THE ROYAL INSTITUTION FOR THE ADVANCEMENT OF LEARNING / MCGILL UNIVERSITY

Published purpose

This project exploits machine learning algorithms in the design of the next generation photonic circuits with enhanced performance. Taking the approach further, machine learning is also exploited in creating a digital twin of silicon photonic fabrication, so that fabrication induced variations can be accurately captured and pre-emptively corrected. Photonic integrated circuits render optical system ideas more practical with increased performance along with reduced size and cost. Silicon photonics, in particular, exploiting the ubiquitous microelectronic miniaturization infrastructure, enables complex optical modules that can be fit into an iPhone, with applications from optical and quantum communications to medical diagnostics. Future applications demand further size reduction and increased functionality. Among the major challenges to achieving these goals, one is to create optical functionalities beyond classical optics. Another is the miniaturization leads to variability in their performance after fabrication.