Christophe Kervazo

Job offers:

• PhD subject about hyperspectral unmixing with Plug-and-Play methods

About me

I received Supélec (France) engineering degree in 2015, and the master of science in Electrical and Computer Engineering from Georgia Institute of Technology (USA) in 2016. From 2016 to 2019, I was a PhD student in the CosmoStat group at CEA Saclay (France), where I worked on the optimization framework for sparse blind source separation, as well as non-linear component separation. I then went for one year in Mons (Belgium) where I worked, as a post-doctoral researcher, on the extension of Nonnegative Matrix Factorization to Linear-Quadratic mixture unmixing.

I am currently an Assistant Professor (maître de conférences) at Télécom Paris (France), in the IMAGES group. I work in two main applicative fields: remote sensing and medical imaging. I focus on interpretable deep learning methods, with an emphasis on several aspects: 1) interpretability of the neural architectures ; 2) uncertainty quantification ; 3) controlled synthetic training set generation.

Research interests

• Deep learning for inverse problems (including in particular works about deep unrolling techniques)

• Remote sensing (hyperspectral imaging and synthetic aperture radar) and medical imaging (tomosynthesis, bioluminescene imaging)

• Uncertainty quantification

• Non-linear component separation methods

• Synthetic training set generation

• Blind Source Separation: sparsity-based approaches and (provably robust) Nonnegative Matrix Factorization

• Optimization: hyper-parameter choice and scalable multi-convex optimization

Contact

Télécom ParisTech,

19 Place Marguerite Perey, 91120 Palaiseau (France)

christophe.kervazo@telecom-paris.fr