2972 research outputs found
Sort by
Wirtschaftsflüchtling
No full textDie von der Bundesrepublik Deutschland unterzeichnete UN-Flüchtlingskonvention von 1951 definiert ‚Flüchtlinge‘ als Personen mit einer „begründeten Furcht vor Verfolgung wegen ihrer Rasse, Religion, Nationalität, Zugehörigkeit zu einer bestimmten sozialen Gruppe oder wegen ihrer politischen Überzeugung“ (Abkommen über die Rechtsstellung der Flüchtlinge 1951: Artikel 1). Die BRD hatte das Recht auf Asyl bereits in ihrem Grundgesetz von 1949 anerkannt: „Politisch Verfolgte genießen Asylrecht“ (Grundgesetz für die Bundesrepublik Deutschland, Artikel 16a Absatz 1). In beiden Grundlagendokumenten ist die individuelle Verfolgung Voraussetzung für den Flüchtlingsstatus.
Der Ausdruck Wirtschaftsflüchtling ist im Gegensatz dazu rechtlich nicht präzise definiert. Er entstand in der Nachkriegszeit im deutschsprachigen Raum aus der Praxis der Asylentscheidungen heraus. Menschen aus dem sogenannten Ostblock, die im Westen Asyl beantragten, von denen aber angenommen wurde, dass sie keine politischen Beweggründe hatten, wurden Wirtschaftsflüchtlinge genannt. In manchen Fällen diente der Begriff als Argument für eine flexiblere Handhabung des Flüchtlingsbegriffs, in anderen Fällen sollte er den Anspruch auf ein Bleiberecht delegitimieren. Mit der Zeit wurde der Begriff immer öfter verwendet, um die Legitimität von Asylgesuchen in Frage zu stellen
‘Digitality For Future’? – The Potential of Digital Practices for Transformative Education in the Area of Climate Change
No full textThe Fridays For Future movement brought the issue of climate crisis an enormous boost in
public awareness and generated great interest in climate-related topics among students.
The particular way in which issues of climate protection or climate adaptation are
negotiated by social media communities ensures their appeal. Two questions arise from this:
Which digital practices do activists of Fridays For Future use to disseminate knowledge on
climate and climate change? How do these practices affect informal learning processes in
the knowledge domain of climate and climate change? Results obtained from a qualitative
content analysis of posts and documents created by members of digital communities of the
Fridays For Future movement and interviews with activists show that processes of identity
formation take place in the communities. If these knowledge practices are taken up and
reflected upon in the classroom, they can be transferred to formal educational processes,
in terms also of transformative learning
Long-Range Dependence in Geometric Models Driven by Poisson Point Processes
No full textIn this thesis, we study limit theorems for two models based on Poisson point processes. The first model is a dynamic Boolean model, obtained by taking the Minkowski sum of a Poisson-distributed collection of lines with a fixed compact set. Our main result describes the asymptotic behavior of the minimum cylinder width required to cover the
d-dimensional unit cube. The second model concerns random geometric graphs projected onto a plane. We investigate the number of edge crossings in such projections and show that, under suitable conditions, the resulting crossing point process converges to a Poisson point process. In both the thermodynamic and dense regimes, we establish a multivariate central limit theorem for the number of crossings and for another Poisson functional, the stress of the graph drawing. A key challenge in analyzing these models is the presence of strong spatial dependence, which complicates the derivation of limiting distributions
Remigration
Full text linkIm Jahr 2024 führte das Bekanntwerden eines Remigrationsplans rechter Kreise in Deutschland zu landesweiten Protesten für den Erhalt einer vielfältigen Gesellschaft. Der Begriff Remigration, fälschlicherweise als neutraler Ausdruck für die freiwillige Rückkehr von Migrant:innen präsentiert, entpuppte sich rasch als Standardvokabel im Grundwortschatz der neo-völkischen Rechten. Diese prägte den Begriff im Rahmen ihres Vorhabens, mit Hilfe von Massendeportationen eine ethnisch definierte Volksgemeinschaft zu schaffen und deren ‚Reinhaltung‘ durch Segregation von als fremd wahrgenommenen Kulturen oder ‚Rassen‘ zu erreichen. Der Remigrationsbegriff ist untrennbar mit völkisch-rassistischen Deportationsplänen verbunden, die eine ‚ethnopluralistische‘ Weltordnung anstreben, die einer globalen Apartheid ähnelt. Diese Kaschierung von Deportationen als ‚Remigration‘ ist keineswegs neu, sondern hat historische Wurzeln in völkisch-rassistischen Praktiken des 19. Jahrhunderts. Abgesehen vom Nationalsozialismus fanden diese Praktiken