Refine
Year of publication
Document Type
- Conference Proceeding (308)
- Article (288)
- Master's Thesis (113)
- Part of a Book (53)
- Book (19)
- Doctoral Thesis (9)
- Report (6)
- Preprint (5)
- Working Paper (4)
- Other (3)
Institute
- Forschungszentrum Mikrotechnik (247)
- Forschungszentrum Business Informatics (149)
- Technik | Engineering & Technology (127)
- Department of Computer Science (Ende 2021 aufgelöst; Integration in die übergeordnete OE Technik) (112)
- Wirtschaft (106)
- Forschungszentrum Energie (79)
- Didaktik (mit 31.03.2021 aufgelöst; Integration ins TELL Center) (37)
- Forschungszentrum Human Centred Technologies (35)
- Soziales & Gesundheit (34)
- Josef Ressel Zentrum für Materialbearbeitung (27)
Language
- English (815) (remove)
Keywords
- Laser ablation (11)
- Y-branch splitter (11)
- arrayed waveguide gratings (11)
- photonics (8)
- Evolution strategy (7)
- Demand side management (6)
- Optimization (6)
- integrated optics (6)
- AWG (5)
- Arrayed waveguide gratings (5)
Many test drives are carried out in the automotive environment. During these test drives many signals are recorded. The task of the test engineers is to find certain patterns (e.g. an emergency stop) in these long time series. Finding these interesting patterns is currently done with rule based processing. This procedure is very time consuming and requires a test engineer with expertise. In this thesis it is examined if the emerging field of machine learning can be used to support the engineers in this task. Active Learning, a subarea of machine learning, is used to train a classifier during the labeling process. Thereby it proposes similar windows to the already labeled ones. This saves the annotator time for searching or formulating rules for the problem. A data generator is worked out to replace the missing labeled data for tests. The custom performance measure “proportion of seen samples” is developed to make the success measurable. A modular software architecture is designed. With that, several combinations of Time Series Classification algorithms and query strategies are compared on artificial data. The results are verified on real datasets, which are open source available. The best performing, but computational intensive solution is an adapted RandOm Convolutional KErnel Transform (ROCKET). The custom query strategy “certainty sampling” shows the best results for highly imbalanced datasets.
Concept of probabilistic modeling for real-time prediction of product quality and design automation
(2018)
Progress in integrated photonics enables development of integrated photonics circuits with new unique properties, circuits of the future, and overcomes current limits in information and communication technologies. The packaging of photonic integrated circuits is necessary for taking them out of research laboratories into real implementation in the information and communication technology applications.
Telecom optical fibers are still being the best transmission medium of digital data and analogue signals for long distance applications. The effective coupling of optical radiation between telecom optical fiber with ten microns core dimension and photonic integrated circuits optical waveguides with submicron dimensions are necessary. To address these challenges, we present our concept of photonics integrated circuit packaging with radio frequency, direct current and fiber array ports with automated active alignment system.
This master thesis investigates a Computational Intelligence-based method for solving PDEs. The proposed strategy formulates the residual of a PDE as a fitness function. The solution is approximated by a finite sum of Gauss kernels. An appropriate optimisation technique, in this case JADE, is deployed that searches for the best fitting parameters for these kernels. This field is fairly young, a comprehensive literature research reveals several past papers that investigate similar techniques.
To evaluate the performance of the solver, a comprehensive testbed is defined. It consists of 11 different Poisson equations. The solving time, the memory consumption and the approximation quality are compared to the state of the art open-source Finite Element solver NGSolve. The first experiment tests a serial JADE. The results are not as good as comparable work in the literature. Further, a strange behaviour is observed, where the fitness and the quality do not match. The second experiment implements a parallel JADE, which allows to make use of parallel hardware. This significantly speeds up the solving time. The third experiment implements a parallel JADE with adaptive kernels. It starts with one kernel and introduce more kernels along the solving process. A significant improvement is observed on one PDE, that is purposely built to be solvable. On all other testbed PDEs the quality-difference is not conclusive. The last experiment investigates the discrepancy between the fitness and the quality. Therefore, a new kernel is defined. This kernel inherits all features of the Gauss kernel and extends it with a sine function. As a result, the observed inconsistency between fitness and quality is mitigated.
