Refine
Year of publication
Document Type
- Conference Proceeding (305) (remove)
Conference Type
- Konferenzartikel (185)
- Konferenz-Abstract (82)
- Sonstiges (25)
- Konferenz-Poster (13)
- Konferenzband (3)
Language
- English (305) (remove)
Has Fulltext
- no (305) (remove)
Is part of the Bibliography
- yes (305) (remove)
Keywords
- RoboCup (32)
- injury (9)
- Machine Learning (8)
- Biomechanik (6)
- Ausbildung (5)
- biomechanics (5)
- running (5)
- ACL (4)
- Additive Manufacturing (4)
- Deep Leaning (4)
Institute
- Fakultät Elektrotechnik und Informationstechnik (E+I) (bis 03/2019) (96)
- Fakultät Maschinenbau und Verfahrenstechnik (M+V) (88)
- Fakultät Elektrotechnik, Medizintechnik und Informatik (EMI) (ab 04/2019) (57)
- Fakultät Medien und Informationswesen (M+I) (bis 21.04.2021) (32)
- Fakultät Wirtschaft (W) (32)
- IMLA - Institute for Machine Learning and Analytics (28)
- INES - Institut für nachhaltige Energiesysteme (12)
- ivESK - Institut für verlässliche Embedded Systems und Kommunikationselektronik (11)
- Fakultät Medien (M) (ab 22.04.2021) (9)
- ACI - Affective and Cognitive Institute (6)
Open Access
- Open Access (305) (remove)
In cardiac resynchronization therapy (CRT) for heart failure, individualization of the AV delay is essential to improve hemodynamics and to minimize non-responder rate. In patients in sinus rhythm having additional disposition to bradycardia, optimization is necessary for both situations, atrial sensing and pacing. Therefore, echo-optimization is the goldstandard but time consuming. Unfortunately, it depends on the particular CRT systems parameter set if the resulting individually optimal AV delays can be programmed or not. Some CRT systems provide a set of AV delays for DDD operation combined with a set of the pace-sense-compensation to optimize the AV delay in DDD and VDD operation. The pace-sense-compensation (PSC) can be defined by the difference of implant-related interatrial conduction intervals in DDD and VDD operation measured in the esophageal left atrial electrogram. In a cohort of 96 CRT patients we found mean PSC of 59-35ms ranging between 0-143ms. As a consequence, allowing 10ms tolerance, AVD optimization is completely impossible in one of the two modes, VDD or DDD operation, in 34 (35%) or 5 (5%) patients with implants restricting the PSC range to 60ms or 100ms, respectively. Thus, we propose companies to provide CRT systems with programmable pace-sense- compensation between 0ms and 150ms.
A recognizable division appears between students with a comprehensive knowledge of the Web and those that are less certain about its resources. This is where, the teaching innovation Web Mentoring: Peer-to-Peer has been developed to help the students to cope better with the demands of media education. Furthermore, this presents the opportunity for master’s degree students to begin mentoring undergraduate students. Mentoring sessions have already been carried out successfully in the previous two semesters and are being presented, evaluated and discussed.
Generative convolutional deep neural networks, e.g. popular GAN architectures, are relying on convolution based up-sampling methods to produce non-scalar outputs like images or video sequences. In this paper, we show that common up-sampling methods, i.e. known as up-convolution or transposed convolution, are causing the inability of such models to reproduce spectral distributions of natural training data correctly. This effect is independent of the underlying architecture and we show that it can be used to easily detect generated data like deepfakes with up to 100% accuracy on public benchmarks. To overcome this drawback of current generative models, we propose to add a novel spectral regularization term to the training optimization objective. We show that this approach not only allows to train spectral consistent GANs that are avoiding high frequency errors. Also, we show that a correct approximation of the frequency spectrum has positive effects on the training stability and output quality of generative networks.
VR-based implementation of interactive laboratory experiments in optics and photonics education
(2022)
Within the framework of a developed blended learning concept, a lot of experience has already been gained with a mixture of theoretical lectures and hands-on activities, combined with the advantages of modern digital media. Here, visualizations using videos, animations and augmented reality have proven to be effective tools to convey learning content in a sustainable way. In the next step, ideas and concepts were developed to implement hands-on laboratory experiments in a virtual environment. The main focus is on the realization of virtual experiments and environments that give the students a deep insight into selected subfields of optics and photonics.