bereits Anwendung im Rahmen rassistischer Segregationspolitiken wie dem transatlantischen Versklavungssystem, dem weißen Siedlungskolonialismus und in Apartheidstaaten wie Südafrika. Seit der Dekolonisierung in den 1960er Jahren propagiert die neo-völkische Rechte ‚Remigration‘ als vermeintliche Lösung für das fiktive Problem einer ‚umgekehrten Kolonisierung‘, das auf einem rassistischen Verschwörungsnarrativ basiert. In diesen sogenannten Remigrationsplänen sind Massendeportationen von nicht-weißen Menschen und deren Unterstützer:innen vorgesehen. Selbst scheinbar nicht-völkische Elemente wie freiwillige Rückkehrprogramme, Startgeld oder die Schaffung von ‚Musterstädten‘ in Afrika lassen sich auf völkische und koloniale Deportationspraktiken und Apartheidsysteme zurückführen
Zero-Click AI Attacks as Emerging Cyber Threat
Full text linkThis working paper analyzes Zero-Click AI Attacks as a new and emerging cyber threat, shows the characteristics of this attack methods and its variants as well as the cyber security measures. Artificial Intelligence (AI) is commonly understood as the ability of machines to perform tasks that normally require human intelligence. A widespread AI application is the Generative AI (GenAI) with Large Language Models (LLMs) that create content based on short instructions (prompts), which are a key vulnerability if malicious instructions are given. Attempts to circumvent restrictions (jailbreaks) are done by prompt injections, i.e., special instructions to AI to provide restricted or malicious content. Typical prompt injections are direct commands like DAN (do anything now), misleading the AI by fictional situations, and reverse psychology to facilitate the creation of malware, viruses, ransomware, and other malicious applications. AI Agents are software programs where LLMs with reasoning ability have increasing autonomy to achieve given objectives. An important feature of AI Agents is the retrieval augmented generation (RAG), where LLMs can access external sources like documents, web sources, or external databases to generate information. Originally, cyber attacks were based on computers that were used to attack other computers. Then, attackers used prompt injections to use AI applications against their targets. In 2025, a new cyber threat emerged where the attacker does not use its own AI against another computer, but attacks instead the local AI model on the target computer. This new attack method is characterized by zero clicks, i.e., the victim does not need to click on anything to execute the attack, and by hidden text which is not legible for humans, but legible for the AI (like white-on-white letters with tiny font of 1pt size). The hidden text (word text or HTML) can be embedded into emails, PDF files, images, videos, websites etc. and contains prompts for the local AI model on the victim computer to exfiltrate sensitive data and to execute other malicious commands. As these prompts are not directly given by the attacker to the targeted AI, they are known as indirect prompt injections. The most prominent examples are EchoLeak and AgentFlayer which exploit the connection between AI Agents and other applications to get access to these applications and their data. However, evaluations in August 2025 have shown that there are lots of possible attack variants and potential targets. This is a relevant cyber threat as every computer with a local AI application/AI Agent can be approached while conventional cyber security is circumvented. Another aspect is the simplicity of attacks and the limited efforts needed by the attackers. Cyber security countermeasures include detection and removal of hidden elements and instructions (prompt sanitization), definition of authorized users who are allowed to prompt the AI, restricting and monitoring of AI connector functions, blocking of AI-generated emails, repositories of malicious websites or whitelisting with use of safe websites only and lists of malicious prompts. In summary, the rapid expansion of AI, AI Agents, multi-agentic systems, and their capabilities results in a rapid growth of cyber threats and attack variants which is an important cyber security issue
Single-file transport in driven lattice gases and continuous-space Brownian dynamics
