Emilio Paolini, our affiliated PhD candidate, was invited to the joint Abdus Salam International Centre for Theoretical Physics (ICTP)–International Atomic Energy Agency (IAEA) School on FPGA-based Systems-on-Chip and its Applications to Nuclear and Scientific Instrumentation.
On November 16th, he gave a lecture about quantization in neural networks, focusing on the advantages and the limitations that this technique can introduce. The lecture included a hands-on session, showing how to implement quantization and the impact on the performance of actual neural networks.Â