The European Research Council (ERC) has awarded its Advanced Grants for groundbreaking research. Two KU Leuven researchers are among this year's recipients: engineer Bart De Moor and philosopher Jan Opsomer. One will attempt to find the best possible mathematical models, the other will seek for traces of Aristotle.
ERC Advanced Grants are awarded to established researchers with a track record of significant research achievements. The Advanced Grants are awarded for a five-year period and they can be worth up to €2.5 million.
All grant applications are assessed on the basis of one single criterion: the excellence of both the project and the researcher. The applications that pass the quality threshold are ranked on the basis of their score, and only the highest-ranked proposals receive funding.
Bart De Moor: the quest for the best mathematical model
Full professor at ESAT (Department of Electrical Engineering)
What exactly will you be investigating over the next five years?
"Every day, thousands of algorithms are used across various domains and for all sorts of purposes in the world of artificial intelligence, without any guarantee that the models used - calculated from the data via numerical algorithms - are the best ones. With this research project, I want to go back to basics and draw on insights from all sorts of subdisciplines in mathematics - from systems theory to algebraic geometry - to find out how users can be certain that they are using the best model from a class of models, and that the results generated by that model will be reproducible time and again."
"In this respect, this ERC grant is an elementary investment that allows us to fund researchers who'll go 'back to the roots', who'll investigate how we can absolutely guarantee the quality and reproducibility of the models. This is deeply fundamental research. Of course, our findings will be applied in many areas. As a research group, we have a long track record in this with our bilateral projects with AI users such as companies and hospitals, as well as through the eight spin-offs we've already launched - these include Cartagenia, Ugentec, Aspect Analytics and Trendminder - and the ones that will follow. Our research results will also be valorised through the recently launched KU Leuven AI Institute (Leuven.AI) in addition to the valorisation that's already happened through the Artificial Intelligence Flanders research programme, which I help coordinate."
Why is it important to use the 'best' mathematical model?
"We are ultimately talking about the reliability of artificial intelligence. Take the healthcare sector, for instance. The judgement of the physician who carries medical responsibility is the deciding factor in establishing the diagnosis. But a physician can also draw on algorithmic models that churn through large amounts of data, both the patient's and that of entire patient populations. Of course, this approach presupposes that the best possible algorithms and models are used, and that they generate reproducible and interpretable results. This is the only way that physicians can trust the assistance of AI."
What was your response when you discovered you had received the grant?
"I am on one of the selection committees for the ERC Starting Grants for young researchers. During one of the selection meetings last week - conducted over video, of course - I suddenly saw the email announcing that I had myself received an ERC Advanced Grant. That was a remarkable concurrence of events: to suddenly get the good news about your own application while you're evaluating the applications of young candidates. A bright spot in these dark times!"
