Nicolas Le Roux

Nicolas Le Roux is an Associate Professor at the Université de Montréal, affiliated with the Department of Computer Science and Operations Research within the Faculty of Arts and Science. He is also an adjunct faculty member at McGill University and part of the Mila – Quebec Institute of Artificial Intelligence. His research focuses on machine learning, optimization, and statistical modeling, particularly in supervised learning and reinforcement learning from an optimization perspective. He has contributed to foundational work on policy gradient methods and surrogate functions in reinforcement learning. Recently, he explores model adaptation, modular architectures, and decentralized AI systems.

Le Roux is a co-recipient of the 2018 Lagrange Prize and a CIFAR AI Chair. His work bridges technical advancements with organizational practices, emphasizing team well-being and inclusive research environments. He has advised three PhD students (Alan Chan, Reyhane Askari Hemmat, and Valentin Thomas) on topics like generative models and optimization in AI agents. Though currently on a career break, he remains active in research and maintains collaborations through Mila.

Key research themes include optimizing policy functions over parameters, modular neural architectures, and ethical AI governance. His recent papers address operator-based policy gradients, decision-aware actor-critic algorithms, and adapter routing for cross-task generalization. Le Roux’s interdisciplinary approach integrates technical innovation with societal impact, advocating for equitable and sustainable AI development.