Information Theoretic Limits of Deep Learning
In this agility project we consider deep learning as a machinery for fine-tuning, enhancing or completely replacing various traditional algorithms in communications based on signal processing and information theory. For example, we utilise deep learning for large-scale channel estimation in wireless networks, simultaneous communication and tracking, or sparse recovery problems. Furthermore, to address new requirements for compression and communication of multimedia signals in future machine-to-machine type scenarios, we consider deep-learning based codecs for IoT applications.
Fellow
Research Group Lead
Postdoctoral Researcher