004 Informatik
Refine
Document Type
- Conference proceeding (457)
- Journal article (116)
- Book chapter (26)
- Doctoral Thesis (12)
- Anthology (7)
- Book (6)
- Patent / Standard / Guidelines (1)
- Issue of a journal (1)
- Working Paper (1)
Is part of the Bibliography
- yes (627)
Institute
- Informatik (531)
- ESB Business School (62)
- Technik (35)
Publisher
- Springer (106)
- Hochschule Reutlingen (88)
- IEEE (86)
- Gesellschaft für Informatik e.V (64)
- Elsevier (52)
- Association for Computing Machinery (39)
- Association for Information Systems (22)
- IARIA (13)
- SciTePress (11)
- RWTH Aachen (10)
With the expansion of cyber-physical systems (CPSs) across critical and regulated industries, systems must be continuously updated to remain resilient. At the same time, they should be extremely secure and safe to operate and use. The DevOps approach caters to business demands of more speed and smartness in production, but it is extremely challenging to implement DevOps due to the complexity of critical CPSs and requirements from regulatory authorities. In this study, expert opinions from 33 European companies expose the gap in the current state of practice on DevOps-oriented continuous development and maintenance. The study contributes to research and practice by identifying a set of needs. Subsequently, the authors propose a novel approach called Secure DevOps and provide several avenues for further research and development in this area. The study shows that, because security is a cross-cutting property in complex CPSs, its proficient management requires system-wide competencies and capabilities across the CPSs development and operation.
This study is about estimating the reproducibility of finding palpation points of three different anatomical landmarks in the human body (Xiphoid Process and the 2 Hip Crests) to support a navigated ultrasound application. On 6 test subjects with different body mass index the three palpation points were located five times by two examiners. The deviation from the target position was calculated and correlated to the fat thickness above each palpation point. The reproducibility of the measurements had a mean error of ≈13.5 mm +- 4 mm, which seems to be sufficient for the desired application field.
Most Question-answering (QA) systems rely on training data to reach their optimal performance. However, acquiring training data for supervised systems is both time-consuming and resource-intensive. To address this, in this paper, we propose TFCSG, an unsupervised similar question retrieval approach that leverages pre-trained language models and multi-task learning. Firstly, topic keywords in question sentences are extracted sequentially based on a latent topic-filtering algorithm to construct unsupervised training corpus data. Then, the multi-task learning method is used to build the question retrieval model. There are three tasks designed. The first is a short sentence contrastive learning task. The second is the question sentence and its corresponding topic sequence similarity judgment task. The third is using question sentences to generate their corresponding topic sequence task. The three tasks are used to train the language model in parallel. Finally, similar questions are obtained by calculating the cosine similarity between sentence vectors. The comparison experiment on public question datasets that TFCSG outperforms the comparative unsupervised baseline method. And there is no need for manual marking, which greatly saves human resources.
While there has been increased digitization of private homes, only little has been done to understand these specific home technologies, how they serve consumers, among other issues. “Smart home technology” (SHT) refer to a wide range of artifacts from cleaning aids to energy advisors. Given this breadth, clarity surrounding the key characteristics and the multi-faceted impact of SHT is needed to conduct more directed research on SHT. We propose a taxonomy to help outline the salient intended outcomes of SHT. Through a process involving five iterations, we analyzed and classified 79 technologies (gathered from literature and industry reports). This uncovered seven dimensions encompassing 20 salient characteristics. We believe these dimensions/characteristics will help researchers and organizations better design and study the impacts of these technologies. Our long-term agenda is to use the proposed taxonomy for an exploratory inquiry to understand tensions occurring when personal and sustainability-related outcomes compete.
In times of climate change and growing urbanization, the way food is produced and consumed also changes. Meanwhile, digitization is transforming farming practices, which also applies to the domestic growing of crops. More and more so-called smart home farms (SHF) are finding their way into private households. This paper conceptualizes the unique nature of enabled smart services and their underlying technology. Following an inductive interpretive approach, this study explores the antecedents of smart home farming practices. Our sample consists of eleven actual smart home farmers. We found six constructs to be of salient importance: expected outcomes related to harvesting, positive feelings, and sustainability; a combination of one's affinity for green and novel technologies; and the smartness and visibility of the enabled services. In the outlook, we present some preliminary thoughts for testing our qualitative findings.
