Avatar

Pierre Gloaguen

Assitant professor

Agroparistech

About me

After obtaining my PhD in statistical modelling at the French research instiute for the sea (IFREMER) in 2015, I have been an assitant professor in Statistics at Agroparistech since 2018.

My teaching mainly focuses on statistical modelling for experimental sciences, starting with the linear model and its extensions, and ending with Markov processes for models in biology and ecology. My research focuses on designing inference algorithms for statistical learning. These algorithm are designed for continuous time models (stochastic differential equations) and state space models (hidden Markov models).

Interests

  • Statistical modelling in movement ecology;
  • Inference for Hidden Markov Models;
  • Inference for partially observed diffusion processes;

Education

  • PhD in statistical modelling, 2012-2015

    IFREMER and Agrocampus-Ouest

Recent Publications

February 2020 Submitted in ICML

Backward continuous importance sampling smoother for partially observed diffusion processes

In this article, we propose a sequential Monte Carlo estimator for generic partially observed diffusion processes. We apply our estimator to multivariate stochastic differential equations for which no competing methods exist. We show that our algorithm shows good computing performance.

PDF Link to the preprint
January 2020 Submitted in Bernoulli

Online pseudo Marginal Sequential Monte Carlo smoother for general state spaces.

In this article, we proove consistency and CLT for general (not only bootstrap) particle smoother (when the transition density can only be approximated). We show the interest of the method on online recursive MLE for partially observed diffusion processes.

PDF Link to the preprint