openHSU (HSU Hamburg)
Not a member yet
7291 research outputs found
Sort by
Enhancing military decision-making through simulation-guided post-training of Large Language Models
No full textThe capabilities of Large Language Models (LLMs) have rapidly evolved, enabling them to perform increasingly complex reasoning tasks. However, while their general reasoning abilities are shaped during large-scale pretraining, domain-specific reasoning such as tactical decision-making in military contexts requires dedicated post-training. This paper introduces a simulation-guided Reinforcement Fine-Tuning (RFT) approach in which reward signals are derived from the outcomes of a combat simulation environment. By embedding a military-grade simulator into the RFT loop via the Group Relative Policy Optimization (GRPO) algorithm, model outputs are evaluated based on tactical effectiveness rather than human annotations or rule-based correctness. A proof-of-concept study demonstrates that the method significantly improves the feasibility and tactical quality of generated Courses of Action (COAs), even under limited training schedules. These findings establish simulation-guided RFT as a promising direction for equipping LLMs with tactically relevant reasoning skills and pave the way toward next-generation decision-support systems in military environments.Vo
Statistisch optimierte akustische Nahfeldholografie (SONAH) für die Ermittlung von Schallfeldgrößen auf der Oberfläche von Schallquellen
No full textIn dieser Arbeit wird die statistisch optimierte akustische Nahfeldholografie (SONAH) zur Rekonstruktion von Schallfeldgrößen auf der Oberfläche vibrierender Strukturen untersucht. Ziel war die inverse Bestimmung der Schallschnelle auf einer Plattenoberfläche anhand zuvor im Nahfeld gemessener Schalldrücke. Hierzu wurde eine bestehende MATLAB-Implementierung erweitert und durch eine Reduktion der Integrationsdimension effizienter gestaltet. In Simulations- und Messreihen konnte gezeigt werden, dass die Methode prinzipiell zur Bestimmung der abgestrahlten Schallleistung geeignet ist. Die Ergebnisse weisen jedoch auf einen frequenzabhängigen, skalierenden Fehler in der Rekonstruktion hin, der den Betrag der Schallschnelle und dadurch auch die Schallleistungsbestimmung beeinflusst. Trotz dieser Einschränkungen konnte die Schallleistung einer Platte mit akzeptabler Übereinstimmung zu normgerechten Messungen nach DIN EN ISO 3745 bestimmt werden. Die Arbeit liefert damit eine Grundlage für den zukünftigen Einsatz der SONAH als berührungsloses Sensorkonzept in aktiven Lärmminderungsmaßnahmen.Vo
Institutionelle Ausdifferenzierung der öffentlichen Aufgabenerfüllung in Deutschland: Ausgliederung, Privatisierung und Rekommunalisierung
No full textVo
A review on machine learning approaches for the prediction of glucose levels and hypoglycemia
No full textType 1 Diabetes (T1D) is an autoimmune disease leading to insulin insufficiency. Thus, patients require lifelong insulin therapy, which has a side effect of hypoglycemia. Hypoglycemia is a critical state of decreased blood glucose levels (BGL) below 70 mg/dL and is associated with increased risk of mortality. Machine learning (ML) models can improve diabetes management by predicting hypoglycemia and providing optimal prevention methods. ML models are classified into regression and classification based, that forecast glucose levels and identify events based on defined labels, respectively. This review investigates state-of-the-art models trained on data of continuous glucose monitoring (CGM) devices from patients with T1D. We compare the models' performance across short-term (15 to 120 min) and long term (3 to more than 24 hours) prediction horizons (PHs). Particularly, we explore: 1) How much in advance can glucose values or a hypoglycemic event be accurately predicted? 2) Which models have the best performance? 3) Which factors impact the performance? and 4) Does personalization increase performance? The results show that 1) a PH of up to 1 hour provides the best results. 2) Conventional ML methods yield the best results for classification and DL for regression. A single model cannot adequately classify across multiple PHs. 3) The model performance is influenced by multivariate datasets and the input sequence length (ISL). 4) Personal data enhances performance but due to limited data quality population-based models are preferred.SMU
Hyperwar and the dehumanisation of war
No full textArtificial intelligence is completely transforming the general picture of warfare and combat operations. These changes not only present the German Armed Forces with technical and organizational challenges but also call into question its entire leadership philosophy. If decisions can only be made at machine speed, the role and self-conecption of soldiers must be taken into account in this process of change. This requires a interdisciplinary discourse, the framework for which must also be developed. This is because the scientific debate about terminology and arguments has a direct impact on human-machine interaction on today’s and tomorrow’s battlefields. Discussions about responsibility and functionality are therefore not theoretical but extremely real turf wars that will help determine how the military will use artificial intelligence in the future.Vo
