13 Zentrale Universitätseinrichtungen
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/14
Browse
Search Results
Item Open Access Reproducing, extending and updating dimensionalty reductions(2021) Debnath, MunmunDimensionality reduction techniques play a key role in data visualization and analysis, as these techniques project high-dimensional data in low-dimensional space by preserving critical information about the data in low-dimensional space. Dimensionality reduction techniques may suffer from various drawbacks, e.g., many dimensionality reduction techniques are missing a natural out-of-sample extension, i.e., the ability to insert additional data points into an existing projection. Therefore when a data set grows and new data points are introduced, the projection has to be recalculated, which often cannot be well related to the previous projection. This thesis proposes a technique based on kernel PCA to reproduce and update the result of dimensionality reduction techniques to overcome the stated problems with better run-time performance. The proposed technique uses an initial projection provided by an arbitrary dimensionality reduction technique as a template of the embedding space. A corresponding kernel matrix is then approximated to project out-of-sample instances. The approach is evaluated on several datasets for reproduction of projections of different dimensionality reduction techniques. It is shown that the proposed technique provides a coherent projection for out-of-sample data, and has a better run-time performance than several other dimensionality reduction techniques.Item Open Access Augmented visualization guidance for wire placement in timber fabrication(2021) Zhu, MarcoAugmented reality has become a broad area over time, which opens up many opportunities, also for industry. Nevertheless, digital solutions have not yet been used for some industrial applications. We focus on the use case “wire placement in timber fabrication”. The task here is to place and fix a cable along a certain route on a timber panel. However, there are several important factors and risks to consider that could cause the task to fail. Usability and efficiency play an essential role in optimally supporting use cases like this. Furthermore, we hope to make the task of this special industrial use case easier for workers by using augmented reality. For such an optimization, the design of a suitable visualization as part of the augmented reality system is indispensable. We constructed possible virtual representations for this use case and developed different suitable prototypes. Based on a user study, we compared the visualizations of these prototypes, evaluated their efficiency and user experience by measuring the workload, usability, task completion time, and determining error rates. By analyzing the developed visualization prototypes, a certain prototype proved to be optimal. With this, the route of the cable is shown as a path on the timber plate and displayed in color. The efficiency differences between this prototype and the other prototypes, supported by augmented reality, are not severe. The use of the augmented reality system turned out to be an essential enrichment. Compared to an equivalent system that is not supported by augmented reality, the developed augmented reality systems showed clear advantages. The suitability of this prototype has been checked and proven through a dry run. The task of the use case could be carried out successfully while the important factors and risks were also considered. The final fabrication has yet to be carried out.Item Open Access On-the-go authoring visualizations through wearable keyboards(2021) Dosdall, SarahNowadays, mobile computing allows users to take advantage of a computer anywhere and anytime. Complex interactions like text entry or programming on-the-go remain challenging. This is due to the lack of input options via keys [FZ94]. One option for solving these problems are chorded keyboards that are connected to the mobile computing device via Bluetooth. The main feature of keyboards is the creation of characters by playing chords and not by single keystrokes. Furthermore, the keyboards have a small size due to their wearability. In this thesis, the chorded keyboard Tap Strap 2 is evaluated as a possible input tool for AVAR, a tool for the visualisation of data in Augmented Reality. For this purpose, a special mapping of the Tap Strap combinations was created that is suitable for programming Pharo code. The focus of the thesis is the design, implementation, and preliminary evaluation of a web application called Learn2Tap. This application is designed to help users learn the Tap Strap combinations and train them to write Pharo code using the Tap Strap. The usability of the Tap Strap and Learn2Tap was evaluated through a 20-day user study. Users were given 15 minutes daily to use Learn2Tap. The outcome of the study was determined through interviews, user tracking, weekly testing and daily saving of learning status. The result of the study is that the users would not choose to use the Tap Strap as an input device because of the difficulties they encountered. However, if they had to learn the combinations of the Tap