Non-convex, non-smooth and stochastic optimization
Optimization algorithms for machine learning
Convergence analysis
Deep learning
Brief Bio:
Camille Castera received the M.Sc. degree in mathematics from INSA Toulouse in France in 2018. He then obtained the Ph.D. degree in applied mathematics in 2021, from the University of Toulouse, working at the IRIT laboratory under the supervision of Cédric Févotte, Edouard Pauwels and Jérôme Bolte. His Ph.D. thesis focused on the design and the analysis of second-order optimization algorithms for training deep neural networks. He is now a postdoctoral researcher in the Mathematical Optimization Group at the University of Tübingen, and a member of the TRINOM-DS project, working with Peter Ochs and Jalal Fadili. His research focuses on second-order optimization algorithms for non-convex and non-smooth machine learning problems.