COMSYS Seminarraum [1]) Referent: Markus Dahlmanns, M.Sc. Lehrstuhl für Informatik 4 (COMSYS) Thema: Identifying Security Issues in the Industrial Internet of Things Abstract: The Industrial Internet of Things (IIoT) allows the collection and communication of data from production processes as well as sending and receiving commands for interaction and actuation to, e.g., increase the efficiency of processes. However, a secure and safe operation is essential to protect, e.g., workers laboring in the operating space of machines and the environment, which were shown to suffer from the misoperation of critical IIoT deployments. As a foundation for such a secure and safe operation IIoT protocols nowadays include security features. However, it is unclear whether these developments of protocol specifications lead to a secure IIoT in practice. In this talk, we address the open research gap of the current uncertainty on IIoT security. To this end, we first set out to assess the security of today's IIoT. Second, we analyze the pitfalls that hinder operators from operating securely despite the existence of secure protocols. Last, we give an overview on novel approaches that help operators secure their deployments. Our first three contributions show from various angles that the majority of Internet-exposed IIoT deployments are insecurely configured, independently of their potential deployment date and the protocol used being either secure-by-design or retrofitted. In our fourth contribution, we trace this problem back to modern technologies like containerization that ease deployment processes but also disguise security issues. Our last two contributions propose two novel mechanisms that can increase the security of the IoT in the future. Our contributions underpin the vast number of issues in IIoT security despite the existence of strong security features in today's protocol specifications and sketch countermeasures to tackle the identified pitfalls that operators tap into when configuring their deployments. Overall, our results encourage shifting from secure-by-design to secure-by-default protocols. Es laden ein: die Dozentinnen und Dozenten der Informatik [1] https://www.comsys.rwth-aachen.de/contact/how-to-get-to-comsys.pdf -- Markus Dahlmanns, M.Sc., Ph.D. Student Chair of Communication and Distributed Systems RWTH Aachen University, Germany tel: +49 241 80 21425 web: https://www.comsys.rwth-aachen.de/team/markus-dahlmanns/

1 0

Research Talk Prof. Romaric Duvignau, 25. Juni 2025, 10:00 Uhr
by Jan Pennekamp 18 Jun '25

18 Jun '25

Dear all, COMSYS is hosting Prof. Romaric Duvignau next week. As part of this visit, he will also give a research talk. More details below. **Title:** Data-Driven and Efficient Algorithms for Enhanced Communication at the Edge **When:** Wednesday, June 25th 10:00-11:00 **Where:** At COMSYS' Seminar Room – Building E3, Room 9007 **Abstract:** Communication is the cornerstone element of many distributed, cyber-physical, or IoT systems (e.g., vehicular networks, smart grid, sensor networks). Transmitting data in the most efficient manner is often a requirement for many applications to reduce costs and boost performance. What is needed in this context are dedicated (i.e., data- or use-case-driven) and efficient algorithms that must be designed, analyzed, and experimentally validated, taking into account both the communication network and the distribution and continuous aspect of the generated data. My research centers on the design, study, and evaluation of those distributed and application-driven algorithms to perform optimization on a continuous basis and decrease computational burden and/or communication costs. In this short presentation, I will focus on recent advancements we have for two specific applications: (1) sharing energy resources among households at the smart grid’s edge and (2) improving the privacy of video streaming. **Speaker Bio:** Romaric Duvignau is an Associate Professor in the Distributed Computing and Systems research group at Chalmers University of Technology, Gothenburg, Sweden. His research focuses on designing data-driven algorithms for distributed applications, addressing challenges in fault tolerance, cybersecurity, privacy, big data and streaming analytics. His current interests include the efficient monitoring of large-scale distributed systems, intelligent data management and decision making in smart grids, federated learning, programmable networks, and network traffic classification. Prof. Duvignau teaches courses on computer networking, distributed systems, and data-driven support for cyber-physical systems and was awarded the Chalmers Pedagogic Prize 2023. You are all cordially invited. Best regards, Jan Pennekamp -- Dr. rer. nat. Jan Pennekamp Communication and Distributed Systems RWTH Aachen University, Germany tel: +49 241 80 21411 web: https://www.comsys.rwth-aachen.de/team/jan-pennekamp/

1 0

Einladung Informatik-Oberseminar Tobias Schulte to Brinke
by Abigail Morrison 12 Jun '25