Cardiac resynchronization therapy is an established therapy for heart failure patients with sinus rhythm, reduced left ventricular ejection fraction and prolongation of QRS duration. The aim of the study was to evaluate ventricular desynchronization with electrical interventricular delay (IVD) to left ventricular delay (LVD) ratio in atrial fibrillation heart failure patients. IVD and LVD were measured by transesophageal posterior left ventricular ECG recording. In atrial fibrillation heart failure patients with prolonged QRS duration, the mean IVD-to-LVD-ratio was 0.84 +/- 0.42 with a range from 0.17 to 2.2 IVD-to-LVD-ratio. IVD-to-LVD-ratio correlated with QRS duration. IVD-to-LVD-ratio may be a useful parameter to evaluate electrical ventricular desynchronization in atrial fibrillation heart failure patients.
VDI Standard 4521: Status
(2016)
VDI Guideline 4521 Part 1: “Inventive problem solving with TRIZ: Part 1 – Fundamentals and definitions” has been published on 2015-04-01. The standard will sharpen the image of TRIZ, facilitate cooperation, and support studying and teaching. It is not a textbook but concisely summarizes basic assumptions of TRIZ and its terminology. It gives an overview on specific methods and tools which will be described in the following parts.
As part of the design education at Offenburg University, the teaching in technical documentation is continuously optimised. In this study, numerous mechanical engineering students, ages 19 to 29, are observed using the eye tracking technology and a video camera while performing various design exercises. The aim of the study is to enhance the students’ ability to read, understand and analyse complex engineering drawings. In one experiment, the students are asked to perform the “cube perspective test” after Stumpf and Fay to assess their ability for mental rotation as part of spatial visualization ability. Furthermore, the students are asked to prepare and give micro presentations on a topic related to their studies. Students have a maximum of 100 s time for these presentations. Thus, they can practise presenting important information in a short amount of time, show their rhetorical skills and demonstrate their acquisition of basic knowledge. During the presentation, the eye movement of a few selected students is recorded to analyse their information acquisition. In a further test, the students’ eye movements are analysed while reading an engineering drawing that consists of multiple views. All the spatial connections have to be included based on the different component views. Including these and their acquired knowledge, the students are asked to identify the correct representation of a component view. Furthermore the subjects are describing the function of an assembly, a parallel gripper and then they are to mentally disassemble the assembly to replace a damaged cylindrical pin. Simultaneously, they are filmed using a video camera to see which terms the students use for the individual technical terms. The evaluation of the eye movements shows that the increasing digitalisation of society and the use of electronic devices in everyday life lead to fast and only selective perceptual behaviour and that students feel insecure when dealing with technical drawings. The analysis of the videos shows a mostly non-technical and inaccurate manner of expression and a poor use of technical terms. The transferability of the achieved results to other technical tasks is part of further investigations.
The University for Children is a very successful event aiming to spark children‧s interest in science, in this particular lecture in Optics and Photonics. It is from brain research that we know about the significant dependence of successful learning on the fun factor. Researchers in this field have shown that knowledge acquired with fun is stored for a longer time in the long-term memory and can be used both more efficiently and more creatively [1], [2]. Such an opportunity to inspire the young generation for science must not be wasted. The world of Photonics and Optics provides us with a nearly inexhaustible source of opportunities of this kind.
Convolutional neural networks (CNN) define the state-of-the-art solution on many perceptual tasks. However, current CNN approaches largely remain vulnerable against adversarial perturbations of the input that have been crafted specifically to fool the system while being quasi-imperceptible to the human eye. In recent years, various approaches have been proposed to defend CNNs against such attacks, for example by model hardening or by adding explicit defence mechanisms. Thereby, a small “detector” is included in the network and trained on the binary classification task of distinguishing genuine data from data containing adversarial perturbations. In this work, we propose a simple and light-weight detector, which leverages recent findings on the relation between networks’ local intrinsic dimensionality (LID) and adversarial attacks. Based on a re-interpretation of the LID measure and several simple adaptations, we surpass the state-of-the-art on adversarial detection by a significant m argin and reach almost perfect results in terms of F1-score for several networks and datasets. Sources available at: https://github.com/adverML/multiLID