Locally robust methods and near-parametric asymptotics (Q1661369)

It is known that the localization of likelihood based methods for regression and density estimation improve the resulting estimators with respect to suitable global risk measures. The authors show that a similar effect can also be observed with respect to robust estimation procedures. They prove that the localized versions of robust density estimation methods perform better with respect to global risk measures based on minimization of Bregman divergence measures.