Daniel Schneider

Al-Harith Farhad, Ioannis Sorokos, Andreas Schmidt et al.

Machine Learning~(ML) has provided promising results in recent years across different applications and domains. However, in many cases, qualities such as reliability or even safety need to be ensured. To this end, one important aspect is to determine whether or not ML components are deployed in situations that are appropriate for their application scope. For components whose environments are open and variable, for instance those found in autonomous vehicles, it is therefore important to monitor their operational situation to determine its distance from the ML components' trained scope. If that distance is deemed too great, the application may choose to consider the ML component outcome unreliable and switch to alternatives, e.g. using human operator input instead. SafeML is a model-agnostic approach for performing such monitoring, using distance measures based on statistical testing of the training and operational datasets. Limitations in setting SafeML up properly include the lack of a systematic approach for determining, for a given application, how many operational samples are needed to yield reliable distance information as well as to determine an appropriate distance threshold. In this work, we address these limitations by providing a practical approach and demonstrate its use in a well known traffic sign recognition problem, and on an example using the CARLA open-source automotive simulator.

Simon Daniel Duque Anton, Daniel Schneider, Hans Dieter Schotten

Artificial Intelligence (AI) increasingly shows its potential to outperform predicate logic algorithms and human control alike. In automatically deriving a system model, AI algorithms learn relations in data that are not detectable for humans. This great strength, however, also makes use of AI methods dubious. The more complex a model, the more difficult it is for a human to understand the reasoning for the decisions. As currently, fully automated AI algorithms are sparse, every algorithm has to provide a reasoning for human operators. For data engineers, metrics such as accuracy and sensitivity are sufficient. However, if models are interacting with non-experts, explanations have to be understandable. This work provides an extensive survey of literature on this topic, which, to a large part, consists of other surveys. The findings are mapped to ways of explaining decisions and reasons for explaining decisions. It shows that the heterogeneity of reasons and methods of and for explainability lead to individual explanatory frameworks.

Marcel Weichel, Martin Reder, Gerit Mühlberg et al.

Sodium-ion batteries employing hard carbon electrodes are considered a drop-in technology for lithium-ion batteries. Electrode drying is a critical manufacturing step, as binder migration during pore emptying impacts the mechanical integrity and electrical performance of the electrode. Existing modeling approaches predominantly rely on the film shrinkage phase in a one dimensional way or neglect the capillary transport, resulting in a lack of physically consistent microstructure resolved predictions of binder migration. In this work, a spatially resolved pore scale continuum model is extended to explicitly describe capillary driven binder transport during pore emptying. The model is applied to hard carbon microstructures with varying mean particle diameters. The simulations reveal that smaller particle sizes lead to a more homogeneous binder distribution, whereas higher evaporation rates and increased surface tension promote stronger binder gradients. Variations in solvent viscosity show only a minor influence on binder migration, as long as no hydrophilic or hydrophobic behavior is present. Finally, the simulations demonstrate that an explicit description of capillary transport and microstructural effects is essential for accurately predicting binder migration and provides a basis for the targeted optimization of electrode drying processes.

Mohammed Naveed Akram, Akshatha Ambekar, Ioannis Sorokos et al.

Reliability estimation of Machine Learning (ML) models is becoming a crucial subject. This is particularly the case when such \mbox{models} are deployed in safety-critical applications, as the decisions based on model predictions can result in hazardous situations. In this regard, recent research has proposed methods to achieve safe, \mbox{dependable}, and reliable ML systems. One such method consists of detecting and analyzing distributional shift, and then measuring how such systems respond to these shifts. This was proposed in earlier work in SafeML. This work focuses on the use of SafeML for time series data, and on reliability and robustness estimation of ML-forecasting methods using statistical distance measures. To this end, distance measures based on the Empirical Cumulative Distribution Function (ECDF) proposed in SafeML are explored to measure Statistical-Distance Dissimilarity (SDD) across time series. We then propose SDD-based Reliability Estimate (StaDRe) and SDD-based Robustness (StaDRo) measures. With the help of a clustering technique, the similarity between the statistical properties of data seen during training and the forecasts is identified. The proposed method is capable of providing a link between dataset SDD and Key Performance Indicators (KPIs) of the ML models.

