TY - CHAP A1 - Simmering, Veronika A1 - Pfeiffer, Thies A1 - Kindsmüller, Martin Christof T1 - The Impact of Embodiment and Output Modality on Learning with Conversational Agents in VR T2 - GI Workshop 2025 on Virtual and Augmented Reality : VR/AR 2025, Lecture Notes in Informatics (LNI), Gesellschaft für Informatik, Bonn 2025 N2 - In Virtual Reality (VR) learning applications, the integration of pedagogical agents is particularly promising, as they function as a virtual social support. With the recent developments in Large Language Models (LLMs), conversational language models are now available that enable natural language interaction in a VR environment. We evaluated four possible designs for LLM-based agents (푛 = 21) in an aircraft-engine training. Focusing on the primary questions when bringing LLM-based agents into VR: whether to use text or speech and whether to embody the conversational agent in 3D or not. Pre- and post-tests were used to measure retention and questionnaires were used to measure the User Experience (UX) of different design variants. Contrary to our hypotheses, retention was higher when using a non-embodied design compared to an embodied design. The output modality did not have a significant impact on the learning success, but it did have an impact on the UX. Y1 - 2025 U6 - https://doi.org/10.18420/vrar2025_12 SN - 2944-7682 SP - 235 EP - 254 PB - Gesellschaft für Informatik ER - TY - CHAP A1 - Querner, Erich A1 - Ewerling, Philipp A1 - Kindsmüller, Martin Christof T1 - Setting the Stage for Collaboration: A Multi-View Table for Touch and Tangible Map Interaction T2 - Mensch und Computer 2025 - Workshopband : MCI-WS11: 18th Workshop “Be-greifbare Interaktion”. Chemnitz. 31. August - 03. September 2025 N2 - We present a multi-user map application based on a novel multiview concept that enables simultaneous and independent interaction with shared geospatial content. Each user operates a personal View Finder and Focus View, color-coded for clarity, while a shared, immutable Context View provides a common reference frame. The system supports both touch-based and tangible interaction techniques, including gesture control, virtual joysticks, and physical objects. Users can flexibly arrange their workspaces on a multi-touch table, supporting both individual exploration and collaborative tasks. Y1 - 2025 U6 - https://doi.org/10.18420/muc2025-mci-demo-318 SP - 1 EP - 4 PB - Gesellschaft für Informatik ER - TY - CHAP A1 - Querner, Erich A1 - Ewerling, Philipp A1 - Kindsmüller, Martin Christof T1 - Touch gets it done, tangibles make it fun? Balancing objective accuracy and user experience T2 - Mensch und Computer 2025 - Workshopband : MCI-WS11: 18th Workshop “Be-greifbare Interaktion”. Chemnitz. 31. August - 03. September 2025 N2 - This study compares Tangible User Interfaces (TUIs) with conventional touch interfaces (CTIs) on multi-touch tables for exploratory multi-user map applications. We developed a prototype that enables multiple users to interact with map content simultaneously and independently. Four interaction methods were implemented and evaluated: gesture-based touch control, widget-based touch joystick control, and two tangible interaction variants (a joystick and car-steering metaphor). A user study with 15 participants was conducted in which users navigated predefined routes with varying difficulty. We collected both objective performance measurements and subjective user assessments. While touch-based methods yielded higher accuracy in objective metrics, TUI-based interactions were rated significantly better by participants in terms of user experience. Notably, the car-steering tangible control was particularly well-received, highlighting how physical, playful interaction can enhance usability despite lower accuracy. These findings contribute to our understanding of how different interaction paradigms support collaborative exploration on multi-touch surfaces. Y1 - 2025 U6 - https://doi.org/https://doi.org/10.18420/muc2025-mci-ws11-164 SP - 1 EP - 6 PB - Gesellschaft für Informatik ER - TY - JOUR A1 - Pilgermann, Michael A1 - Vielhauer, Claus A1 - Petrov, Denis A1 - Dittmann, Jana A1 - Krätzer, Christian T1 - Attribution im IT-Grundschutz des BSI: Ausweitung der Detektion von Cyberangriffen im IT-Sicherheitsprozess JF - Datenschutz und Datensicherheit (DuD) N2 - Inzwischen hat sich die Einsicht durchgesetzt, dass Cyberangriffe nicht komplett verhindert werden können. Stattdessen sollen die möglichen Auswirkungen von Vorfällen sinnvoll reduziert werden. Damit steigt die Bedeutung der Erkennung von Sicherheitsvorfällen, um schnell reagieren und das Schadensausmaß begrenzen zu können. Damit die Fortschritte bei der Erkennung künftig auch in der Praxis angewendet werden können, ist eine Verankerung in den entsprechenden Vorgehensweisen erforderlich. Dies erfolgt nachfolgend am Beispiel des IT-Grundschutzes, dessen Methoden in der deutschsprachigen IT-Sicherheitscommunity verbreitet sind. Y1 - 2025 U6 - https://doi.org/10.1007/s11623-025-2173-5 VL - 49 IS - 11 SP - 731 EP - 736 PB - Springer ER - TY - JOUR A1 - Khuntsaria, Irakli A1 - Gachechiladze, Konstantine A1 - Liparta, Elene A1 - Kobulashvili, Tamar A1 - Jorbenadze, Mariam A1 - Revazishvili, Tamaz A1 - Mamiseishvili, Ana A1 - Kekelia, Elene A1 - Topuridze, Marina A1 - Sturua, Lela A1 - Schrader, Thomas A1 - Kldiashvili, Ekaterina T1 - Strategic implementation of HPV vaccination to eliminate cervical cancer in Georgia JF - Public Health in Practice N2 - Objectives Human Papillomavirus (HPV) is a prevalent sexually transmitted infection and a leading cause of cervical cancer. In Georgia, cervical cancer ranks as the fifth most common cancer among women, with approximately 330 new cases and 200 deaths reported annually. Despite the availability of effective HPV vaccines, national vaccination coverage remains low. This study aimed to evaluate HPV vaccination coverage, analyze cervical cancer incidence trends, and model the potential impact of increased vaccination uptake on cancer prevention outcomes in Georgia. Study design A retrospective observational study was conducted using national health registry data and modeling projections to assess the burden of cervical cancer and estimate the effect of scaled vaccination coverage. Methods National health databases were used to analyze HPV vaccination rates and cervical cancer incidence. Descriptive statistics, chi-square tests, and linear regression were applied to identify trends and disparities. Additionally, a dynamic transmission model was developed to simulate the 10-year impact of increasing HPV vaccination coverage on cervical cancer incidence. Results The crude cervical cancer incidence rate was 15.7 per 100,000 women, with an age-standardized rate of 10.6 per 100,000. In 2022, only 38 % of eligible girls aged 13–18 received the first HPV vaccine dose, and 26 % completed the second dose. Regional disparities in vaccination and screening were noted, and overall screening coverage declined to 13,890 women screened in 2022. Modeling showed that increasing vaccine coverage to 60 % could reduce cervical cancer incidence by 50 % (preventing ∼163 cases); coverage of 80 % and 90 % could reduce incidence by 70 % and 85 %, preventing 228 and 276 cases, respectively. Conclusion Low HPV vaccination uptake in Georgia (38 % first dose and 26 % dull coverage) and declining screening limit cervical cancer prevention. Modeling shows that increasing vaccination coverage to 60–90 % could prevent 163–276 cases over the next decade. Strengthening vaccination and screening strategies is essential to move forward elimination. Y1 - 2025 U6 - https://doi.org/10.1016/j.puhip.2025.100679 VL - 2025 IS - 10 SP - 1 EP - 4 PB - Elsevier ER - TY - CHAP A1 - Weich, Patrick A1 - Lobachev, Oleg T1 - PIN-a-Boo: Revealing Smartphone PINs via Segmentation and Hand Skeleton Tracking from Video Feeds T2 - WSCG 2025 Proceedings : Computer Science Research Notes - CSRN N2 - It is crucial to improve smartphone security, given the prevalence of sensitive information stored on them. This study presents an attack strategy that reveals smartphone PIN entries using computer vision and pattern recognition techniques. By leveraging modern segmentation and hand skeleton tracking, our method accurately