15 September 2020
With a financial package of 7.5M, the Quebec component of the DEEL (DEpendable & Explainable Learning) project was launched on September 14 as part of a press event, organized on the sidelines of the CRIAQ annual general meeting.
This is a major project, the innovations of which “will undoubtedly have real impacts for the aerospace industry, both here and internationally”, emphasized Université Laval President, Sophie D’Amours. The initiative addresses one of the key issues in artificial intelligence (AI), for the aerospace industry and beyond: certification.
Compared to traditional computing – where the programmer tells the machine what to do to solve a task – in machine learning it is the machine itself that weaves its way. The programmer provides the computer with a large amount of data and a problem to solve. This makes it difficult to fully understand the path taken by the machine and, therefore, it is almost impossible to guarantee that the systems developed are flawless. “This inability to explain and certify AI systems limits their use, especially in critical systems such as on-board aircraft systems, as well as in a host of areas where the life and safety of people are at stake,” explains François Laviolette, professor and director of the Big Data Research Center at Université Laval and scientist in charge of the DEEL-Quebec project.
Over the next 5 years, 20 research teams from 5 Quebec universities (Polytechnique Montreal, UQAM, Université de Montréal, McGill University and Université Laval) will work alongside four Quebec giants in the aerospace sector (Thales Canada, CAE, Bombardier and Bell Textron Canada), together with the support of the Consortium for Research and Innovation in Aerospace in Quebec (CRIAQ) and IVADO to meet this challenge.