No full textStochastic particle transport processes occur in a wide variety of systems, ranging from physics, chemistry, and biology to technological applications. Geometric constraints often prevent particles from overtaking one another. For example, particles confined to a channel can be too large to pass each other. This causes the order of particles to be preserved, making it a single-file transport process. In this thesis, representative lattice models and models with continuous-space dynamics are investigated for this type of transport. New simulation methods for single-file dynamics of interacting particles with a hard core in general force fields are developed. Clusters of particles are moved in these novel Brownian cluster dynamics simulations. They are particularly efficient at high particle densities and allow also for studies of deterministic motion in the absence of thermal noise. For Brownian motion of hard spheres, an adhesive contact interaction is shown to speed up subdiffusion at large times, while slowing down normal diffusion at short times. These phenomena are explained based on the isothermal compressibility and the distribution of particle cluster sizes, leading to a full description of the time-dependent mean squared displacement through scaling relations. For driven lattice gases with repulsive nearest-neighbor interactions, stochastic motion of domain walls separating coexisting extremal current phases is investigated. The width of the fluctuations of the domain wall position grows subdiffusively at short times. At long times, the width saturates at a value growing sublinearly with the system size. This implies that the saturated width diverges in the thermodynamic limit of infinite system size, while the width relative to the system size becomes infinitely sharp. This phenomenon is referred to as weak pinning of the domain wall. The subdiffusive behavior is traced back to long-range anticorrelated current fluctuations and the weak pinning effect to an effective restoring force on the domain wall toward its mean position. It is shown that an Ornstein-Uhlenbeck process with a long-time anticorrelated noise term accounts for the overall scaling behavior of the domain wall fluctuations. To corroborate the findings, extensive simulations are performed for determining correlations between density and current fluctuations in driven lattice gases and for Brownian motion of hard spheres in both equilibrium and nonequilibrium steady states. Two different scaling behaviors are observed: scaling according to the Edwards-Wilkinson universality class occurs in equilibrium systems, and it applies also to the special case of nonequilibrium steady states of driven hard spheres in flat potentials. For nonequilibrium steady states in systems with nonlinear dependence of current on particle density, the scaling behavior of the correlation functions falls into the Kardar-Parisi-Zhang universality class
Urban 3D Reconstruction in Remote Sensing via Deep Learning and Dataset Enhancement Strategies
No full textModelling the profile of a city has been widely studied by the research community, particularly
in remote sensing. By using sensors located on airborne and satellite platforms, it is possible
to retrieve data such as optical/infrared images, radar and laser measurements, etc. Many of
these sensors can be used to compute the 3D profile of the scene. Radar and LiDAR are able
to measure the distance with high accuracy, but the reconstruction might be sparse, include
outliers and uses expensive technology. Images on the contrary are relatively cheaper and
capture geometric details, useful for a dense reconstruction. Nonetheless, the reconstruction
depends on the matching capabilities of the applied algorithm, as the depth has to be computed
from the displacement of corresponding pixels in the images.
Before the deep learning solutions, algorithms such as Semi-Global Matching or those based
on Structure from Motion used to lead the reconstruction benchmarks. These conventional
algorithms can be implemented on any set of images without any prior knowledge of the scene
and the refinement process, which benefit from geometric principles to detect inconsistencies and
occlusions, generate an accurate digital surface models with few remaining outliers. However,
conventional approaches fail in complicated areas such as those with poor texture, repetitive
patters, reflective surfaces, that are common in remote sensing imagery.
In contrast, deep learning approaches deal better with complicated areas and by using contextual
information, they are able to reconstruct a smooth 3D profile with few outliers and high accuracy.
Yet, learning based algorithms might fail if the differences between the training and testing
sets are large. In addition, neural networks require a large amount of quality data for a robust
training, which is not easy to collect for remote sensing platforms. What is more, ground truth
might still be obtained with laser but for smaller regions, leading to domain shifts.