identifies and analyzes finger movement patterns, even when partially obscured. We can reliably infer the entered PIN by combining these movement patterns with the smartphone’s position and the on-screen keypad layout. This approach significantly enhances shoulder-surfing attacks, requiring only a video recording of the entry process. Our attack requires much less specialized expertise, making it more accessible. We conclude by analyzing the method’s potential impact and its implications for public safety. Y1 - 2025 U6 - https://doi.org/10.24132/CSRN.2025-31 SN - 464-4617 SN - 464-4625 SP - 293 EP - 304 PB - WSCG CY - Plzen ER - TY - JOUR A1 - Pilgermann, Michael A1 - Stein, Stefan A1 - Schrader, Thomas A1 - Weber, Simon T1 - Angriffserkennung beim Betrieb von KRITIS gemäß IT-Sicherheitsgesetz 2.0 JF - Datenschutz und Datensicherheit N2 - Die Krankenhaus-IT sieht sich einer stetig zunehmenden Bedrohung der Sicherheit der Patientendaten und des Krankenhausbetriebes gegenüber. Der Gesundheitssektor gehört zu den Kritischen Infrastrukturen, die zunehmend strengeren Regelungen unterliegen. Jüngste Regulierung ist das IT-Sicherheitsgesetz 2.0, das unter anderem Maßnahmen zur Detektion von Angriffen fordert. Der Beitrag stellt vor, wie eine branchenspezifische Umsetzung dieser Anforderung aussehen könnte. Y1 - 2021 U6 - https://doi.org/10.1007/s11623-021-1525-z VL - 45 SP - 733 EP - 737 ER - TY - JOUR A1 - Stein, Stefan A1 - Pilgermann, Michael A1 - Sedlmayr, Martin T1 - Systematic Evaluation of Manufacturer Disclosure Statements for Medical Device Security (MDS2) to Strengthen Hospital OT Security Measures – Lessons Learned JF - Studies in health technology and informatics N2 - The growing number of connected medical devices in hospitals poses serious operational technology (OT) security challenges. Effective countermeasures require a structured analysis of the communication interfaces and security configurations of individual devices. State of the art: Although Manufacturer Disclosure Statements for Medical Device Security (MDS2, Version 2019) offer relevant information, they are rarely integrated into cybersecurity workflows. Existing studies are limited in scope and lack scalable methodologies for systematic evaluation. Concept: This study analyzed 209 MDS2 documents and 161 security white papers to extract structured information on ports, protocols, and protective measures. Over 52,000 question–answer pairs were converted into a machine-readable format using customized parsing and validation routines. The aim was to establish whether this dataset could inform risk assessments and future applications involving Large Language Models (LLMs). Implementation: The analysis revealed 367 distinct ports, including common protocols such as HTTPS (443), DICOM (104), and RDP (3389), as well as vendor-specific proprietary ports. Approximately 40% of the devices used over 20 ports, indicating a broad attack surface. OCR errors and inconsistent formatting required manual corrections. A consolidated dataset was developed to support clustering, comparison across vendors and versions, and preparation for downstream LLM use, particularly via structured SBOM and configuration data. Lessons learned: Although no model training was conducted, the structured dataset can support AI-based OT security workflows. The findings highlight the critical need for up-to-date, machine-readable manufacturer data in standardized formats and schemas. Such information could greatly enhance the automation, comparability, and scalability of hospital cybersecurity measures. Y1 - 2025 U6 - https://doi.org/10.3233/SHTI251404 VL - 331 SP - 256 EP - 264 ER - TY - JOUR A1 - Weber, Simon B. A1 - Stein, Stefan A1 - Pilgermann, Michael A1 - Schrader, Thomas T1 - Attack Detection for Medical Cyber-Physical Systems – A Systematic Literature Review JF - IEEE Access N2 - The threat situation due to cyber attacks in hospitals is emerging and patient life is at risk. One significant source of potential vulnerabilities is medical cyber-physical systems (MCPS). Detecting intrusions in this environment faces challenges different from other domains, mainly