Search

With the increasing prevalence of security incidents, the adoption of deception-based defense strategies has become pivotal in cyber security. This work addresses the challenge of scalability in designing honeytokens, a key component of such defense mechanisms. The manual creation of honeytokens is a tedious task. Although automated generators exists, they often lack versatility, being specialized for specific types of honeytokens, and heavily rely on suitable training datasets. To overcome these limitations, this work systematically investigates the approach of utilizing Large Language Models (LLMs) to create a variety of honeytokens. Out of the seven different honeytoken types created in this work, such as configuration files, databases, and log files, two were used to evaluate the optimal prompt. The generation of robots.txt files and honeywords was used to systematically test 210 different prompt structures, based on 16 prompt building blocks. Furthermore, all honeytokens were tested across different state-of-the-art LLMs to assess the varying performance of different models. Prompts performing optimally on one LLMs do not necessarily generalize well to another. Honeywords generated by GPT-3.5 were found to be less distinguishable from real passwords compared to previous methods of automated honeyword generation. Overall, the findings of this work demonstrate that generic LLMs are capable of creating a wide array of honeytokens using the presented prompt structures., Comment: 12 pages

This paper investigates the shift in crowdsourcing towards self-managed enterprises of crowdworkers (SMECs), diverging from traditional platform-controlled models. It reviews the literature to understand the foundational aspects of this shift, focusing on identifying key factors that may explain the rise of SMECs, particularly concerning power dynamics and tensions between Online Labor Platforms (OLPs) and crowdworkers. The study aims to guide future research and inform policy and platform development, emphasizing the importance of fair labor practices in this evolving landscape., Comment: 12 pages, 5 figures, ICEIS 2024 - 2024 International Conference on Enterprise Information Systems

Widespread adoption of AI for medical decision making is still hindered due to ethical and safety-related concerns. For AI-based decision support systems in healthcare settings it is paramount to be reliable and trustworthy. Common deep learning approaches, however, have the tendency towards overconfidence under data shift. Such inappropriate extrapolation beyond evidence-based scenarios may have dire consequences. This highlights the importance of reliable estimation of local uncertainty and its communication to the end user. While stochastic neural networks have been heralded as a potential solution to these issues, this study investigates their actual reliability in clinical applications. We centered our analysis on the exemplary use case of mortality prediction for ICU hospitalizations using EHR from MIMIC3 study. For predictions on the EHR time series, Encoder-Only Transformer models were employed. Stochasticity of model functions was achieved by incorporating common methods such as Bayesian neural network layers and model ensembles. Our models achieve state of the art performance in terms of discrimination performance (AUC ROC: 0.868+-0.011, AUC PR: 0.554+-0.034) and calibration on the mortality prediction benchmark. However, epistemic uncertainty is critically underestimated by the selected stochastic deep learning methods. A heuristic proof for the responsible collapse of the posterior distribution is provided. Our findings reveal the inadequacy of commonly used stochastic deep learning approaches to reliably recognize OoD samples. In both methods, unsubstantiated model confidence is not prevented due to strongly biased functional posteriors, rendering them inappropriate for reliable clinical decision support. This highlights the need for approaches with more strictly enforced or inherent distance-awareness to known data points, e.g., using kernel-based techniques., Comment: Keywords: probabilistic inference, uncertainty estimation, uncertainty quantification, epistemic uncertainty, clinical prognosis, electronic health records

Cyber Physical Systems (CPS) enable new kinds of applications as well as significant improvements of existing ones in numerous different application domains. A major trait of upcoming CPS is an increasing degree of automation up to the point of autonomy, as there is a huge potential for economic success as well as for ecologic and societal improvements. However, to unlock the full potential of such (cooperative and automated) CPS, we first need to overcome several significant engineering challenges, where safety assurance is a particularly important one. Unfortunately, established safety assurance methods and standards do not live up to this task, as they have been designed with closed and less complex systems in mind. This paper structures safety assurance challenges of cooperative automated CPS, provides an overview on our vision of dynamic risk management and describes already existing building blocks.

The zeitgeist of the digital era has been dominated by an expanding integration of Artificial Intelligence~(AI) in a plethora of applications across various domains. With this expansion, however, questions of the safety and reliability of these methods come have become more relevant than ever. Consequently, a run-time ML model safety system has been developed to ensure the model's operation within the intended context, especially in applications whose environments are greatly variable such as Autonomous Vehicles~(AVs). SafeML is a model-agnostic approach for performing such monitoring, using distance measures based on statistical testing of the training and operational datasets; comparing them to a predetermined threshold, returning a binary value whether the model should be trusted in the context of the observed data or be deemed unreliable. Although a systematic framework exists for this approach, its performance is hindered by: (1) a dependency on a number of design parameters that directly affect the selection of a safety threshold and therefore likely affect its robustness, (2) an inherent assumption of certain distributions for the training and operational sets, as well as (3) a high computational complexity for relatively large sets. This work addresses these limitations by changing the binary decision to a continuous metric. Furthermore, all data distribution assumptions are made obsolete by implementing non-parametric approaches, and the computational speed increased by introducing a new distance measure based on the Empirical Characteristics Functions~(ECF).

This work discusses the origin of temperature rise during the collision welding process. The different physical irreversible and reversible mechanisms which act as heat sources are described: isentropic compression work, shock dissipation, plasticity, and phase transitions. The temperature increase due to these effects is quantified in a continuum mechanics approach, and compared to predictions of atomistic molecular dynamics simulations. Focusing on a single impact scenario of 1100 aluminium at 700 m/s, our results indicate that shock heating and plastic work only effect a temperature rise of 100 K, and that the effects of phase change are not significant. This temperature rise cannot explain welding. In consequence, the relevant mechanism which effects bonding in collision welding must be due to the jet, which is only formed at oblique impact angles.

Intensive research is carried out on two-dimensional materials, in particular molybdenum disulfide, towards high-performance transistors for integrated circuits. Fabricating transistors with ohmic contacts is challenging due to the high Schottky barrier that severely limits the transistors' performance. Graphene-based heterostructures can be used in addition or as a substitute for unsuitable metals. We present lateral heterostructure transistors made of scalable chemical vapor-deposited molybdenum disulfide and chemical vapor-deposited graphene with low contact resistances of about 9 k${\Omega}$${\mu}$m and high on/off current ratios of 10${^8}$. We also present a theoretical model calibrated on our experiments showing further potential for scaling transistors and contact areas into the few nanometers range and the possibility of a strong performance enhancement by means of layer optimizations that would make transistors promising for use in future logic circuits., Comment: 39 pages

In the field of network security, with the ongoing arms race between attackers, seeking new vulnerabilities to bypass defense mechanisms and defenders reinforcing their prevention, detection and response strategies, the novel concept of cyber deception has emerged. Starting from the well-known example of honeypots, many other deception strategies have been developed such as honeytokens and moving target defense, all sharing the objective of creating uncertainty for attackers and increasing the chance for the attacker of making mistakes. In this paper a methodology to evaluate the effectiveness of honeypots and moving target defense in a network is presented. This methodology allows to quantitatively measure the effectiveness in a simulation environment, allowing to make recommendations on how many honeypots to deploy and on how quickly network addresses have to be mutated to effectively disrupt an attack in multiple network and attacker configurations. With this optimum, attacks can be detected and slowed down with a minimal resource and configuration overhead. With the provided methodology, the optimal number of honeypots to be deployed and the optimal network address mutation interval can be determined. Furthermore, this work provides guidance on how to optimally deploy and configure them with respect to the attacker model and several network parameters.