12 Jun '25

+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Montag, 16. Juni 2025, 15.00 Uhr Ort: Raum 9222, Gebäude E3, Informatik-Zentrum Referent: Tobias Schulte to Brinke, MSc. Lehrstuhl Informatik 3 Thema: Analysis of Information Processing and Memory Prerequisites for Temporal Difference Learning in Cortical Neural Network Models Abstract: This doctoral thesis delves into the computational intricacies of the human brain, exploring the capabilities of cortical microcircuit models, the extent of their information processing capacity, and their role in memory and temporal difference learning through the use of cortico-striatal populations. Central to this exploration is the study of spiking neural networks (SNN). This research aims to gain a deeper understanding of the structural and neuronal influences on information processing in these networks and at the same time to provide a guideline for their analysis. In the first part, a network model of a cortical column introduced in a previous paper is reproduced and extended. These analyses show that the specific, data-based connectivity improves computational performance by sharpening the clarity of internal representations rather than increasing the duration of information retention as previously described. Moving beyond traditional task-based evaluations, the second part introduces a novel application of the information processing capacity (IPC) metric to SNN. This approach provides a comprehensive profile of the functions computed by SNN, encompassing memory and nonlinear processing. The study methodically examines various encoding mechanisms and their impact on the IPC and shows that the metric is indicative of the performance in tasks with different demands of nonlinear processing and memory. This exploration not only extends the utility of the IPC metric to more complex neural networks but also offers a deeper insight into their computational capabilities. The third part of the thesis tests a hypothesis about the computation of temporal difference errors in the brain, focusing on two distinct populations of cortical layer 5 neurons: the crossed corticostriatal (CCS) and corticopontine (CPn0 cells. By implementing network models based on these populations and evaluating their memory capabilities through the lens of the IPC, the research supports, at least for continuous rate networks, the proposed role of these neurons in the computation of temporal difference errors. However, the spiking network models pose a greater challenge and exhibit little ability to memorize previous inputs in our experiments. In summary, this work not only confirms and extends existing research results, but also develops new methods for analyzing SNN. It lays a solid foundation for future studies of the brain's computational processes and enriches the field of computational neuroscience with advanced tools and methods for exploring the intricate workings of biologically inspired neural network models. Es laden ein: die Dozentinnen und Dozenten der Informatik -- Prof. Dr. Abigail Morrison Institute for Advanced Simulation (IAS-6) Jülich Research Center & Computer Science 3 - Software Engineering RWTH Aachen http://www.fz-juelich.de/inm/inm-6 http://www.se-rwth.de Office: +49 2428 8097504 Fax # : +49 2461 61-9460 Pronouns: she/her ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------ Forschungszentrum Jülich GmbH 52425 Jülich Sitz der Gesellschaft: Jülich Eingetragen im Handelsregister des Amtsgerichts Düren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Stefan Müller Geschäftsführung: Prof. Dr. Astrid Lambrecht (Vorsitzende), Dr. Stephanie Bauer (stellv. Vorsitzende), Prof. Dr. Ir. Pieter Jansens, Prof. Dr. Laurens Kuipers ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------

1 0

[sekretariate] Einladung: Informatik-Oberseminar Lutz Klinkenberg
by Elke Ohlenforst 05 Jun '25

05 Jun '25

+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Freitag, 20. Juni 2025, 09:00 Uhr Ort: Raum 9222 (Informatikzentrum E3, Ahornstraße 55) Referent: Lutz Klinkenberg, M.Sc. Lehrstuhl für Informatik 2 Thema: Analysis of Probabilistic Programs Using Generating Functions Abstract: Probabilistic programming languages have emerged as a powerful tool for modeling complex and uncertain systems. Widely applied in fields like artificial intelligence, machine learning, robotics but also in non-technical domains as, e.g., healthcare, social and climate sciences, economic and psychology, these languages allow us to succinctly represent probabilistic processes, including those involving loops, recursion, and other intricate structures. However, reasoning about such programs, particularly those with infinite loops or non- terminating behavior, poses significant challenges. This dissertation introduces a novel approach to address these challenges by developing a denotational probability generating function (PGF) semantics for discrete probabilistic programs with loops. This semantics provides an exact and compact representation of program behaviors, enabling exact Bayesian inference where prior methods often relied on approximations. A central contribution of this work is the introduction of invariant-based reasoning techniques, which facilitate the systematic derivation of posterior distributions and hence also properties like expectations and higher-order moments, even for possibly unbounded loops. Furthermore, we present a syntactic characterization of a class of proba- bilistic programs for which the distributions they describe can be expressed as rational closed-form generating functions. This characterization ensures that exact inference is preserved within this class, allowing for efficient alge- braic manipulation and direct computation of quantities such as tail bounds and moments. By linking program syntax to the algebraic tractability of their semantics, this work establishes a foundational framework for exact reasoning in probabilistic programming. This framework not only enhances the capacity for exact Bayesian inference enabling decision-making under uncertainty but also broadens the applicability of probabilistic programming to domains where formal verification is inevitable. It combines theoretical rigor with practical utility, offering a unified perspective on the structure and behavior of discrete probabilistic programs. Es laden ein: die Dozentinnen und Dozenten der Informatik

1 0