Hence, the first step to set a reliable framework to evaluate reconstruction algorithms is to
provide high quality data. As this is expensive in a real scenario, this study proposes the use of
a pipeline to generate large amounts of synthetic data to train stereo matching and multi-view
stereo (MVS) networks. Since the data is rendered from software, accurate ground truth is
available. Moreover, as the software allows editions of the virtual scene, the urban growth can
be simulated, which helps to create data for additional tasks like change detection.
A reliable dataset allows to set up experiments to evaluate the quality of the reconstruction
algorithms. This dissertation considers two main research directions to design these experiments.
On the one hand, it is important to explore the advantages of both the conventional and the
learning based solutions, which are evaluated for the stereo matching case. On the other
hand, a comparison between the stereo and MVS algorithms is conducted. Intuitively, using
complementary information as MVS does might produce a more robust result, but stereo
methods have been more studied and have a simplified matching case. Therefore, conventional
and learnable, stereo and MVS algorithms are analysed with reliable datasets to assess how these
contribute to the 3D reconstruction task. Furthermore, an alternative case to fuse height values
into a final digital surface model is explored, where the confidence for the values predicted by
the neural networks is estimated and used to guide the fusion.
Valuable insights into the urban 3D reconstruction were obtained from the carried out experiments. The generation of datasets from real and synthetic scenarios facilitated the analysis
of the capabilities of the tested algorithms. Despite the well-known problem of the domain
gap, the networks trained on the generated datasets produced good reconstruction results in
complex regions. Buildings and man-made structures benefit from the synthetic models, but
for vegetation and natural elements the algorithms exhibit a lower performance because such
elements are simplified in the 3D modelling.
Among the methods tested, stereo matching approaches computed reconstructions that were
less prone to outliers, while the MVS was more robust for edge discontinuities. However,
learning algorithms estimate a value for each pixel in the input images, but the reliability of this
estimation should still be assessed. By pre-selecting the predicted values based on a confidence
estimation, the accuracy of the fusion was improved for the stereo matching case. Yet, this
fusion strategy needs to be further explored to generalize to MVS methods as well
Langlebigkeitsforschung und KI
No full textDieses Arbeitspapier analysiert die jüngsten Fortschritte in der Alterungsforschung, der Entwicklung von Langlebigkeitsmedikamenten und diesbezüglichen Werkzeugen der Künstlichen Intelligenz (KI) sowie die potenziellen medizinischen und gesellschaftlichen Auswirkungen. Derzeit liegt die theoretische Lebenserwartung des Menschen bei 120 Jahren, wobei die tatsächliche Lebenserwartung deutlich geringer ist. Angesichts des demografischen Wandels mit dem erwarteten Arbeitskräftemangel und den Problemen der Sozialsysteme besteht sowohl ein medizinisches als auch ein gesellschaftliches Interesse an Langlebigkeitsmedikamenten, die zu einem längeren und gesünderen Leben führen. Altern ist ein biologischer Prozess mit fortschreitendem Verfall, der zu Funktionseinschränkungen und schließlich zum Tod führt. Nach aktuellem Wissensstand ist Altern eine Anreicherung alter (seneszenter) Zellen mit Funktionsstörungen, chronischen Entzündungen und daraus resultierenden altersbedingten Erkrankungen: Der Rückgang des Redoxfaktors Nicotinamidadenindinukleotid NAD sowie Umwelteinflüsse verursachen oxidativen Stress, der Zellen und Gene schädigen und die Sekretion von Entzündungsfaktoren fördern kann, die als senescence-associated secretory phenotype (SASP) bekannt sind. Die Entzündung ist steril und wird als „Inflammaging“ (Entzündung und Alterung) bezeichnet. Das tiefere Verständnis des Alterns förderte die Suche nach Langlebigkeitsmedikamenten wie Senolytika, Senomorphika und Telomerasemodulatoren. Senolytika induzieren selektiv die Apoptose (Selbstzerstörung) seneszenter Zellen, und klinische Studien mit Dasatinib, Quercetin und Fisetin wurden