On April 14, 2025, the Faculty of Electronic Engineering, University of Niš, had the pleasure of hosting Prof. Davide Bertozzi from the University of Manchester for a specialized training session focused on Dynamically-Reconfigurable Deep Learning Accelerators.

During this expert visit, Prof. Bertozzi shared valuable insights into the design and implementation of dynamic neural network models—models capable of adapting their structure or parameters in response to input during inference. The session also explored the hardware/software stack required to support such adaptive systems, with a particular emphasis on the open-source, industry-grade NVDLA architecture.

This event offered attendees a unique opportunity to deepen their understanding of advanced AI acceleration techniques and engage in discussion on the challenges and possibilities of deploying adaptive models in real-world scenarios.

The visit was organized as part of the AIDA4Edge project, with support from the European Union and UK Research and Innovation.

We thank Prof. Bertozzi for his visit and inspiring contribution, and we look forward to further collaboration and knowledge exchange in the field of intelligent edge computing.