I am an alumni from the École Normale Supérieure de Cachan, near Paris. I graduated from UC Berkeley with a PhD in Statistics in January 2009. My PhD advisor at Berkeley was Pr. Michael Jordan. After my PhD, I was a post-doc in the WILLOW group. After my post-doc I had a researcher position in the WILLOW and then in the SIERRA group group until December 2012.
My field is Machine Learning and I am interested by its many applications in various application domains. Over time I have collaborated on projects in computer vision, text modelling, computational biology and brain imaging. Nowadays, I am interested by urban modelling and geomatics.
On the more theoretical side, my research focusses on sparsity and the problem of large scale variable selection, that is the problem of learning models from a potentially large number of indicators, predictors, descriptors or variables, when it is assumed that only a small number of them are actually necessary for the learning problem considered. This so-called sparsity assumption is crucial to be able to learn models from limited amount of data. I have worked on formulations for sparsity and sparsity with specific structure as convex problems, both on the optimization and on the statistical side. I try to use structured sparsity to explore more efficiently large feature spaces, and to uncover the latent structure of signals.
I am also interested by optimization techniques for large scale learning, matrix learning, kernel methods and graphical models techniques such as Markov random fields and Bayesian networks.