Monitoring military vehicle depots
No full textThis paper presents a CNN-based approach for the localization and counting of closely parked military vehicles in Synthetic Aperture Radar (SAR) imagery. A significant challenge in this task is the separation of small, densely positioned vehicles with partially overlapping signatures. To address this, the problem is formulated as a segmentation task, and a U-Net model is trained using point-level annotations to predict vehicle locations. To mitigate the impact of class imbalance between vehicle positions and the background, the
Tversky loss function is employed, which applies a greater penalty for missed detections of the underrepresented class. Finally, vehicle counts are derived from the predicted segmentation masks by counting connected components. To validate the approach, SAR simulation is used to generate a synthetic dataset comprising labeled SAR image data that depicts military vehicle depots from different perspectives. This dataset enables model training and allows for a comprehensive evaluation of the overall concept. The results demonstrate that the method successfully detects, separates and counts densely parked vehicles. The model trained on simulated data provides a suitable foundation for the subsequent exploitation of real image data in the future.Vo
In tension between traditional gender roles and societal change : academic career prospects for bourgeois girls in the Weimar Republic
No full textDie Weimarer Republik erwies sich als eine Zeit des gesellschaftlichen Wandels. Reformen im Bildungswesen und Veränderungen des Arbeitsmarktes eröffneten neue Perspektiven für Mädchen im Hinblick auf Berufstätigkeiten, insbesondere in sozialen, pädagogischen und kaufmännischen Berufen. Diese Entwicklung war jedoch von Ambivalenzen geprägt, da traditionelle Rollenvorstellungen ihren Einfluss auf Karrierewege und soziale Akzeptanz nicht verloren. Die vorliegende Arbeit analysiert anhand von zeitgenössischen Fotografien und biografischen Zeugnissen, welche akademischen Perspektiven sich für bürgerliche Mädchen im Spannungsfeld von Emanzipationsbestrebungen, Weiblichkeitsidealen, Bildungsexpansion, traditionellen Geschlechterrollen und gesellschaftlichen Restriktionen ergaben. Dabei wird deutlich, dass die beruflichen Chancen bürgerlicher Frauen zwar wuchsen, aber strukturelle Ungleichheiten und normative Barrieren fortbestanden.Vo
A proposed paradigm for imputing missing multi-sensor data in the healthcare domain
No full textChronic diseases such as diabetes pose significant management challenges, particularly due to the risk of complications like hypoglycemia, which require timely detection and intervention. Continuous health monitoring through wearable sensors offers a promising solution for early prediction of glycemic events. However, effective use of multisensor data is hindered by issues such as signal noise and frequent missing values. This study examines the limitations of existing datasets and emphasizes the temporal characteristics of key features relevant to hypoglycemia prediction. A comprehensive analysis of imputation techniques is conducted, focusing on those employed in state-of-the-art studies. Furthermore, imputation methods derived from machine learning and deep learning applications in other healthcare contexts are evaluated for their potential to address longer gaps in time-series data. Based on this analysis, a systematic paradigm is proposed, wherein imputation strategies are tailored to the nature of specific features and the duration of missing intervals. The review concludes by emphasizing the importance of investigating the temporal dynamics of individual features and the implementation of multiple, feature-specific imputation techniques to effectively address heterogeneous temporal patterns inherent in the data.SMU
Risk-aware autonomous defence with generative AI
No full textThe integration of artificial intelligence into industrial cybersecurity creates unique challenges at the intersection of technical efficiency, accountability, and ethics—especially in critical infrastructures governed by regulations like NIS2 and ISO 27001. We introduce AIOD, an autonomous cyber defence system that combines large language models (LLM), retrieval-augmented generation (RAG), and a risk-based decision framework derived from ISO 27001. Unlike conventional Endpoint Detection and Response (EDR) or Security Operations Center (SOC) playbooks, AIOD autonomously interprets incidents, selects organisational risk policies, and generates executable mitigation code under auditable constraints. Evaluated in a controlled Industry 4.0-inspired testbed with replayed events, AIOD achieved sub-minute containment of port scans and sub-10-minute mitigation for malware surrogates (EICAR-based), demonstrating potential for near-real-time response under simplified but representative conditions. Our findings highlight not only the technical potential of LLM-based cyber defence but also raise critical questions around safety, the creative interpretation of vague policies, and human oversight—underscoring the need for governance frameworks to responsibly deploy such systems in cybersecurity and critical infrastructure.Vo