3633 research outputs found
Sort by
DANSEN: Database acceleration on native computational storage by exploiting NDP
No full textThis paper introduces DANSEN, the hardware accelerator component for neoDBMS, a full-stack computational storage system designed to manage on-device execution of database queries/transactions as a Near-Data Processing (NDP)-operation. The proposed system enables Database Management Systems (DBMS) to oload NDP-operations to the storage while maintaining control over data through a native storage interface. DANSEN provides an NDP-engine that enables DBMS to perform both low-level database tasks, such as performing database administration, as well as high-level tasks like executing SQL, on the smart storage device while observing the DBMS concurrency control. Furthermore, DANSEN enables the incorporation of custom accelerators as an NDP-operation, e.g., to perform hardware-accelerated ML inference directly on the stored data. We built the DANSEN storage prototype and interface on an Ultrascale+HBM FPGA and fully integrated it with PostgreSQL 12. Experimental results demonstrate that the proposed NDP approach outperforms software-only PostgreSQL using a fast of-the-shelf NVMe drive, and signiicantly improves the end-to-end execution time of an aggregation operation (similar to Q6 from CH-benCHmark, 150 million records) by ≈10.6×. The versatility of the proposed approach is also validated by integrating a compute-intensive data analytics application with multi-row results, outperforming PostgreSQL by ≈1.5×
Data analysis of non-invasive ballistocardiographic sensors
No full textUnintrusive health monitoring systems is important when continuous monitoring of the patient vital signals is required. In this paper, signals obtained from accelerometers placed under a bed are processed with ballistocardiography algorithms and compared with synchronized electrocardiographic signals
Gewinnung professoralen Personals durch datengetriebene Personas
No full textThis paper explores the application of People Analytics in
recruiting professors for universities of applied sciences. Using data-driven personas, the research project aims to identify and communicate the different paths and connections leading candidates to a professorship. The authors introduce the concept of personas, describe the underlying data source and derive an example for the current project
Sustainable innovations, knowledge and the role of proximity: A systematic literature review
Get PDFInnovations can substantially contribute to the transformation toward sustainability if they induce a positive social and/or environmental impact. Such sustainable innovations differ considerably from conventional, purely economic innovations. The main difference stems from the different knowledge bases necessary for the development of these innovations. These knowledge bases are widely dispersed across different actors from business, academia, government, and civil society. Following the innovation system approach, we look at actor constellations, linkages between actors, and knowledge flows within networks that generate sustainable innovations. For this purpose, we conduct a systematic literature review, focusing on the concept of proximity and its five dimensions (geographical, cognitive, institutional, organizational, and social proximity). The results show that all proximity dimensions, as well as the interdependencies between them, are relevant for analyzing knowledge flows leading to sustainable innovations. The interplay of the different proximity dimensions can be described via two mechanisms, one being reinforcement and the other one being either substitution or overlap. We conclude that for the occurrence of radical, systemic innovations, which have the potential of altering the prevailing socio‐economic paradigm toward greater sustainability, a combination of low cognitive and low (micro‐) institutional proximity combined with high organizational, social, or geographical proximity, appears particularly conducive
Zero-shot strike: testing the generalisation capabilities of out-of-the-box LLM models for depression detection
No full textDepression is a significant global health challenge. Still, many people suffering from depression remain undiagnosed. Furthermore, the assessment of depression can be subject to human bias. Natural Language Processing (NLP) models offer a promising solution. We investigated the potential of four NLP models (BERT, Llama2-13B, GPT-3.5, and GPT-4) for depression detection in clinical interviews. Participants (N = 82) underwent clinical interviews and completed a self-report depression questionnaire. NLP models inferred depression scores from interview transcripts. Questionnaire cut-off values for depression were used as a classifier for depression. GPT-4 showed the highest accuracy for depression classification (F1 score 0.73), while zero-shot GPT-3.5 initially performed with low accuracy (0.34), improved to 0.82 after fine-tuning, and achieved 0.68 with clustered data. GPT-4 estimates of symptom severity PHQ-8 score correlated strongly (r = 0.71) with true symptom severity. These findings demonstrate the potential of AI models for depression detection. However, further research is necessary before widespread deployment can be considered
Exploring machine learning diagnostic decisions based on wideband immittance measurements for otosclerosis and disarticulation
Get PDFBackground: Wideband acoustic immittance (WAI) and wideband tympanometry (WBT) are promising approaches to improve diagnosis accuracy in middle-ear diagnosis, though due to significant interindividual difference, their analysis and interpretation remains challenging. Recent approaches have come up, implementing machine learning (ML) or deep learning classifiers trained with measured WAI or WBT data for the classification of otitis media or otosclerosis. Also, first approaches have been made in identifying important regions from the WBT data, which the classifiers used for their decision-making.
