Stefan Müller

Compression of Deep Neural Networks: Development of approaches to compress Deep Neural Networks to reduce their model size (up to compression factors 15-25). Thereby ressources like memory-consumption/footprint, coumputation and energy cost are also reduced tremendously. That makes the deployment of Deep Neural Networks and Deep Learning on mobile and embedded devices much more affordable. Technologies: Python, Caffe, CUDA

node.js, C++

node.js, C++, REST

Intelligente/autonome/verteilte Informationssysteme, Deep Learning, Informationssicherheit, Software Design und Realisierung

Industrielle Automatisierung, Software Engineering, Digitale Bildverarbeitung