Matthias Rottmann obtained the habilitation with his research on AI
With the oral defense, Dr. Matthias Rottmann has successfully completed his habilitation in mathematics on April 21, 2023. The title of his habilitation thesis is "Uncertainty and Robustness of Deep Learning". There, he deals predominantly with the quantification of uncertainty in deep neural networks and its impact in the field of perception. On the one hand, this has application in downstream tasks such as active learning and sensor fusion. On the other hand, the developed methods are applied in different tasks of image processing, from image classification over object detection to semantic segmentation as well as instance segmentation.
The figure shows the semantic segmentation of a tractor on the left. The tractor does not belong to any of the classes used when training the neural network. On the right you can see the uncertainty detection. Green stands for low uncertainty and red for high uncertainty.
Last modified: 11.12.2023