Methods: Two classifiers, a convolutional neural network (CNN) and the ML algorithm extreme gradient boosting (XGB), are trained on artificial data obtained with a finite-element ear model providing the middle-ear measurements energy reflectance (ER), pressure reflectance phase, impedance amplitude and phase. The performance of both classifiers is evaluated by cross-validation on artificial test data and by classification of real measurement data from the literature using the metrics macro-recall and macro-F1 score. The feature contributions are quantified using the feature importance ‘gain’ for XGB and deep Taylor decomposition for CNN.
Results: In the cross-validation with artificial data, the macro-recall and macro-F1 scores are similar, namely 91.2% for XGB and 94.5% for CNN. For the classification with real measurement data the macro-recall and macro-F1-score were 81.8% and 38.2% (XGB) and 81.0% and 54.8% (CNN), respectively. The key features identified are ER between 600–1,000 Hz together with impedance phase between 600–1,000 Hz for XGB and ER up to 1,500 Hz for CNN.
Conclusions: We were able to show that the applied classifiers CNN and XGB trained with simulated data lead to a reasonably well performance on real data. We conclude that using simulation-based WAI data can be a successful strategy for classifier training and that XGB can be applied to WAI data. Furthermore, ML interpretability algorithms are useful to identify relevant key features for differential diagnosis and to increase confidence in classifier decisions. Further evaluation using more measured data, especially for pathological cases, is essential
Digitaler Zwilling - Technologie, Maschine, Prozess - Wo hakt es derzeit noch in der flexiblen Fertigung?
No full textDie digitale Abbildung von Maschinen und Prozessen zur Entwicklung, Auslegung, Simulation und Optimierung schreitet immer weiter voran. Steigende Leistungsfähigkeit von Rechnerkapazitäten und der Einsatz lernender Algorithmen erlaubt die immer exaktere Wiedergabe der einzelnen Komponenten, Maschinen und Abläufen. Aufgrund der Komplexität von Werkzeugmaschinen besteht aber dennoch erheblicher Bedarf zur Integration zusätzlichen Wissens in die bei der Erstellung von Digitalen Zwillingen verwendeten Daten
Textiles on the path to sustainability and circularity: results of application tests in the business-to-business sector
Get PDFThe textile sector is responsible for a number of environmental impacts, e.g., climate change, and is not pursuing sustainable production and consumption patterns. Due to the increasing quantities of textiles, their share is rising, and a trend reversal from a linear to a circular and sustainable textile chain is needed. This article presents the background, methodological approach and results of a participatory textile development model. In the commercial B2B sector, three textile prototypes were developed together with users and trialled over several months in three application areas. Textile development took into account the requirements of fibre regeneration in the product design and focused on innovative more sustainable chemical recycling solutions. The three sustainably aligned textiles were subjected to spectroscopic and textile–technological tests. The sustainability tool screening life cycle assessments analysed their environmental profile and compared it with reference textiles that are used as the standard. Overall, it is clear that the three textiles can match conventional reference textiles in terms of quality and have considerable environmental benefits compared to the reference textiles. The user survey did identify concerns about a high artificial fibre content, although a general rejection of recycled fibres was not observed. The results show that a sustainable transformation is possible but must start with the fibre composition; recycling, on the other hand, is of minor importance
Speech recognition for the sterile interaction with information systems in the surgical area : towards an optimized interaction paradigm with HIS data on large displays in the sterile field
Get PDFThis paper describes how automatic speech recognition (ASR) can be used to explore electronic patient records (EPR) in a sterile environment. As an information display, we used a system that was developed for the presentation of patient data and for supporting surgical hand disinfection. The speech recognition was performed using the Mozilla Web Speech API. A room microphone, a Jabra Box, and a directional microphone, a RØDE NT-USB Mini, were tested at various positions. Interactions with the EPR are triggered with a list of commands. The interaction was working as indented and test persons reported the system as intuitive
Towards examiner-independent reproducible abdominal ultrasound examinations : A system approach targeting changing examination settings like in ambulatory scenarios
Get PDFAbstract
The proposed system aims at reproducibility in ultrasound follow-up examinations by storing the position of the ultrasound transducer in relation to acquired images. The system utilizes an electromagnetic tracking system. It includes a guidance feature to assist in placing the transducer correctly on the patient. Evaluation of the system involved technical accuracy tests on a phantom and accuracy tests on human subjects. The results showed that the technical accuracy of the system met the required criteria, but the transducer positioning error in realistic scenarios was above the threshold. Usability testing indicated potential benefits for medical training and provided suggestions for improving the user interface. Overall, the system shows promise for further development and can already be used for training ultrasound examinations