Peter Schlicht

TL;DR: An extensive empirical evaluation validates major improvement of the trade-off between model performance and communication which could be beneficial for numerous decentralized learning applications, such as autonomous driving, or voice recognition and image classification on mobile phones.

TL;DR: This paper adapts a known domain adaptation approach to work in an unsupervised fashion for semantic segmentation on high resolution data and provides some analysis of the learned representations.

TL;DR: This procedure yields an almost plug and play post-processing tool to rate the prediction quality of semantic segmentation networks on segment level, especially relevant for monitoring neural networks in online applications like automated driving or medical imaging where reliability is of utmost importance.

TL;DR: This work addresses the problem with the use of a redundant teacher-student framework, consisting of a static teacher network, a static student network, and a constantly adapting student network to show that a significant robustness increase of student DNNs against adversarial attacks is achieveable, while maintaining semantic segmentation quality at a reasonably high level.