Robust Estimation, Random Matrix Theory and Applications to Signal Processing
Frederic Pascal
CentraleSupelec, Paris

This talk deals with general problems of covariance matrix estimation with applications in signal processing. Assuming random interferences (noise, outliers, missing data¿), we focus on different problems of detection and estimation: detection of an object of interest and estimation of its parameters. Nowadays, the amount of observations or data is bigger and bigger, due to the improvement of technology or due to the use of new data. In the general statistical framework of robust estimation theory, we will show how classical results can be extended to high dimensional data by using the Random Matrix Theory for large dimensional problems.