Al-Harith Farhad, Ioannis Sorokos, Mohammed Naveed Akram et al.

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).

Raia Abu Ahmad, Rana Abdulla, Tilahun Abedissa Taffa et al.

Shared tasks are powerful tools for advancing research through community-based standardised evaluation. As such, they play a key role in promoting findable, accessible, interoperable, and reusable (FAIR), as well as transparent and reproducible research practices. This paper presents an updated overview of twelve shared tasks developed and hosted under the German National Research Data Infrastructure for Data Science and Artificial Intelligence (NFDI4DS) consortium, covering a diverse set of challenges in scholarly document processing. Hosted at leading venues, the tasks foster methodological innovations and contribute open-access datasets, models, and tools for the broader research community, which are integrated into the consortium's research data infrastructure.

Adrian Lindenmeyer, Malte Blattmann, Stefan Franke et al.

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.

Verena Biener, Travis Gesslein, Daniel Schneider et al.

Virtual Reality (VR) has the potential to support mobile knowledge workers by complementing traditional input devices with a large three-dimensional output space and spatial input. Previous research on supporting VR knowledge work explored domains such as text entry using physical keyboards and spreadsheet interaction using combined pen and touch input. Inspired by such work, this paper probes the VR design space for authoring presentations in mobile settings. We propose PoVRPoint -- a set of tools coupling pen- and touch-based editing of presentations on mobile devices, such as tablets, with the interaction capabilities afforded by VR. We study the utility of extended display space to, for example, assist users in identifying target slides, supporting spatial manipulation of objects on a slide, creating animations, and facilitating arrangements of multiple, possibly occluded, shapes. Among other things, our results indicate that 1) the wide field of view afforded by VR results in significantly faster target slide identification times compared to a tablet-only interface for visually salient targets; and 2) the three-dimensional view in VR enables significantly faster object reordering in the presence of occlusion compared to two baseline interfaces. A user study further confirmed that the interaction techniques were found to be usable and enjoyable.

Simon Daniel Duque Anton, Daniel Fraunholz, Daniel Krohmer et al.

Operational Technology (OT)-networks and -devices, i.e. all components used in industrial environments, were not designed with security in mind. Efficiency and ease of use were the most important design characteristics. However, due to the digitisation of industry, an increasing number of devices and industrial networks is opened up to public networks. This is beneficial for administration and organisation of the industrial environments. However, it also increases the attack surface, providing possible points of entry for an attacker. Originally, breaking into production networks meant to break an Information Technology (IT)-perimeter first, such as a public website, and then to move laterally to Industrial Control Systems (ICSs) to influence the production environment. However, many OT-devices are connected directly to the Internet, which drastically increases the threat of compromise, especially since OT-devices contain several vulnerabilities. In this work, the presence of OT-devices in the Internet is analysed from an attacker's perspective. Publicly available tools, such as the search engine Shodan and vulnerability databases, are employed to find commonly used OT-devices and map vulnerabilities to them. These findings are grouped according to country of origin, manufacturer, and number as well as severity of vulnerability. More than 13000 devices were found, almost all contained at least one vulnerability. European and Northern American countries are by far the most affected ones.

Mahdi Razzaghpour, Adwait Datar, Daniel Schneider et al.

Cooperative driving systems, such as platooning, rely on communication and information exchange to create situational awareness for each agent. Design and performance of control components are therefore tightly coupled with communication component performance. The information flow between vehicles can significantly affect the dynamics of a platoon. Therefore, both the performance and the stability of a platoon depend not only on the vehicle's controller but also on the information flow Topology (IFT). The IFT can cause limitations for certain platoon properties, i.e., stability and scalability. Cellular Vehicle-To-Everything (C-V2X) has emerged as one of the main communication technologies to support connected and automated vehicle applications. As a result of packet loss, wireless channels create random link interruption and changes in network topologies. In this paper, we model the communication links between vehicles with a first-order Markov model to capture the prevalent time correlations for each link. These models enable performance evaluation through better approximation of communication links during system design stages. Our approach is to use data from experiments to model the Inter-Packet Gap (IPG) using Markov chains and derive transition probability matrices for consecutive IPG states. Training data is collected from high fidelity simulations using models derived based on empirical data for a variety of different vehicle densities and communication rates. Utilizing the IPG models, we analyze the mean-square stability of a platoon of vehicles with the standard consensus protocol tuned for ideal communication and compare the degradation in performance for different scenarios.