due to the heterogeneity of devices, the diversity of connectivity types, and the variety of terminology. To summarize existing results, we conducted a structured literature review (SLR) following the guidelines of Kitchenham et al. for SLRs in software engineering. We developed six research questions regarding detection approach, detection location, included features, adversarial focus, utilized datasets, and intrusion prevention. We identified that most researchers focused on an anomaly-based detection approach at the network layer. The primary focus was on the detection of malicious insiders. While several researchers used publicly available datasets for training and testing their algorithms, the lack of suitable datasets resulted in the development of testbeds consisting of various medical devices. Based on the results, we formulated five future research topics. First, the special conditions of hospital networks, the MCPS deployed within them, and the contrasts to other IT and OT environments should be examined. Thereupon, MCPS-specific datasets should be created that allow researchers to address the health domain’s unique requirements and possibilities. At the same time, endeavors aimed at standardization in this area should be supported and expanded. Moreover, the use of medical context for attack detection should be further explored. Last but not least, efforts for MCPS-tailored intrusion prevention should be intensified. This way, the emerging threat landscape can be addressed, IT security in hospitals can be improved, and patient health can be protected. Y1 - 2023 U6 - https://doi.org/10.1109/ACCESS.2023.3270225 IS - 11 SP - 41796 EP - 41815 PB - IEEE ER - TY - JOUR A1 - Jamil, Mahnoor A1 - Mihajloska Trpcheska, Hristina A1 - Popovska-Mitrovikj, Aleksandra A1 - Dimitrova, Vesna A1 - Creutzburg, Reiner T1 - Advancing Image Spam Detection: Evaluating Machine Learning Models Through Comparative Analysis JF - Applied Sciences N2 - Image-based spam poses a significant challenge for traditional text-based filters, as malicious content is often embedded within images to bypass keyword detection techniques. This study investigates and compares the performance of six machine learning models—ResNet50, XGBoost, Logistic Regression, LightGBM, Support Vector Machine (SVM), and VGG16—using a curated dataset containing 678 legitimate (ham) and 520 spam images. The novelty of this research lies in its comprehensive side-by-side evaluation of diverse models on the same dataset, using standardized dataset preprocessing, balanced data splits, and validation techniques. Model performance was assessed using evaluation metrics such as accuracy, receiver operating characteristic (ROC) curve, precision, recall, and area under the curve (AUC). The results indicate that ResNet50 achieved the highest classification performance, followed closely by XGBoost and Logistic Regression. This work provides practical insights into the strengths and limitations of traditional, ensemble-based, and deep learning models for image-based spam detection. The findings can support the development of more effective and generalizable spam filtering solutions in multimedia-rich communication platforms. KW - spam detection KW - image spam KW - machine learning KW - Support Vector Machine KW - XGBoost KW - Logistic Regression KW - ResNet50 KW - LightGBM KW - VGG16 Y1 - 2025 U6 - https://doi.org/10.3390/app15116158 SN - 2076-3417 VL - 15 IS - 11 PB - MDPI ER - TY - INPR A1 - Tuset-Peiro, Pere A1 - Pilgermann, Michael A1 - Pegueroles, Josep A1 - Vilajosana, Xavier T1 - Assessing Cybersecurity of Internet-Facing Medical IT Systems in Germany & Spain Using OSINT Tools N2 - This paper investigates cybersecurity threats in medical IT (Information Technology) systems exposed to the Internet. To that end, we develop a methodology and build a data processing pipeline that allows to gather data from different OSINT (Open Source Intelligence) sources, and processes it to obtain relevant cybersecurity metrics. To validate its operation and usefulness, we apply it to two countries, Germany and Spain, allowing to study the main threats that affect