University of Waterloo

Published purpose

The Project will enable improved training of deep neural networks, build models for AI-enabled COVID-19 diagnosis and risk stratification through applying few-shot learning methods, and address current limitations of the low quantity of annotated COVID-19 positive data. The Project will accelerate the development of an open-source, open-access diagnostic platform. The potential for an epidemic or pandemic caused by a high impact respiratory pathogen is increasing and the frequency of outbreaks of newly emerging diseases is rising. Exploration of few-shot learning techniques to enable improved training of deep neural networks has the potential to apply to other emerging epidemics and pandemics when the available data is limited.