The early involvement of experiences gained through intelligence and data analysis is becoming increasingly important in order to develop new products, leading to a completely different conception of product creation, development and engineering processes using the advantages that the dedication of the digital twin entails. Introducing a novel stage gate process in order to be holistically anchored in learning factories adopting idea generation and idea screening in an early stage, beta testing of first prototypes, technical implementation in real production scenarios, business analysis, market evaluation, pricing, service models as well as innovative social media portals. Corresponding product modelling in the sense of sustainability, circular economy, and data analytics forecasts the product on the market both before and after market launch with the interlinking of data interpretation nearby in real-time. The digital twin represents the link between the digital model and the digital shadow. Additionally, the connection of the digital twin with the product provides constantly updated operating status and process data as well as mapping of technical properties and real-world behaviours. A future-networking product, by embedded information technology with the ability to initiate and carry out one's own further development, is able to interact with people and environments and thus is relevant to the way of life of future generations. In today's development work for this new product creation approach, on one hand, "Werk150" is the object of the development itself and on the other hand the validation environment. In the next step, new learning modules and scenarios for trainings at master level will be derived from these findings.
Das Ziel dieser Arbeit war die Umsetzung eines Wahrnehmungsensors für Softwareagenten, die über ein virtuelles Menschmodell in einer dreidimensionalen Umgebung agieren. Hierbei sollen die Agenten über den Sensor in der Lage sein, semantische Informationen zu geometrischen Objekten in der Umgebung zu erhalten. Hierfür wurden zwei Verfahren umgesetzt, die das menschliche Sehen simulieren, indem Objekte erkannt werden, wenn diese innerhalb eines Sichtfelds liegen. Ein Problem, das dabei gelöst werden muss, ist die Identifizierung möglicher Verdeckungen der Objekte. Ein Ansatz, dieses Problem zu lösen, ist der Ray-Tracing Ansatz, welcher für das erste Verfahren umgesetzt wurde. Das zweite Verfahren verwendet den Occlusion-Culling Ansatz. Auswertungen beider Verfahren haben gezeigt, dass der Ray-Tracing Ansatz eine schnellere Laufzeit aufweist, der Occlusion-Culling Ansatz jedoch mehr unverdeckte Objekte im Sichtfeld erkennt.
Nowadays, the importance of early active patient mobilization in the recovery and rehabilitation phase has increased significantly. One way to involve patients in the treatment is a gamification-like approach, which is one of the methods of motivation in various life processes. This article shows a system prototype for patients who require physical activity because of active early mobilization after medical interventions or during illness. Bedridden patients and people with a sedentary lifestyle (predominantly lying in bed) are also potential users. The main idea for the concept was non-contact system implementation for the patients making them feel effortless during its usage. The system consists of three related parts: hardware, software, and game application. To test the relevance and coherence of the system, it was used by 35 people. The participants were asked to play a video game requiring them to make body movements while lying down. Then they were asked to take part in a small survey to evaluate the system's usability. As a result, we offer a prototype consisting of hardware and software parts that can increase and diversify physical activity during active early mobilization of patients and prevent the occurrence of possible health problems due to predominantly low activity. The proposed design can be possibly implemented in hospitals, rehabilitation centers, and even at home.
Monitoring heart rate and breathing is essential in understanding the physiological processes for sleep analysis. Polysomnography (PSG) system have traditionally been used for sleep monitoring, but alternative methods can help to make sleep monitoring more portable in someone's home. This study conducted a series of experiments to investigate the use of pressure sensors placed under the bed as an alternative to PSG for monitoring heart rate and breathing during sleep. The following sets of experiments involved the addition of small rubber domes - transparent and black - that were glued to the pressure sensor. The resulting data were compared with the PSG system to determine the accuracy of the pressure sensor readings. The study found that the pressure sensor provided reliable data for extracting heart rate and respiration rate, with mean absolute errors (MAE) of 2.32 and 3.24 for respiration and heart rate, respectively. However, the addition of small rubber hemispheres did not significantly improve the accuracy of the readings, with MAEs of 2.3 bpm and 7.56 breaths per minute for respiration rate and heart rate, respectively. The findings of this study suggest that pressure sensors placed under the bed may serve as a viable alternative to traditional PSG systems for monitoring heart rate and breathing during sleep. These sensors provide a more comfortable and non-invasive method of sleep monitoring. However, the addition of small rubber domes did not significantly enhance the accuracy of the readings, indicating that it may not be a worthwhile addition to the pressure sensor system.
