Overview
Visual computing is the research area at the intersection of computer graphics, computer vision, and machine learning. Our group works in this broader area using ideas from physically-based and geometric modeling to develop new machine learning methods and to better understand existing approaches.
Active Research areas:
- Understanding deep learning
- Machine learning methods, in particular for geometric data
- Interdisciplinary applications, in particular to natural sciences / computational physics
The main area of expertise of our group has been statistical methods for "shape understanding", such as reconstructing geometric information from sensor input, modeling of shape spaces, and generative models for 3D geometry. In recent years, such methods have been mostly based on deep neural networks, which offer unprecedented performance and generalization abilities. Therefore, a lot of our recent activities have focused on improving our understanding of deep learning methods, in particular, by linking them to pattern formation in dynamical systems. Applications of statistical data analysis and machine learning methods to interdisciplinary applications, in particular in physics, have been another focus area. We have also build expertise in scalable algorithms for visualizing and processing of large data sets.