Contexte et atouts du poste
The PhD thesis will take place at the Université Grenoble Alpes in the Inria-AIRSEA team. This project is funded by Numpex, the French exascale supercomputing program (
Mission confiée
Bayesian optimal sensor placement is critical in various applications, particularly in scenarios where data acquisition is expensive (satelite observation, buoys in the ocean, underground drill etc). The primary challenge lies in determining the optimal locations where to observe the system in order to best inference a specific parameter of interest. While linear models and Gaussian priors are well-understood and relatively straightforward to handle, the problem becomes significantly more complex when dealing with models that are numerically costly to evaluate. This is especially true for large-scale, nonlinear and nonGaussian systems for which evaluating the numerical model is prohibitively expensive.
Recently, a gradient-based approach has been proposed to alleviate this computational burden. The strategy behind this approach is to minimize a bound of the so-called Expected Information Gain (EIG), which is relatively easy to work with, rather than minimizing the EIG itself. In principle, this bound serves as a surrogate for the EIG which providing a computationally favorable way to guide the sensor placement. This is because the error-bound can be evaluated and optimized much more efficiently than the actual error, which requires numerous expensive numerical simulations of the numerical model.
The objective of this project is to address various numerical aspects associated with the gradient-based solution for the Bayesian optimal sensor placement problem. The project has three main goals:
Avantages
Rémunération
1st and 2nd year: 2100 euros gross salary /month
3rd year: 2190 euros gross salary / month