begonnen. Senomorphika verändern das Sekretionsmuster seneszenter Zellen, ohne diese zu zerstören; eine Gruppe sind entzündungshemmende Medikamente wie Rapamycin, Steroide usw., während eine andere Gruppe Wachstumsfaktorhemmer wie Metformin und Resveratrol sind. Andere Forscher konzentrieren sich auf die Telomere und das Reparaturenzym Telomerase. Während viele Substanzen wie Vitamine, Antioxidantien und Probiotika dafür bekannt sind, zu einem gesünderen Leben beizutragen, konnte nicht nachgewiesen werden, dass sie das menschliche Leben signifikant und zuverlässig verlängern. Darüber hinaus wurden einige Substanzen bisher nur an Tieren getestet und müssen noch am Menschen getestet werden, während andere bereits für andere Zwecke am Menschen verwendet wurden, Nebenwirkungen aber eine Umwidmung dieser Medikamente verhindern. Aus diesen Gründen werden künstliche Intelligenz (KI)-Tools zunehmend eingesetzt, um die Medikamentenentwicklung voranzutreiben und in den nächsten Jahren große Fortschritte in der Forschung zu Langlebigkeitsmedikamenten zu erzielen. Im Jahr 2025 präsentierten OpenAI und das Start-up Retro Biosciences mit GPT-4b ein fortschrittliches KI-Modell, das die Wirksamkeit von Stammzellentwicklungsproteinen, den sogenannten Yamanaka-Faktoren, um das 50-fache steigerte. Dies ist ein bedeutender Schritt in Richtung Gewebereparatur und -erneuerung. Andere Forscher erhoffen sich von den KI-Fortschritten noch weit größere Erfolge. Obwohl dies noch Spekulation ist, rückt die Wahrscheinlichkeit einer Lebensverlängerung nun näher. OpenAI erwartet, im Jahr 2026 eine kreative KI auf den Markt zu bringen, die selbstständig neue Dinge entwickeln kann. Sollte einer kreativen KI ein großer Durchbruch gelingen, könnte ein neues Langlebigkeitsmedikament beispielsweise die Lebenserwartung um 20 Jahre erhöhen, was zu einer Verdoppelung der Rentendauer führen würde, dann allerdings eine Kürzung der Rentenzahlungen oder eine Verlängerung der Arbeitszeit durch Anhebung des Renteneintrittsalters auf 90 Jahre erfordern würde. Dies könnte den Arbeitskräftemangel abmildern, aber auch zu einer langfristigen Blockade von Führungspositionen in Politik und Wirtschaft führen. Zudem ist unklar, wer wirklich bis 90 arbeiten könnte, da Medikamente zur Steigerung des Lebensalters zwar altersbedingte Krankheiten reduzieren können, nicht aber andere Krankheiten. Staaten könnten zu dem Schluss kommen, dass dies nicht zu bewerkstelligen ist und der Zugang eingeschränkt werden sollte, was ebenfalls zu erheblichen sozialen Spannungen führen könnte. Auch aus ethischer Sicht ist die Entscheidung, wer länger leben darf, äußerst kritisch. Zusammenfassend lässt sich sagen: Medikamente zur Steigerung des Lebensalters könnten dem Einzelnen durch ein längeres und gesünderes Leben nützen, aber auch erhebliche strukturelle Belastungen für die Gesellschaft mit sich bringen. Die rasanten Fortschritte bei KI-Tools für die Genom- und Proteinanalyse sowie die Medikamentenforschung werden die Entwicklung von Medikamenten zur Steigerung des Lebensalters beschleunigen
AI Robotics and Humanoid Robots
No full textThis working paper analyzes the recent advances in AI-enabled robotics and humanoid robots, the economic and societal impact as well as the safety aspects of robots and their AI models. According to the definition of the International Standardization Organization ISO, a robot is a programmed actuated mechanism with a degree of autonomy to perform locomotion, manipulation, or positioning. The rise of advanced AI models in the 2020ies and the successful integration into robots has led to rise in AI-enabled robotics and contributed to the release of commercially available humanoid robots, i.e., general-purpose, bipedal robots modeled after the human form, which can now be used in real-world productions. More than 4 million industrial robots were already globally operating in 2024. The global industrial robotics market was close to USD 34 billion in 2024 with an expected annual growth rate of almost 10% until 2030. There are three different types of AI which contribute to robotics, the analytic, physical, and generative AI. The analytic AI can analyze large amount data such as sensor data and improves pattern recognition. The physical AI simulates real-world environments which allows the training of robots in realistic settings. The robots can train and adapt