The thesis closes with a proposal for further investigations. The concepts here should be reconsidered by using better performing optimisation algorithms from the literature, like CMA-ES. Beyond that, an adaptive scheme for the collocation points could be tested. Finally, the fitness function should be further examined.
Compression of ultrashort laser pulses via gated multiphoton intrapulse interference phase scans
(2014)
Complementarities and synergies of quadruple helix innovation design in smart city development
(2020)
Increased urbanization trends are stimulating regional needs to support transitions from urban environments to smart cities, using its holistic perspective as a source to innovation. Strong relations between smart cities, urban and regional development, are getting increased attention both at policy and implementation level, providing fertile ground for execution of the new European policy frameworks that supports quadruple helix approaches to innovation. Smart specialization strategies (RIS3) encompass such initiatives, placing ICT and collaboration between academia, industry, government, and citizen at the center of urban innovation. However, there is still lack of research on effects of such approaches to innovation, involving both quadruple helix clusters and ICT in utilizing innovation potentials for developing smart cities. This study aims to increase the understanding on how quadruple helix urban innovation strengthens competitiveness of regions by improving its local smart areas – RIS3. We identified smart specialization patterns and applied comparative benchmark between nine smallmedium sized urban regions in Central Europe. Building on these results, the study provides an overview of the effects of RIS3 strategies implemented through quadruple helix innovation clusters on competitiveness of regions and Smart City development.
A concept for a recommender system for the information portal swissmom is designed in this work. The challenges posed by the cold start problem and the pregnancy-related temporal interest changes need to be considered in the concept. A state-of-the-art research on recommender systems is conducted to evaluate suitable models for solving both challenges. The explorative data analysis shows that the article's month of pregnancy is an important indicator of how relevant an article is to a user. Neither collaborative filtering, content-based filtering, hybrid models, nor context-aware recommender systems are applicable because the user's pregnancy phase is unknown in the available data. Therefore, the proposed recommender system concept is a case-based model that recommends articles which belong to the same gestation phase as the currently viewed article.
This recommender system requires that the month of pregnancy, in which an article is relevant, is known for each article. However, this information is only available for 31% of all articles about pregnancy. Consequently, this work looks for an approach to predict the month of gestation based on the article text. The challenges with this are that only few training data are available, and the article texts of the various months of pregnancy often contain the same terms, considering all articles are about pregnancy. A keyword-based approach using the TF-IDF model is compared with a context-based approach using the BERT model. The results show that the context-based approach outperforms the keyword-based approach.
Comparison of silicon nitride based 8-channel 100-GHz AWGs applying different waveguide structures
(2019)
This paper presents design and simulation of 8-channel, 100-GHz AWGs based on Si/SiO2/SiN/SiOx material platform. For the designs, two different waveguide structures were used, i.e. ridge and rib waveguides. AWGs were designed for central wavelength of 850 nm applying AWG-Parameters tool. The simulations were performed applying FEM and BPM methods in RSoft and PHASAR photonic tools. The simulation results show considerably lower losses but slightly higher channel crosstalk when applying rib waveguides.
Comparison of silicon nitride based 1x8 Y-branch splitters applying different waveguide structures
(2019)
This paper presents design, simulation and optimization of 1x8 Y-branch power splitters based on Si/SiO2/SiN/SiOx material platform. For the designs, two different waveguide structures were used, i.e. ridge and rib waveguides. The splitters were designed for 850 nm spectral optical window and the simulations were performed applying FEM and BPM methods in RSoft photonic tool. The aim of this work was to find minimum physical dimensions of the designed splitters occupying minimal space on PIC chip. The optimization was done with regards to high symmetrical splitting ratio and low insertion loss. Finally, the optical properties of both splitters were studied and compared with each other.