Research

Interests

I am interested in mathematical statistics, probability of random dynamical systems and applied mathematical modelling.

  • Stochastic algorithms: modern issues in Big Data are commonly dealt with on-line optimization algorithm. Kurdyka-Lojasiewicz inequality
  • Nonparametric statistics: minimax and adaptive estimation/classification from a frequentist point of view. Nonparametric Bayes. Inverse Problems
  • High dimensional problems: variable selection, sparse recovery, algorithms of model selection and optimal sampling, Sequential prediction.
  • Stochastic processes: Ergodic results, trajectorial properties, large deviations, optimization.
  • Interface: mathematical modelling, self interacting behaviour, memory differential equations, applications of random models to biology, genetics, finance, behavioral sciences. Data science applied to other fields of research

Grants

 

  • Head of the project COSAL Combining Optimization and Stochastic Algorithms for large scale Learning. (2018-2019) Funded by the fondation Jacques Hadamard. 40 K€.
  • Head of the math wp of the project COMPUTREAT Ibrutinib effects on CL leukemia and follicular lymphoma : random modelling, statistical inference and forecasting. (2016-2019) Funded by Plan Cancer – Systems biology – 20 K€ + 1 Post Doc (80 k€).
  • Head of the project TOXIMATH, TOXIcity and MATHematics of immune system. (2014-2017) Funded by the Fondation Recherche Innovation Thérapeutique Cancérologie (Plan Cancer – Oncopole ). 20 K€ + 1 post doc (80 k€).
  • I was the Principal Investigator of the ANR Demos project. (2010-2014)

This project concerns some research on DEformable MOdels in Statistics for signal and image analysis. It is funded by the Programme Jeunes Chercheurs of the Agence Nationale de la Recherche (ANR). This project involves several researchers from the Institut de Mathématiques de Toulouse and the University Paul Sabatier. The project members are Jérémie Bigot, Clément Marteau and Jean-Michel Loubes.

  • Patrick Cattiaux and I obtained an « AO3″ dissertation grand, as well as several funds  with a multidisciplinary collaboration with the Dr Salvatore Valitutti Inserm team. The project aims to provide a sharp modelisation of the predator/preys system of tumoral vs CTL interaction. The PhD thesis is 50% host atthe Inserm Laboratory, and 50% at the IMT.
  • I intensely worked with Thalès (Thales Alenia Space) and the CNES on a project of Satellite guided Navigation. Our project sharply callibrated several error measurements of the Galileo system using Extreme Value Theory. We first built a supervised algorithm with EVT to estimate accuracy of the system in 2009. In 2010, we used a rare event simulation method of splitting and compared this approach with Petri’s network. The last year (2011), we have developed an unsupervised automatic detection method of integrity loss. This last algorithm is now in charge on the Galileo system (oO! ;)). The project members were Jean-Marc Azaïs, Agnès Lagnoux and Cécile Mercadier.