Daniel Schneider, Verena Biener, Alexander Otte et al.

An increasing number of consumer-oriented head-mounted displays (HMD) for augmented and virtual reality (AR/VR) are capable of finger and hand tracking. We report on the accuracy of off-the-shelf VR and AR HMDs when used for touch-based tasks such as pointing or drawing. Specifically, we report on the finger tracking accuracy of the VR head-mounted displays Oculus Quest, Vive Pro and the Leap Motion controller, when attached to a VR HMD, as well as the finger tracking accuracy of the AR head-mounted displays Microsoft HoloLens 2 and Magic Leap. We present the results of two experiments in which we compare the accuracy for absolute and relative pointing tasks using both human participants and a robot. The results suggest that HTC Vive has a lower spatial accuracy than the Oculus Quest and Leap Motion and that the Microsoft HoloLens 2 provides higher spatial accuracy than Magic Leap One. These findings can serve as decision support for researchers and practitioners in choosing which systems to use in the future.

Erik Armengaud, Georg Macher, Alexander Massoner et al.

The open and cooperative nature of Cyber-Physical Systems (CPS) poses new challenges in assuring dependability. The DEIS project (Dependability Engineering Innovation for automotive CPS. This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 732242, see http://www.deis-project.eu) addresses these challenges by developing technologies that form a science of dependable system integration. In the core of these technologies lies the concept of a Digital Dependability Identity (DDI) of a component or system. DDIs are modular, composable, and executable in the field facilitating (a) efficient synthesis of component and system dependability information over the supply chain and (b) effective evaluation of this information in-the-field for safe and secure composition of highly distributed and autonomous CPS. The paper outlines the DDI concept and opportunities for application in four industrial use cases.

Daniel Reti, Daniel Fraunholz, Janis Zemitis et al.

Deception technology has proven to be a sound approach against threats to information systems. Aside from well-established honeypots, decoy elements, also known as honeytokens, are an excellent method to address various types of threats. Decoy elements are causing distraction and uncertainty to an attacker and help detecting malicious activity. Deception is meant to be complementing firewalls and intrusion detection systems. Particularly insider threats may be mitigated with deception methods. While current approaches consider the use of multiple decoy elements as well as context-sensitivity, they do not sufficiently describe a relationship between individual elements. In this work, inter-referencing decoy elements are introduced as a plausible extension to existing deception frameworks, leading attackers along a path of decoy elements. A theoretical foundation is introduced, as well as a stochastic model and a reference implementation. It was found that the proposed system is suitable to enhance current decoy frameworks by adding a further dimension of inter-connectivity and therefore improve intrusion detection and prevention.

Daniel Schneider, Daniel Fraunholz, Daniel Krohmer

Honeypots are a well-studied defensive measure in network security. This work proposes an effective low-cost honeypot that is easy to deploy and maintain. The honeypot introduced in this work is able to handle commands in a non-standard way by blocking them or replying with an insult to the attacker. To determine the most efficient defense strategy, the interaction between attacker and defender is modeled as a Bayesian two-player game. For the empirical analysis, three honeypot instances were deployed, each with a slight variation in its configuration. In total, over 200 distinct sessions were captured, which allows for qualitative evaluation of post-exploitation behavior. The findings show that attackers react to insults and blocked commands in different ways, ranging from ignoring to sending insults themselves. The main contribution of this work lies in the proposed framework, which offers a low-cost alternative to more technically sophisticated and resource-intensive approaches.