medical IT systems in these countries. Our initial findings reveal that 20% of German hosts and 15% of Spanish hosts tagged as medical devices have at least one CVE (Common Vulnerabilities and Exposures) with a CVSS (Common Vulnerability Scoring System) graded as critical (i.e., value 8 or greater). Moreover, we found that 74% of CVEs found in German hosts are dated from earlier than 2020, whereas for Spanish hosts the percentage is 60%. This indicates that medical IT systems exposed to the Internet are seldom updated, which further increases their exposure to cyberthreats. Based on these initial findings, we finish the paper providing some insights on how to improve cybersecurity of these systems. Y1 - 2025 U6 - https://doi.org/10.20944/preprints202503.1340.v1 SP - 190 EP - 197 ER - TY - JOUR A1 - Stein, Stefan A1 - Pilgermann, Michael A1 - Weber, Simon A1 - Sedlmayr, Martin T1 - Leveraging MDS2 and SBOM data for LLM-assisted vulnerability analysis of medical devices JF - Computational and Structural Biotechnology Journal N2 - This study investigated the use of a semi-automated, Retrieval-Augmented Generation (RAG)-based multi-agent architecture to analyze security-relevant data and assemble specialized exploitation paths targeting medical devices. The input dataset comprised device-specific sources, namely, the Manufacturer Disclosure Statement for Medical Device Security (MDS2) documents and Software Bills of Materials (SBOMs), enriched with public vulnerability databases, including Common Vulnerabilities and Exposures (CVE), Known Exploited Vulnerabilities (KEV), and Metasploit exploit records. The objective was to assess whether a modular, Large Language Model (LLM)-driven agent system could autonomously correlate device metadata with known vulnerabilities and existing exploit information to support structured threat modeling. The architecture follows a static RAG design based on predefined prompts and fixed retrieval logic, without autonomous agent planning or dynamic query adaptation. The developed Vulnerability Intelligence for Threat Analysis in Medical Security (VITAMedSec) system operates under human-prompted supervision and successfully synthesizes actionable insights and exploitation paths without requiring manual step-by-step input during execution. Although technically coherent results were obtained under controlled conditions, real-world validation remains a critical avenue for future research. This study further discusses the dual-use implications of such an agent-based framework, its relevance to patient safety in medical device cybersecurity, and the broader applicability of the proposed architecture to other critical infrastructure sectors. These findings emphasize both the technical potential and ethical responsibility for applying semi-automated AI workflows in medical cybersecurity contexts. Y1 - 2025 U6 - https://doi.org/10.1016/j.csbj.2025.07.012 VL - 28 SP - 267 EP - 280 PB - Elsevier ER - TY - CHAP A1 - Wagner, Robin A1 - Kitzelmann, Emanuel A1 - Boersch, Ingo T1 - Mitigating Hallucination by Integrating Knowledge Graphs into LLM Inference – a Systematic Literature Review T2 - Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop) N2 - Large Language Models (LLMs) demonstrate strong performance on different language tasks, but tend to hallucinate – generate plausible but factually incorrect outputs. Recently, several approaches to integrate Knowledge Graphs (KGs) into LLM inference were published to reduce hallucinations. This paper presents a systematic literature review (SLR) of such approaches. Following established SLR methodology, we identified relevant work by systematically search in different academic online libraries and applying a selection process. Nine publications were chosen for indepth analysis. Our synthesis reveals differences and similarities of how the KG is accessed, traversed, and how the context is finally assembled. KG integration can significantly improve LLM performance on benchmark datasets and additionally to mitigate hallucination enhance reasoning capabilities, explainability, and access to domain-specific knowledge. We also point out