themselves (Self-learning) which overcomes the limitations of traditional programming. The generative AI can generate new content which is based on algorithms, training data e.g. from the internet and machine learning. The physical integration of AI into robots (embodied AI as AI inside a physical form) combines now the learning cycles of robots with the advances of AI which accelerates the progress of AI-enabled robotics. Humanoid robots are generally equipped with locomotion (gross motor skills), dexterity (fine motor skills), and intelligence provided by embodied AI. In future, robots may build chips, other robots, and entire productions, e.g., Foxconn and Nvidia intend to deploy humanoid robots at a Foxconn plant in Houston to produce Nvidia AI servers from 2026 on. The advances led to the first commercially available humanoid robots that can be used in industry, but also as service robots in transportation and logistics, but can also act in the service robot market in hospitality, healthcare (diagnosis, treatment, care), agriculture; there is also a growing number of non-humanoid cleaning robots. Experts expect an increase of humanoid robots from 13 million units in 2035 to 648 million units in 2050. In late 2024, over 50 types of humanoid robots were known already. This indicates that the AI transformation is no longer limited to specialist functions in the IT and finance sector. Safety issues are malfunctions and aggressive behavior of humanoid robots. The combination of AI with robots makes AI safety to a matter of robot safety as well. Attackers could gain control over robots by exploits of AI vulnerabilities which could be already demonstrated. New AI attack methods significantly facilitate cyber attacks and can bypass conventional cybersecurity measures, e.g., the use of hidden text which is not legible for humans, but legible for the AI. The coming ChatGPT5 is expected to be significantly faster and stronger. Furthermore, novel AI architectures may reduce the need for reasoning protocols for AI models which could result in loss of control for humans. In summary, with a rapidly advancing AI as platform technology and the growing practical experience with robots, a massive increase of AI-enabled robots including humanoid robots is expected in the coming years. The rapid advance of AI and robotics will amplify the economic benefits, but also safety risks and the societal impact
Interaction of Ultrashort Pulses with Polar Oxide Nanoparticles Studied by Nonlinear Optical Spectroscopy and Microscopy
Full text linkThe primary objective of this work is to explore the nonlinear optical phenomena
arising from the interaction between ultrashort pulses and polar oxide nanoparticles. The
research focuses on nonlinear optical spectroscopy and is complemented by applications
in nonlinear optical microscopy. Alongside the comprehensive characterization of the
nonlinear optical properties of various polar oxide materials using ultrashort pulses, this
work also investigates how these pulses interact with nanoparticle ensembles. By employing
time-resolved fs-pump-probe spectroscopy, the interaction is interrogated by the
nonlinear optical signal response, allowing for the deconvolution of the temporal evolution
of the pulses within the material. Systematic characterizations of the nonlinear optical
properties allow for the assessment of the application potential of these nanoparticles in
nonlinear optical microscopy. The concept of tunable high-energy widefield microscopy is
demonstrated, leveraging the interaction of high-energy ultrashort pulses with polar oxide
nanoparticles. This technique enables imaging under widefield illumination, utilizing the
nanoparticles as efficient, spectrally tunable nonlinear optical markers.
In addition, the work addresses the question of how fundamental optical properties and
effects known from bulk polar oxide materials can be studied in nanoparticle ensemble
measurements. To this end, existing methods for the spectroscopic characterization of
nanoparticles are conceptually extended, enabling the investigation of light-induced and
temperature-dependent absorption phenomena known from bulk materials herein explored
in nanocrystalline materials. This approach is especially valuable for studying materials
that lack a bulk equivalent, such as those synthesized through bottom-up methods