
Post-Graduate Stories
Marouane Il Idrissi Will Seek to Unlock the “Explainability and Interpretability of Black-box Models”
As a 2025 CANSSI Distinguished Postdoctoral Fellow, Marouane Il Idrissi will take part in a comprehensive program that involves teaching, interdisciplinary or applied collaboration, professional development, and a research project that aims to understand the inner workings of black-box models like deep neural networks. He will work under the supervision of Professor Arthur Charpentier (Université du Québec à Montréal) and Professor Marie-Pier Côté (Université Laval).
Program: CANSSI Distinguished Postdoctoral Fellowship
Region: National
Date: 2025–2027
Project Focus Areas
Explainability and Interpretability are crucial for predictive models because they serve as a bridge between advanced machine learning techniques and their real-world applications. While black-box models like deep neural networks often exhibit remarkable predictive performance, their inner workings can remain inscrutable, making their decisions challenging to trust, explain, and debug. In high-stakes decision contexts such as healthcare, finance and insurance, or autonomous vehicles, the ability to understand these models is paramount for accountability, fairness, and regulatory compliance.
Interpreting black-box models becomes significantly more challenging when predictive features are correlated. Correlation among features can lead to multicollinearity, in which case the unique contribution of each variable to the model’s predictions is difficult to discern. In such cases, interpretability techniques must disentangle the intricate web of feature interactions, further underscoring the importance of robust methods for understanding these complex, real-world machine learning systems.
Marouane Il Idrissi’s research will draw on real-world insurance datasets. For applications, he will use public domain databases as much as possible. During the project, he will receive training in statistical methodology and actuarial science and he will have opportunities to teach and mentor undergraduate and graduate students. In addition, he will have opportunities to help organize workshops and present at national and international conferences.


Getting to Know Marouane
Marouane Il Idrissi completed his PhD in Applied Mathematics at the Université de Toulouse in 2024 under the supervision of Jean-Michel Loubes, Fabrice Gamboa, Nicolas Bousquet, and Bertrand Iooss. His dissertation focused on the development of interpretability methods in machine learning for the certification of artificial intelligence related to critical systems.
The primary objective of this work was to enhance the transparency and accountability of black-box models deployed in high-stakes domains, such as energy production and distribution, where the reliability and safety of artificial intelligence (AI)–driven decisions are crucial.
In his application to the CANSSI Distinguished Postdoctoral Fellowships (CDPF) program, Marouane highlighted his diverse academic journey—from an initial focus on economics to advanced statistics and applied mathematics—and his professional experiences across multiple critical sectors, including insurance, healthcare, and energy, which he credits with deepening his understanding of the real-world implications of AI.
He also expressed a strong interest in teaching and stated, “My goal as an educator is to equip students with the technical skills and ethical awareness needed to navigate the evolving landscape of data-driven decision-making responsibly and effectively.”
Marouane views the CDPF program not only as a research opportunity but also as “a unique chance to achieve my goal of becoming a long-term, prolific member of the Canadian academic community.”
I view this fellowship not only as a research opportunity but also as a unique chance to achieve my goal of becoming a long-term, prolific member of the Canadian academic community.
About the Supervisors
Arthur Charpentier
Arthur Charpentier, PhD, Fellow of the French Institute of Actuaries, is a Full Professor at the Université du Québec à Montréal (UQAM), Montreal, Canada, and Université de Rennes, in France. He is a member of the editorial board of the Journal of Risk and Insurance, the ASTIN Bulletin, and Risks. He edited, a few years ago, Computational Actuarial Science with R (CRC), and more recently wrote Insurance, Biases, Discrimination and Fairness (Springer). He is also the former director of the Data Science for Actuaries program of the French Institute of Actuaries. He is a Louis Bachelier Fellow, and his recent work is about climate change and predictive modelling insurance, more specifically in the context of fairness and discrimination.
He has a PhD in Applied Mathematics from KU Leuven (2006, Belgium) and an HDR (Habilitation) in Applied Mathematics from the Université de Rennes (2016, France).
Marie-Pier Côté
Marie-Pier Côté is an Assistant Professor in the School of Actuarial Science at Université Laval and a Fellow of the Society of Actuaries. She holds an Educational Leadership Chair (ELC) in Big Data Analysis for Actuarial Science – Intact.
Professor Côté received a Master’s degree (2014) and a PhD (2018) in Statistics from McGill University. Her research interests include insurance risk dependence modelling and the development of statistical learning models for pricing and reserves in general insurance. She has co-authored several articles and collaborates closely with partners in the insurance industry in her research and she is a member of the Big Data Research Center at Université Laval.