current limitations and outline directions for future work. KW - LLMs KW - hallucination KW - knowledge graphs KW - inference KW - literature review Y1 - 2025 UR - https://aclanthology.org/2025.acl-srw.53.pdf U6 - https://doi.org/10.18653/v1/2025.acl-srw.53 SP - 795 EP - 805 PB - Association for Computational Linguistics CY - Vienna ER - TY - JOUR A1 - Neubert, Tom A1 - Schueler, Eric A1 - Ullrich, Henning A1 - Buxhoidt, Laura A1 - Vielhauer, Claus T1 - Extended Analysis, Detection and Attribution of Steganographic Embedding Methods in Network Data of Industrial Controls Systems JF - International Journal on Advances in Security N2 - Since the last decade, it is well known that Industrial Control Systems (ICS) are under attack and attackers nowadays increasingly use stealthy malware (i.e., stegomalware) imple- mented by steganographic embedding methods to in- and exfil- trate hidden information. Unfortunately, current mechanisms to distinguish between network steganographic embedding methods and embedded message types need improvement for a potential attribution of attackers. For the analysis of steganographic em- bedding methods which are utilized in stealthy malware, the work presented in this paper builds upon a state-of-the-art analysis testbed proposed earlier, which is recapitulated here. It offers the opportunity to analyze network steganographic embedding methods in ICS to elaborate methods to detect and distinguish between them to gain forensic information for attribution of potential attackers and their methods. In this work, we introduce a novel machine learning based approach to distinguish between five selected embedding methods and two embedded message types. We use the analysis testbed to evaluate and determine the accuracy of the novel approach compared to a state-of-the-art approach. In our extensive evaluation, our novel approach has shown to be able to distinguish between network steganographic embedding methods with an average accuracy of 85.7%, which is an improvement in comparison to the state-of-the-art by +5.9% and enables a more accurate attribution of attackers. Additionally, the novel approach is able to improve the accuracy of distinction between embedding method and embedded message type by +9.3% in comparison to the evaluated state-of-the-art approach. Y1 - 2025 VL - 18 IS - 1&2 SP - 112 EP - 122 PB - IARIA ER - TY - CHAP A1 - Vielhauer, Claus A1 - Loewe, Fabian A1 - Pilgermann, Michael T1 - Towards Modeling Hidden & Steganographic Malware Communication based on Images T2 - IH&MMSEC '25: ACM Workshop on Information Hiding and Multimedia Security, San Jose, CA, USA, 2025 N2 - Recently, an increasing number of IT security incidents involving malware, which makes use of hidden and steganographic channels for malicious communication (a.k.a. as "stegomalware"), can be observed in the wild. Especially the use of images to hide malicious code is rising. In consideration of this shift, a new model is proposed in this paper, which aims to help security professionals to identify and analyze incidents revolving around steganographic malware in the future. The model focuses on practical aspects of steganalysis of communication data to elaborate linking properties to previous code analysis knowledge. The model features two distinct roles that interact with a knowledge base which stores malware features and helps building a context for the incident. For evaluation, two image steganography malware types are chosen from popular databases (malpedia and MITRE ATT&CK®), which are analyzed in multiple steps including steganalysis and code analysis. It is conceptually shown, how the extracted features can be stored in a knowledge base for later use to identify stegomalware from communication data without the need of a thorough code analysis. This allows to uncover previously hidden meta-information about the examined malicious programs, enrich the incident’s forensic context traces and thus allows for thorough forensic insights, including attribution and improved preventive security measures in the future. Y1 - 2025 UR - https://dl.acm.org/doi/pdf/10.1145/3733102.3733152 U6 - https://doi.org/10.1145/3733102.3733152 SP - 52 EP - 63 ER - TY - JOUR A1 - Prinz, Alexander A1 - Orlowski, Katja A1 - Beck, Eberhard A1 - Witte, Kerstin T1 - Serious Games for Seniors With Dementia: A Pilot Study JF - Journal of Applied Gerontology N2 - Dementia, marked by cognitive decline, significantly impacts daily life. With global prevalence rising, traditional treatments manage symptoms but have side effects and offer no cure. Non-pharmacological interventions, like serious games, are gaining importance. This study assesses the feasibility and benefits of serious games for people with mild to moderate dementia over a 10-week intervention. Sixty-one patients were recruited, with 35 completing the study. The intervention included six games focusing on physical and cognitive training. Outcome measures were motor function, cognitive assessments, quality of life, and depression. Results showed significant improvements in dynamic balance (p = .013) but no significant changes in other measures. The findings suggest that serious games are feasible and can improve motor functions like balance. However, short intervention periods may limit their impact on cognitive function and quality of life. Longer interventions and personalized game designs are recommended for greater benefits. Y1 - 2025 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:522-opus4-33440 VL - 2025 SP - 1 EP - 11 PB - SAGE ER - TY - CHAP A1 - Kitzelmann, Emanuel T1 - Lernen mit Neuronalen Netzen T2 - Furbach, U., Kitzelmann, E., Michaeli, T., Schmid, U. (Hrsg.) Künstliche Intelligenz für Lehrkräfte. N2 - Sitz der menschlichen Intelligenz ist das Gehirn. Was läge also näher, als Erkenntnisse aus der Hirnforschung für die Erzeugung künstlicher Intelligenz zu nutzen? Künstliche neuronale Netze simulieren in einer vereinfachten und abgewandelten Form den grundlegenden Aufbau des Gehirns und sind als eine leistungsfähige Variante des maschinellen Lernens vielseitig einsetzbar, um komplexe Funktionen, z. B. in der Bild- und Sprachverarbeitung, aus großen Mengen von Daten zu lernen. Viele aktuelle und Aufsehen erregende Anwendungen der künstlichen Intelligenz – leistungsfähige ChatBots und Bildgeneratoren, Bilderkennung in Fahrassistenzsystemen, die Erkennung von Krankheitsbildern in der bildgebenden Diagnostik, Prognosen im Finanzwesen und viele mehr – beruhen auf künstlichen neuronalen Netzen. Dieses Kapitel behandelt den Grundaufbau eines einzelnen künstlichen Neurons und eines mehrschichtigen neuronalen Netzes. Es wird erklärt, wie ein neuronales Netz Eingaben verarbeitet und wie es mit Daten trainiert werden kann, um eine bestimmte Funktion auszuführen. Y1 - 2024 U6 - https://doi.org/10.1007/978-3-658-44248-4_5 SP - 53 EP - 67 PB - Springer ER - TY - INPR A1 - Pepe, Alberto A1 - Lasenby, Joan A1 - Buchholz, Sven T1 - STAResNet: A Network in Spacetime Algebra to solve Maxwell's PDEs T2 - arXiv N2 - We introduce STAResNet, a ResNet architecture in Spacetime Algebra (STA) to solve Maxwell's partial differential equations (PDEs). Recently, networks in Geometric Algebra (GA) have been demonstrated to be an asset for truly geometric machine learning. In [1], GA networks have been employed for the first time to solve partial differential equations (PDEs), demonstrating an increased accuracy over real-valued networks. In this work we solve Maxwell's PDEs both in GA and STA employing the same ResNet architecture and dataset, to discuss the impact that the choice of the right algebra has on the accuracy of GA networks. Our study on STAResNet shows how the correct geometric embedding in Clifford Networks gives a mean square error (MSE), between ground truth and estimated fields, up to 2.6 times lower than than obtained with a standard Clifford ResNet with 6 times fewer trainable parameters. STAREsNet demonstrates consistently lower MSE and higher correlation regardless of scenario. The scenarios tested are: sampling period of the dataset; presence of obstacles with either seen or unseen configurations; the number of channels in the ResNet architecture; the number of rollout steps; whether the field is in 2D or 3D space. This demonstrates how choosing