Camille Castera

Camille Castera is an Associate Professor in optimization and machine learning at the Mathematical Institute of the University of Bordeaux (IMB). Her academic journey includes a postdoctoral position at the University of Tübingen (2022-2024) working on the TRINOM-DS project under Peter Ochs and Jalal Fadili, and a PhD at the IRIT laboratory in Toulouse supervised by Cédric Févotte, Edouard Pauwels, and Jérôme Bolte as part of the FACTORY project.

Her research focuses on optimization algorithms for machine learning, particularly second-order methods with applications to deep learning, and the emerging field of learning optimization algorithms. Her work spans continuous optimization (convex & non-convex, non-smooth, stochastic), learning to optimize, and optimization for deep learning with automatic hyper-parameter tuning.

Her recent publications demonstrate significant contributions to the field of optimization algorithms, particularly in developing novel approaches that bridge classical optimization techniques with learning-based methods. Her research shows a clear trajectory toward creating more adaptive and versatile optimization algorithms that can generalize beyond their training settings.

At the University of Bordeaux, she teaches advanced courses including Large-scale optimization (Master 2), Introduction to deep learning (Master 2), Convex optimization (Master 1), and Image processing (Licence 3), while also supervising student projects on image processing and machine learning.

Her technical work is supported by substantial implementation efforts, with publicly available code repositories demonstrating practical applications of her theoretical contributions to optimization algorithms.