MEMS gyroscopes stability is crucial for long-term inertial navigation. This thesis explores the stability of the MEMS gyroscopes specific to the mode-split architecture, commonly used in mass market inertial sensors. It focuses on the offset stability of those devices and analyzes both systematic and stochastic offset contribution. One example of systematic effects is the change in environmental temperature, which can be compensated through system understanding. Thus, a temperature model of the gyroscope offset was developed and compared against the measurements of forty research devices. At the same time, stochastic contributions define the best-achievable limit for offset stability. To uncover the root causes of the bias instability, a scale factor instability model was derived and combined with a phase space gyroscope model. The scale factor instability model carries the information of the amplitude change of the signals in a gyroscope system, whereas the phase models the phase noise shaping. The combination of the two models resulted in a bias instability model applicable to both rate and quadrature signals. Based on the model prediction, bias instability improvement of up to 40 % was achieved, through the reduction of the PLL input phase flicker noise. The bias instability model was identified sensorindividually on four triaxial research gyroscope devices, equivalent to having twelve single axis devices. The proposed scale factor and bias instability models provide a measurement methodology to distinguish different flicker noise contributions. This methodology was successfully applied to available research devices at Robert Bosch GmbH, successfully identifying the dominant bias instability (flicker) noise sources. Depending on the system performance either temperature or flicker noise may be the limiting factors for long-term gyroscope rate signal stability. Hence, both the temperature and bias instability model represent a step towards gyroscope stability improvement, and in turn more precise navigation.

In today’s technological world, data collection, preservation and transfer has become one of the most important means of means of interpreting information and gaining knowledge from it. Particularly in modern medicine, medical data, which is collected through various technological means, such as smartwatches, tablets, medical devices and through the input of medical personnel into various software solutions, are essential for providing effective and patient-oriented medical treatment. However, this life-saving information is fragmented across different platforms and areas (homes, hospitals, doctors) and cannot be consolidated for further treatment. This lack of access to important external knowledge sources, such as clinical trials and cutting-edge research, leads to incomplete utilization of the available potential. This data could support medical staff in providing a coherent and rapid basis for decision-making based on personal data. This plays a particularly important role in rescue operations. In the short time available to the rescue service, the new knowledge about the situation, the patient Medical vital signs such as blood pressure, ECG and heart rate help to classify the patient on site and treat them correctly. Technical solutions could help here by using hybrid knowledge models to bundle medical knowledge and use artificial intelligence to provide support in decision-making for rescue personnel. This thesis deals with the design of an innovative hybrid knowledge graph model for the integration of (non-) medical medical knowledge into a knowledge graph developed for medicine. The aim is to use intelligent knowledge fusion methods to integrate different knowledge sources (e.g. historical data, vital parameters), to enable more efficient, evidence-based patient treatment to make them usable and thus reduce the complexity of combining them. In the long term, the digital twin is to be used as a knowledge model, to combine knowledge streams and formulate new findings (with the help of DT services). This can initiate a knowledge reform in medicine medical knowledge, whereby distributed medical expertise accumulates added value for patients patients, medical staff and society.

Eating is never just a matter of the palate – it is a medial, technical, and cultural event. This issue explores how digital sensors, AI, and social media are reshaping our perception of taste, smell, and texture. From electronic noses and machine tasting to food photography and AI-generated recipes, culinary culture emerges as a field where senses, media, and technologies intersect in new ways.

The rapid digitalization of the information age has led to challenges such as information overload, underscoring the critical role of Recommender Systems (RecSys) in organizing and delivering relevant content. Despite their ubiquity, the development of RecSys remains resource-intensive and labor-intensive. Additionally, the environmental impact of RecSys has emerged as a critical yet underexplored concern as modern approaches increasingly employ resource-intensive architectures. The field of Automated Machine Learning (AutoML) has demonstrated significant success in streamlining model development for general machine learning tasks, lowering the barrier of entry for researchers and practitioners by reducing necessary manual labor and expertise while increasing performance. Motivated by their advancements, we investigate the transfer of AutoML principles to RecSys in the framework of Automated Recommender Systems (AutoRecSys) in this dissertation. We focus on solving the automated algorithm selection problem, due to its high relevance for efficient RecSys modeling, and address significant optimization confounders in AutoRecSys. We propose an algorithm selection framework that addresses dataset limitations through community-contributed metadata while offloading computationally intensive meta-learning tasks to server-side components for efficient client-side deployment. Additionally, we provide the first analysis of algorithm selection for ranking prediction tasks with implicit feedback datasets, showing the correlation with ground-truth performance for traditional and AutoML-based meta-models. We quantify the impact of randomness during data splitting, showing that it leads to significant performance deviations unless mitigated through repeated experiments or cross-validation. Furthermore, we analyze the suitability of top-N metrics for optimization, showing that their use in validation does not introduce confounding effects in RecSys evaluation. This reinforces the reliability of conventional evaluation methodologies. Beyond automation and evaluation concerns, we conduct the first comprehensive investigation of the environmental impact of RecSys experiments. We reveal that modern deep learning-based papers emit 42 times more CO2 equivalents than papers employing traditional approaches. Furthermore, we introduce a software tool for measuring and reporting energy consumption in RecSys experiments, enabling researchers to understand and report their environmental impact. Finally, we synthesize our contributions and demonstrate that automated algorithm selection can amortize its environmental impact through widespread adoption. To summarize, this dissertation lays the foundation for future research in algorithm selection through the AutoRecSys framework, further reduces uncertainties for RecSys evaluation methodologies, and helps the RecSys community to understand and address their environmental impact for a sustainable future.