Strap a second time, they would use Learn2Tap again.Item Open Access System zur visuellen Analyse for Daten der kontinuierlichen Glukosemessung von Diabetes-Patienten(2021) Kaya, MuhammedDie vorliegende Arbeit beschäftigt sich mit der visuellen Analyse von Daten der kontinuierlichen Glukosemessung. Hierzu wird ein Visualisierungssystem entwickelt, welches in der Lage ist, Muster und Ausreißer in den Daten mithilfe von verschiedenen Visualisierungen darzustellen und dazugehörige Statistiken zur Auswertung der Daten zu generieren. Erreicht wird dies mithilfe von unüberwachtem maschinellen Lernen in Kombination mit klassischen Zeitreihen-Visualisierungen. Zunächst werden theoretische Grundlagen gemeinsam mit aktuellen Visualisierungsansätzen aus verschiedenen verwandten Arbeiten vermittelt. Darauf basierend wird ein Visualisierungsansatz für die Analyse der Daten vorgestellt. Diese beinhalten Kastengrafiken, Heatmaps, Linien-, Säulen-, Kreis- und Streudiagramme. Schließlich wird dazu ein Visualisierungssystem realisiert, welches in Form von drei Fallstudien evaluiert wird. Dafür werden reale Daten verwendet, die aus der kontinuierlichen Glukosemessung verschiedener Patienten/-innen entstanden sind. Die Ergebnisse der Fallstudien konnten zeigen, dass sich der Einsatz von unüberwachten maschinellen Lernverfahren in Kombination mit klassischen Zeitreihen-Visualisierungen für die Erkennung von Mustern und Ausreißern in den Daten verwenden lassen.Item Open Access Immersive Miniaturkarten auf der HoloLens 2(2021) Scheurenbrand, PatrickMit dem Aufkommen von Augmented-Reality-Headsets (AR-Headsets) für Normalverbraucher entstehen neue Möglichkeiten und es stellen sich neue Fragen. Ist es beispielsweise möglich, mit AR-Headsets Kartendarstellung im Vergleich zu aktuellen Methoden grundlegend zu verbessern? Ziel dieser Arbeit war es herauszufinden, wie traditionelle Karten mit der Hilfe von AR-Headsets wieder Hololens 2, verbessert werden können. Dabei wurden unterschiedliche Kartendarstellungsarten entwickelt und untersucht. Ein großer Teil der Arbeit beschäftigt sich mit einem Kartenmodus, in welchem der Nutzer selbst in der Karte steht. Um zu verhindern, dass der Nutzer Räume verdeckt,werden diese von ihm weggeschoben. Dafür wurde einerseits eine Fisheye-Verzerrung verwendet. Anderseits wurde noch versucht, mittels der Physik-Engine die Räume zu verschieben. Die dadurch entstandenen Probleme werden besprochen und unterschiedliche Lösungsansätze werden vorgestellt. Zum Schluss wurden die Kartendarstellungsarten mittels einer Benutzerstudie evaluiert.Item Open Access Improving the MPI-IO performance of applications with genetic algorithm based auto-tuning(2021) Bagbaba, Ayse; Wang, XuanParallel I/O is an essential part of scientific applications running on high-performance computing systems. Under- standing an application’s parallel I/O behavior and identifying sources of performance bottlenecks require a multi-layer view of the I/O. Typical parallel I/O stack layers offer many tunable parameters that can achieve the best possible I/O performance. However, scientific users do often not have the time nor the experience for investigating the proper combination of these parameters for each application use-case. Auto-tuning can help users by automatically tuning I/O parameters at various layers transparently. In auto-tuning, using naive strategy, running an application by trying all possible combinations of tunable parameters for all layers of the I/O stack to find the best settings is an exhaustive search through the huge parameter space. This strategy is infeasible because of the long execution times of trial runs. In this paper, we propose a genetic algorithm-based parallel I/O auto-tuning approach that can hide the complexity of the I/O stack from users and auto-tune a set of parameter values for an application on a given system to improve the I/O performance. In particular, our approach tests a set of parameters and then, modifies the combination of these parameters for further testing based on the I/O performance. We have validated our model using two I/O benchmarks, namely IOR and MPI-Tile-IO. We achieved an increase in I/O bandwidth of up to 7.74×over the default parameters for IOR and 5.59× over the default parameters for MPI-Tile-IO.Item Open Access Lustre I/O performance investigations on Hazel Hen : experiments and heuristics(2021) Seiz, Marco; Offenhäuser, Philipp; Andersson, Stefan; Hötzer, Johannes; Hierl, Henrik; Nestler, Britta; Resch, MichaelWith ever-increasing computational power, larger computational domains are employed and thus the data output grows as well. Writing this data to disk can become a significant part of runtime if done serially. Even if the output is done in parallel, e.g., via MPI I/O, there are many user-space parameters for tuning the performance. This paper focuses on the available parameters for the Lustre file system and the Cray MPICH implementation of MPI I/O. Experiments on the Cray XC40 Hazel Hen using a Cray Sonexion 2000 Lustre file system were conducted. In the experiments, the core count, the block size and the striping configuration were varied. Based on these parameters, heuristics for striping configuration in terms of