I hold a Bachelor's degree in Medical Informatics and a Master's degree in Media Informatics. Currently I am pursuing my PhD in the field of Software Visualization.
Fields of Research:
Modern endeavors in the natural sciences create data in bigger quantities and higher complexity than ever before in human history. Space missions for example gather data on celestial bodies and the fundamentals of the universe, high-energy particle collision experiments amass petabytes of data per year, and even the annual check-up at a doctor’s office generates considerable quantities of data. Gaining insight and not just information from these datasets is a difficult task.
Modern techniques from the field of computer graphics provide tools to efficiently visualize datasets. With their abilities to do parallel computations on a massive scale, modern GPUs can enable interactive exploration and handling of datasets at interactive framerates.
The combination of the natural science's data with the techniques of the field of computer graphics is Scientific Visualization. Algorithms for surface and volume rendering allow to create meaningful three-dimensional visualizations of experimentally gathered spatial data. Visualizing what's underneath a patient's skin in a non-invasive manner, simply by joining multiple MRI images is an example of these techniques.
A limiting factor for the usefulness and feasability of these approaches is performance. A simulation needs to terminate to provide results. A visual analytics system must respond to user input, no matter how big the underlying data is. Assessing and visualizing the performance of the rendering techniques are critical steps in increasing it, and thus the usefulness/feasability of applications using these techniques.
In my dissertation, I want to look at how to measure, gather, visualize, and then analyze performance data of current scientific visualization approaches. This not only requires a deep understanding of scientific visualization techniques (like volume and surface rendering), computer graphics (rendering techniques, rendering pipeline, software optimization), performance metrics (runtime statistics, profiling), but also knowledge about how to visualize complex information (information visualization, visual comparison, multivariate data, descriptive statistics). Benchmarking current rendering algorithms with actual datasets of the SciVis community, and using the results as the basis for developing a visual comparison system for multivariate runtime data might be one step to do this. I have already taken first steps in this direction by focussing my research so far on the visual comparison of multivariate runtime statistics and dynamic graphs.
SoftwareDynamics²: Fine-Grained Evolution of Software Behavior
Tarner, Hagen; Frick, Veit; Pinzger, Martin; Beck, Fabian: Exploring Visual Comparison of Multivariate Runtime Statistics. In: 9th Symposium on Software Performance 2018. Hildesheim 2018. PDF Citation Abstract Details
To understand program behavior or find performance bottlenecks in their software, developers need tools to efficiently compare runtime statistics collected across multiple executions. As there is a variety of useful metrics, a good visualization needs to be able to handle multivariate data and highlight the most important differences between multiple versions. We identify three scenarios for the comparison of execution-relevant changes, and explore possible visualizations of the gathered multivariate runtime statistics.
A list of courses I helped to realize:
- Project Group ViVaSD - Visualizing Variability in Software Dynamics
- Project Group Talking Realities
- Lecture and Exercise Modelle der Informatik
- Lecture and Exercise Computergrafik
- Lecture and Exercise Informations- und Softwarevisualisierung
- Seminar Software Performance Monitoring
- Seminar Software Visualization