Simon D Duque Anton, Daniel Fraunholz, Daniel Schneider

The internet landscape is growing and at the same time becoming more heterogeneous. Services are performed via computers and networks, critical data is stored digitally. This enables freedom for the user, and flexibility for operators. Data is easier to manage and distribute. However, every device connected to a network is potentially susceptible to cyber attacks. Security solutions, such as antivirus software or firewalls, are widely established. However, certain types of attacks cannot be prevented with defensive measures alone. Offensive security describes the practice of security professionals using methods and tools of cyber criminals. This allows them to find vulnerabilities before they become the point of entry in a real attack. Furthermore, following the methods of cyber criminals enables security professionals to adapt to a criminal's point of view and potentially discover attack angles formerly ignored. As cyber criminals often employ freely available security tools, having knowledge about these provides additional insight for professionals. This work categorises and compares tools regarding metrics concerning maintainability, usability and technical details. Generally, several well-established tools are available for the first phases, while phases after the initial breach lack a variety of tools.

Verena Biener, Daniel Schneider, Travis Gesslein et al.

Virtual Reality (VR) has the potential to transform knowledge work. One advantage of VR knowledge work is that it allows extending 2D displays into the third dimension, enabling new operations, such as selecting overlapping objects or displaying additional layers of information. On the other hand, mobile knowledge workers often work on established mobile devices, such as tablets, limiting interaction with those devices to a small input space. This challenge of a constrained input space is intensified in situations when VR knowledge work is situated in cramped environments, such as airplanes and touchdown spaces. In this paper, we investigate the feasibility of interacting jointly between an immersive VR head-mounted display and a tablet within the context of knowledge work. Specifically, we 1) design, implement and study how to interact with information that reaches beyond a single physical touchscreen in VR; 2) design and evaluate a set of interaction concepts; and 3) build example applications and gather user feedback on those applications.

Travis Gesslein, Verena Biener, Philipp Gagel et al.

Virtual Reality (VR) can enhance the display and interaction of mobile knowledge work and in particular, spreadsheet applications. While spreadsheets are widely used yet are challenging to interact with, especially on mobile devices, using them in VR has not been explored in depth. A special uniqueness of the domain is the contrast between the immersive and large display space afforded by VR, contrasted by the very limited interaction space that may be afforded for the information worker on the go, such as an airplane seat or a small work-space. To close this gap, we present a tool-set for enhancing spreadsheet interaction on tablets using immersive VR headsets and pen-based input. This combination opens up many possibilities for enhancing the productivity for spreadsheet interaction. We propose to use the space around and in front of the tablet for enhanced visualization of spreadsheet data and meta-data. For example, extending sheet display beyond the bounds of the physical screen, or easier debugging by uncovering hidden dependencies between sheet's cells. Combining the precise on-screen input of a pen with spatial sensing around the tablet, we propose tools for the efficient creation and editing of spreadsheets functions such as off-the-screen layered menus, visualization of sheets dependencies, and gaze-and-touch-based switching between spreadsheet tabs. We study the feasibility of the proposed tool-set using a video-based online survey and an expert-based assessment of indicative human performance potential.

Daniel Schneider, Alexander Otte, Travis Gesslein et al.

Physical keyboards are common peripherals for personal computers and are efficient standard text entry devices. Recent research has investigated how physical keyboards can be used in immersive head-mounted display-based Virtual Reality (VR). So far, the physical layout of keyboards has typically been transplanted into VR for replicating typing experiences in a standard desktop environment. In this paper, we explore how to fully leverage the immersiveness of VR to change the input and output characteristics of physical keyboard interaction within a VR environment. This allows individual physical keys to be reconfigured to the same or different actions and visual output to be distributed in various ways across the VR representation of the keyboard. We explore a set of input and output mappings for reconfiguring the virtual presentation of physical keyboards and probe the resulting design space by specifically designing, implementing and evaluating nine VR-relevant applications: emojis, languages and special characters, application shortcuts, virtual text processing macros, a window manager, a photo browser, a whack-a-mole game, secure password entry and a virtual touch bar. We investigate the feasibility of the applications in a user study with 20 participants and find that, among other things, they are usable in VR. We discuss the limitations and possibilities of remapping the input and output characteristics of physical keyboards in VR based on empirical findings and analysis and suggest future research directions in this area.