Sleep analysis using a Polysomnography system is difficult and expensive. That is why we suggest a non-invasive and unobtrusive measurement. Very few people want the cables or devices attached to their bodies during sleep. The proposed approach is to implement a monitoring system, so the subject is not bothered. As a result, the idea is a non-invasive monitoring system based on detecting pressure distribution. This system should be able to measure the pressure differences that occur during a single heartbeat and during breathing through the mattress. The system consists of two blocks signal acquisition and signal processing. This whole technology should be economical to be affordable enough for every user. As a result, preprocessed data is obtained for further detailed analysis using different filters for heartbeat and respiration detection. In the initial stage of filtration, Butterworth filters are used.
Theoretical foundation, effectiveness, and design artefact for machine learning service repositories
(2022)
Machine learning (ML) has played an important role in research in recent years. For companies that want to use ML, finding the algorithms and models that fit for their business is tedious. A review of the available literature on this problem indicates only a few research papers. Given this gap, the aim of this paper is to design an effective and easy-to-use ML service repository. The corresponding research is based on a multi-vocal literature analysis combined with design science research, addressing three research questions: (1) How is current white and gray literature on ML services structured with respect to repositories? (2) Which features are relevant for an effective ML service repository? (3) How is a prototype for an effective ML service repository conceptualized? Findings are relevant for the explanation of user acceptance of ML repositories. This is essential for corporate practice in order to create and use ML repositories effectively.
The tale of 1000 cores: an evaluation of concurrency control on real(ly) large multi-socket hardware
(2020)
In this paper, we set out the goal to revisit the results of “Starring into the Abyss [...] of Concurrency Control with [1000] Cores” and analyse in-memory DBMSs on today’s large hardware. Despite the original assumption of the authors, today we do not see single-socket CPUs with 1000 cores. Instead multi-socket hardware made its way into production data centres. Hence, we follow up on this prior work with an evaluation of the characteristics of concurrency control schemes on real production multi-socket hardware with 1568 cores. To our surprise, we made several interesting findings which we report on in this paper.
In this paper, we propose a radical new approach for scale-out distributed DBMSs. Instead of hard-baking an architectural model, such as a shared-nothing architecture, into the distributed DBMS design, we aim for a new class of so-called architecture-less DBMSs. The main idea is that an architecture-less DBMS can mimic any architecture on a per-query basis on-the-fly without any additional overhead for reconfiguration. Our initial results show that our architecture-less DBMS AnyDB can provide significant speedup across varying workloads compared to a traditional DBMS implementing a static architecture.
In our initial DaMoN paper, we set out the goal to revisit the results of “Starring into the Abyss [...] of Concurrency Control with [1000] Cores” (Yu in Proc. VLDB Endow 8: 209-220, 2014). Against their assumption, today we do not see single-socket CPUs with 1000 cores. Instead, multi-socket hardware is prevalent today and in fact offers over 1000 cores. Hence, we evaluated concurrency control (CC) schemes on a real (Intel-based) multi-socket platform. To our surprise, we made interesting findings opposing results of the original analysis that we discussed in our initial DaMoN paper. In this paper, we further broaden our analysis, detailing the effect of hardware and workload characteristics via additional real hardware platforms (IBM Power8 and 9) and the full TPC-C transaction mix. Among others, we identified clear connections between the performance of the CC schemes and hardware characteristics, especially concerning NUMA and CPU cache. Overall, we conclude that no CC scheme can efficiently make use of large multi-socket hardware in a robust manner and suggest several directions on how CC schemes and overall OLTP DBMS should evolve in future.
In this paper, we present a new approach for achieving robust performance of data structures making it easier to reuse the same design for different hardware generations but also for different workloads. To achieve robust performance, the main idea is to strictly separate the data structure design from the actual strategies to execute access operations and adjust the actual execution strategies by means of so-called configurations instead of hard-wiring the execution strategy into the data structure. In our evaluation we demonstrate the benefits of this configuration approach for individual data structures as well as complex OLTP workloads.