the right algebra in Clifford networks is a crucial factor for more compact, accurate, descriptive and better generalising pipelines. Y1 - 2024 UR - https://arxiv.org/pdf/2408.13619 SP - 1 EP - 17 PB - arXiv ER - TY - CHAP A1 - Pepe, Alberto A1 - Lasenby, Joan A1 - Buchholz, Sven T1 - CGAPoseNet+GCAN: A Geometric Clifford Algebra Network for Geometry-aware Camera Pose Regression T2 - Winter Conference on Applications of Computer Vision, Waikoloa, Hawaii, US, 2024 N2 - We introduce CGAPoseNet+ GCAN, which enhances CGAPoseNet, an architecture for camera pose regression, with a Geometric Clifford Algebra Network (GCAN). With the addition of the GCAN we obtain a geometry-aware pipeline for camera pose regression from RGB images only. CGAPoseNet employs Clifford Geometric Algebra to unify quaternions and translation vectors into a single mathematical object, the motor, which can be used to uniquely describe camera poses. CGAPoseNet solves the issue of balancing rotation and translation components in the loss function, and can obtain comparable results to other approaches without the need of expensive tuning of the loss function or additional information about the scene, such as 3D point clouds, which might not always be available. CGAPoseNet, however, like several approaches in the literature, only learns to predict motor coefficients, and it is unaware of the mathematical space in which predictions sit in and of their geometrical meaning. By leveraging recent advances in Geometric Deep Learning, we modify CGAPoseNet with a GCAN: proposals of possible motor coefficients associated with a camera frame are obtained from the InceptionV3 backbone, and the GCAN downsamples them to a single motor through a sequence of layers that work in G_ 4, 0. The network is hence geometry-aware, has multivector-valued inputs, weights and biases and preserves the grade of the objects that it receives in input. CGAPoseNet+ GCAN has almost 4 million fewer trainable parameters, it reduces the average rotation error by 41% and the average translation error by 8.8% compared to CGAPoseNet. Similarly, it reduces rotation and translation errors by 32.6% and 19.9%, respectively, compared to the best performing PoseNet strategy. CGAPoseNet+ GCAN reaches the state-of-the-art results on 13 commonly employed datasets. To the best of our knowledge, it is the first experiment in GCANs applied to the problem of camera pose regression. Y1 - 2024 U6 - https://doi.org/10.1109/WACV57701.2024.00646 SP - 6593 EP - 6603 PB - IEEE ER - TY - CHAP A1 - Pepe, Alberto A1 - Buchholz, Sven A1 - Lasenby, Joan T1 - Clifford Group Equivariant Neural Network Layers for Protein Structure Prediction T2 - Northern Lights Deep Learning Conference, VOL 233, 09-11 Januar 2024, Tromso, Norwegen N2 - We employ Clifford Group Equivariant Neural Network (CGENN) layers to predict protein coordinates in a Protein Structure Prediction (PSP) pipeline. PSP is the estimation of the 3D structure of a protein, generally through deep learning architectures. Information about the geometry of the protein chain has been proven to be crucial for accurate predictions of 3D structures. However, this information is usually flattened as machine learning features that are not representative of the geometric nature of the problem. Leveraging recent advances in geometric deep learning, we redesign the 3D projector part of a PSP architecture with the addition of CGENN layers . CGENNs can achieve better generalization and robustness when dealing with data that show rotational or translational invariance such as protein coordinates, which are independent of the chosen reference frame. CGENNs inputs, outputs, weights and biases are objects in the Geometric Algebra of 3D Euclidean space, i.e. G3,0,0, and hence are interpretable from a geometrical perspective. We test 6 approaches to PSP and show that CGENN layers increase the accuracy in term of GDT scores by up to 2.1\%, with fewer trainable parameters compared to linear layers and give a clear geometric interpretation of their outputs. Y1 - 2024 UR - https://proceedings.mlr.press/v233/pepe24a/pepe24a.pdf SP - 1 EP - 7 PB - PMLR ER -