Philipp Oettershagen, Florian Achermann, Benjamin Müller et al.

Today, low-altitude fixed-wing Unmanned Aerial Vehicles (UAVs) are largely limited to primitively follow user-defined waypoints. To allow fully-autonomous remote missions in complex environments, real-time environment-aware navigation is required both with respect to terrain and strong wind drafts. This paper presents two relevant initial contributions: First, the literature's first-ever 3D wind field prediction method which can run in real time onboard a UAV is presented. The approach retrieves low-resolution global weather data, and uses potential flow theory to adjust the wind field such that terrain boundaries, mass conservation, and the atmospheric stratification are observed. A comparison with 1D LIDAR data shows an overall wind error reduction of 23% with respect to the zero-wind assumption that is mostly used for UAV path planning today. However, given that the vertical winds are not resolved accurately enough further research is required and identified. Second, a sampling-based path planner that considers the aircraft dynamics in non-uniform wind iteratively via Dubins airplane paths is presented. Performance optimizations, e.g. obstacle-aware sampling and fast 2.5D-map collision checks, render the planner 50% faster than the Open Motion Planning Library (OMPL) implementation. Test cases in Alpine terrain show that the wind-aware planning performs up to 50x less iterations than shortest-path planning and is thus slower in low winds, but that it tends to deliver lower-cost paths in stronger winds. More importantly, in contrast to the shortest-path planner, it always delivers collision-free paths. Overall, our initial research demonstrates the feasibility of 3D wind field prediction from a UAV and the advantages of wind-aware planning. This paves the way for follow-up research on fully-autonomous environment-aware navigation of UAVs in real-life missions and complex terrain.

Daniel Schneider, Jens Grubert

Around-device interaction techniques aim at extending the input space using various sensing modalities on mobile and wearable devices. In this paper, we present our work towards extending the input area of mobile devices using front-facing device-centered cameras that capture reflections in the human eye. As current generation mobile devices lack high resolution front-facing cameras we study the feasibility of around-device interaction using corneal reflective imaging based on a high resolution camera. We present a workflow, a technical prototype and an evaluation, including a migration path from high resolution to low resolution imagers. Our study indicates, that under optimal conditions a spatial sensing resolution of 5 cm in the vicinity of a mobile phone is possible.

Daniel Schneider, Jens Grubert

Smartphones are a popular device class for mobile Augmented Reality but suffer from a limited input space. Around-device interaction techniques aim at extending this input space using various sensing modalities. In this paper we present our work towards extending the input area of mobile devices using front-facing device-centered cameras that capture reflections in the cornea. As current generation mobile devices lack high resolution front-facing cameras, we study the feasibility of around-device interaction using corneal reflective imaging based on a high resolution camera. We present a workflow, a technical prototype and a feasibility evaluation.

Betty Cheng, Kerstin Eder, Martin Gogolla et al.

A self-adaptive software system modifies its behavior at runtime in response to changes within the system or in its execution environment. The fulfillment of the system requirements needs to be guaranteed even in the presence of adverse conditions and adaptations. Thus, a key challenge for self-adaptive software systems is assurance. Traditionally, confidence in the correctness of a system is gained through a variety of activities and processes performed at development time, such as design analysis and testing. In the presence of selfadaptation, however, some of the assurance tasks may need to be performed at runtime. This need calls for the development of techniques that enable continuous assurance throughout the software life cycle. Fundamental to the development of runtime assurance techniques is research into the use of models at runtime (M@RT). This chapter explores the state of the art for usingM@RT to address the assurance of self-adaptive software systems. It defines what information can be captured by M@RT, specifically for the purpose of assurance, and puts this definition into the context of existing work. We then outline key research challenges for assurance at runtime and characterize assurance methods. The chapter concludes with an exploration of selected application areas where M@RT could provide significant benefits beyond existing assurance techniques for adaptive systems.