Chairholders
Profile
François Laviolette
Department of Computer Science and Software Engineering
Université Laval
Chair title
NSERC/Intact Financial Corporation Industrial Research Chair in Machine Learning for Insurance
Chair program
Industrial Research Chairs program
Role
Senior Chairholder since 2018
Summary
Machine learning is the study of algorithms that will allow a computer system to learn how to perform a task without being explicitly programmed to do so. In recent years, this field of research has experienced significant growth. It has led to several major artificial intelligence applications, and it can effectively add value to big data. Despite these recent successes, many areas of application remain closed to the use of machine learning due to the lack of interpretability of the learning models. The work of Dr. Laviolette, chairholder of this Industrial Research Chair, focuses mainly on the fundamental aspects of machine learning, which naturally includes interpretability. Research conducted under his direction has led to impressive interpretability successes in bioinformatics. Dr. Laviolette’s research has also led to significant advances in the understanding of deep neural networks, with his most important contribution being the domain adversarial neural network. As part of the Chair’s research, the interpretability of machine learning models will be explored from new angles: specifically, by exploring new interpretable rule-based models, by establishing the basis for interpretability in neural networks and by developing the networks’ potential to learn data transformations to enhance their interpretability.
The Chair’s program will provide Intact Financial Corporation with the necessary technological means to tackle the big data revolution. Automating insurance operations with machine learning is likely to be the industry standard of the future. With increasing international competition and the arrival of new fully virtual players such as Lemonade and Tangerine, it is now crucial for Canadian insurers to modernize their practices, which is one of Intact Financial Corporation’s main motivations. Yet, the adoption and usefulness of machine learning methods in many of Intact Financial Corporation’s business processes require a certain degree of transparency at the model level. For example, risk modelling with an efficient, but opaque, model is already very impressive because it can achieve excellent performance. On the other hand, understanding how the model relates risk factors to achieve this performance brings additional value from the point of view of business intelligence. Intact Financial Corporation’s commitment to addressing these issues and to advancing machine learning in insurance recently resulted in the creation of DataLab, a group whose mandate includes supporting the Chair by sharing its experience in insurance data.
The proposed research on machine learning interpretability will not only contribute to improving the application of machine learning methods in the insurance industry, but it will also contribute to the development of more interpretable artificial intelligence applications in many other sectors, such as medical diagnostics, where the understanding and validation of predictions is absolutely necessary. The issue of more interpretable machine learning is thus fundamental in the adoption of critical decision systems based on machine learning algorithms. Many researchers around the world have recently focused their work on this issue. By expanding the expertise available in Canada to include interpretable machine learning, the advances made within the Chair’s program will have an impact on several fields of application, thus strengthening Canada’s position as a world leader in artificial intelligence, machine learning and big data.
Partner
- Intact Financial Corporation
Contact information
Department of Computer Science and Software Engineering
Université Laval
