Le lundi 31 janvier 2011 à 15:00 - UM2 - Bât 09 - Salle 331 (3ème étage)Sophie Dabo-Niang
We are interested in estimating the regression function non-parametrically when the data are measured with error. We study the behavior of the kernel density and regression estimators when the data are noisy. This is less standard than classical density and regression estimation problems. We applied these estimations to improve time series prediction in some situations and case of finite dimensional or functional data.