Professor Foutse Khomh and his team from Polytechnique Montréal's Department of Computer Engineering and Software Engineering has been awarded a Canada-CIFAR Chair in AI.
This 5-year chair supports the work of Professor Khomh and members of the SWAT (SoftWare Analytics and Technologies) lab at Polytechnique Montréal, who are developing tools to deconstruct machine learning algorithms to ensure their efficiency, reliability, security, and safety.
Even if artificial intelligence (AI) algorithms "learn" by themselves, the fact remains that their skills must be validated by humans - that's where the SWAT team comes in.
"Models sometimes work very well in the research environment where they have been trained, but before you take them into production (i.e. the real world), you have to ensure that they don't have any biases or issues that will make them unreliable when presented with new data," explains Professor Khomh.
To accomplish this task, the team is developing techniques and software tools to assess how well an algorithm is learning, and how close it is to reaching its learning limits. The techniques and tools developed by the SWAT team help to ensure the quality of machine learning-based software systems throughout their life cycle, from the design phase to production.
Objective: AI certification
This work is of great interest to various sectors where AI is used as a decision-making aid. Professor Khomh offers the example of the medical sector, where AI tools are already using images to identify which tissues are likely to be cancerous.
His group is also interested in the use of AI in aeronautics, with the aim of enabling the certification of AI systems by government agencies.
"To certify a system, you have to demonstrate that you understand how it works extremely well, and be able to guarantee that it's trustworthy," says the researcher, who is also affiliated with the Institut québécois d'intelligence artificielle (MILA) and the Institut de valorisation des données (IVADO). "The tools developed by my team allow us to better understand how AI systems function, to identify failures within these systems, to suggest corrective measures, and to provide guarantees about their functioning, which permits us to get closer to the certification objective. However, a lot remains to be done.
Canada-CIFAR Chairs in AI have been awarded annually since 2017, and a total of 114 Canadian professors have benefited to date. In addition to Professor Foutse Khomh, four other MILA-affiliated professors have been awarded a Canada-CIFAR Chair this year.
Further, in addition to Professor Khomh, two other researchers at Polytechnique hold Canada-CIFAR AI Chairs. The latter two professor are Full Professor Christopher Pal and Assistant Professor Sarath Chandar Anbil Parthipan, both of whom are also part of the Department of Computer Engineering and Software Engineering.
Congratulations to Professor Khomh!
Learn more
Professor Foutse Khomh expertise https://www.polymtl.ca/expertises/en/khomh-foutse
SWAT Laboratory website http://swat.polymtl.ca/
Department of Computer Engineering and Software Engineering website https://polymtl.ca/gigl (in French)
CIFAR website https://cifar.ca
Receive the Funding Newswire each Tuesday, our Policy Bulletin each Thursday, and news about bridging Europe’s east-west innovation gap twice a month in The Widening. 
A unique international forum for public research organisations and companies to connect their external engagement with strategic interests around their R&D system.