core count and block size were determined, yielding up to a 32-fold improvement in write rate compared to the default. This corresponds to 85 GB/s of the peak bandwidth of 202.5 GB/s. The heuristics are shown to be applicable to a small test program as well as a complex application.Item Open Access Visualization of differences in perception caused by vision deficiency(2021) Amann, MarcoSimulation of visual impairments may be used to counter design exclusion and improve the accessibility of products. Whilst first-hand experience of the simulated scene is important, having to manually look for differences caused by vision deficiencies, let alone judge their severity, is time-consuming and error-prone. In this work, we develop an error metric that tracks errors and uncertainty throughout the stages of a simulator for visual impairments. This allows us to visualize errors and uncertainty independently in all tracked dimensions. We add more advanced view modes to the simulator to enable visualization of our error metric in combination with the input and output images. We further extend the simulator to include simulations for strabismus, astigmatism and retinal ganglion cells. By simulating several combinations of vision deficiency, we found that a visualization of the proposed metric can be used to identify problematic areas in a scene. Depending on the use case, one may need to select different combination functions generating scalar values, e.g. the aggregated standard deviation of the RGBXY vector. With a comparison of our metric to SSIM, we found that our metric can cope better with displaced features of an image but may produce blurred visualizations. We conclude that although our metric imposes considerable performance penalties on the simulator, it has advantages compared to approaches based exclusively on input and output images.Item Open Access Visualization for human-AI collaborative music composition(2021) Rau, SimeonWe propose an AI-assisted approach based on interactive visualizations to support users in composing music and getting insights into the AI through hyperparameter analysis. Our user-centered approach allows the user to better control the composition by steering the AI’s suggestions. We use symbolic music data and piano rolls as visual music notation for easier understanding for amateur users and interaction with the notes of a melody. As the user requests multiple possible continuation for a given seed melody, and also multiple continuations for each of the previous continuations, a tree or graph structure of melodies occurs. We visualize this structure with an icicle plot, where the nodes are represented by a piano roll, to show the hierarchical structure of the melody samples. To add sorting options for easier sample selection, while still displaying the structure, we added links between the nodes. Both visualizations enable listening to selected melodies. For larger numbers of generated suggestions, we added a similarity-preserving scatterplot to visualize all samples at the same time with different glyphs representing melody samples. The scatterplot improves the efficiency of sample selection, as similar samples are close together and the user can disregard entire neighborhoods if one sample does not fit at all. We support brushing the scatterplot to select neighborhoods for which we then show visual aggregations to allow for insights into groups. To evaluate our design, we conducted a pair analytics study with two participants with limited musical knowledge. Both participants were able to quickly create compositions they liked and found our approach helpful. They also learned new things about the AI, like the influence of the hyperparameter temperature on the resulting melody.Item Open Access Augmented reality drum kit feedback visualization(2021) Lammert, JonasThis thesis explores several visualizations that aim to assist aspiring drummers with learning the drums using projector-based augmented reality technology. It aims to provide feedback on their training sessions when they do not have a professional teacher available. This feedback is projected directly onto the instrument, so the user does not have to transfer the information from e.g. a screen to their drums and can save cognitive resources. We give feedback in real-time and retrospectively. For the real-time feedback, we have to consider a visualization that keeps the information lightweight, to not divert too much attention away from playing the instrument, so we use simple encodings like color or a speedometer. For retrospective feedback, we use more complex visualizations like a pie chart. Our approach can give feedback on timing and dynamics of the played pattern. We evaluated our visualizations with three case studies, one that focuses on comparing the different visualizations, one that evaluates their efficiency in identifying errors, and one that investigates the influence of the visualization regarding errors made over time. We conclude that the visualizations help the user identify their errors, however, we did not have enough data to truly verify that they reduce errors in one’s performance over time, or if one visualization has a better performance than the others.