Forecasting demand is challenging. Various products exhibit different demand patterns. While demand may be constant and regular for one product, it may be sporadic for another, as well as when demand occurs, it may fluctuate significantly. Forecasting errors are costly and result in obsolete inventory or unsatisfied demand. Methods from statistics, machine learning, and deep learning have been used to predict such demand patterns. Nevertheless, it is not clear for what demand pattern, which algorithm would achieve the best forecast. Therefore, even today a large number of models are used to forecast on a test period. The model with the best result on the test period is used for the actual forecast. This approach is computationally and time intensive and, in most cases, uneconomical. In our paper we show the possibility to use a machine learning classification algorithm, which predicts the best possible model based on the characteristics of a time series. The approach was developed and evaluated on a dataset from a B2B-technical-retailer. The machine learning classification algorithm achieves a mean ROC-AUC of 89%, which emphasizes the skill of the model.
Machine learning (ML) techniques are rapidly evolving, both in academia and practice. However, enterprises show different maturity levels in successfully implementing ML techniques. Thus, we review the state of adoption of ML in enterprises. We find that ML technologies are being increasingly adopted in enterprises, but that small and medium-size enterprises (SME) are struggling with the introduction in comparison to larger enterprises. In order to identify enablers and success factors we conduct a qualitative empirical study with 18 companies in different industries. The results show that especially SME fail to apply ML technologies due to insufficient ML knowhow. However, partners and appropriate tools can compensate this lack of resources. We discuss approaches to bridge the gap for SME.
Im Rahmen der wissenschaftlichen Vertiefung an der Hochschule Reutlingen befasst sich diese Arbeit mit der Untersuchung der Anforderungen und der Machbarkeit zur computergestützten Erkennung der Deutschen Gebärdensprache (DGS) und des deutschen Fingeralphabets. Die Erkenntnisse aus dieser Arbeit dienen als Grundlage zur Entwicklung eines Systems zur Übersetzung von Gebärden der DGS oder des Fingeralphabets in die deutsche Schriftsprache. Zunächst werden grundlegende Informationen zu Geschichte, Aufbau und Grammatik der DGS und des Fingeralphabets aufgeführt. Die Erkennung der Gebärden soll durch optische Bewegungssensoren erfolgen. Hierfür werden unterschieliche Sensortypen betrachtet und verglichen. Im weiteren Verlauf erfolgt die Analyse der benutzerspezifischen und technischen Anforderungen. Erstere basieren auf der Befragung einer Fokusgruppe aus gehörlosen und hörenden Menschen aus dem Bereich der Gehörlosen-, Schwerhörigen- und Sprachbehindertenpädagogik. Abgeleitet aus den Informationen der Anforderungsanalyse ergibt sich, bis zu einem gewissen Grad, die Machbarkeit aus technischer und benutzerspezifischer Sicht. Abschließend erfolgen die Zusammenfassung der Anforderungen, welche an das zu entwickelnde System gestllt werden, sowie eine Handlungsempfehlung für die Entwicklung eines Prototyps.
The basic idea behind a wearable robotic grasp assistancesystem is to support people that suffer from severe motor impairments in daily activities. Such a system needs to act mostly autonomously and according to the user’s intent. Vision-based hand pose estimation could be an integral part of a larger control and assistance framework. In this paper we evaluate the performance of egocentric monocular hand pose estimation for a robot-controlled hand exoskeleton in a simulation. For hand pose estimation we adopt a Convolutional Neural Network (CNN). We train and evaluate this network with computer graphics, created by our own data generator. In order to guide further design decisions we focus in our experiments on two egocentric camera viewpoints tested on synthetic data with the help of a 3D-scanned hand model, with and without an exoskeleton attached to it.We observe that hand pose estimation with a wrist-mounted camera performs more accurate than with a head-mounted camera in the context of our simulation. Further, a grasp assistance system attached to the hand alters visual appearance and can improve hand pose estimation. Our experiment provides useful insights for the integration of sensors into a context sensitive analysis framework for intelligent assistance.
Additive Manufacturing is increasingly used in the industrial sector as a result of continuous development. In the Production Planning and Control (PPC) system, AM enables an agile response in the area of detailed and process planning, especially for a large number of plants. For this purpose, a concept for a PPC system for AM is presented, which takes into account the requirements for integration into the operational enterprise software system. The technical applicability will be demonstrated by individual implemented sections. The presented solution approach promises a more efficient utilization of the plants and a more elastic use.