Robust estimation of a regression function in exponential families

We observe pairs of independent (but not necessarily i.i.d.) random variables and tackle the problem of estimating the conditional distributions of given for all . Even though these might not be true, we base our estimator on the assumptions that the data are i.i.d.\ and the conditional distributions of given belong to a one parameter exponential family with parameter space given by an interval . More precisely, we pretend that these conditional distributions take the form for some that belongs to a VC-class of functions with values in . For each , we estimate by a distribution of the same form, i.e.\ , where is a well-chosen estimator with values in . We show that our estimation strategy is robust to model misspecification, contamination and the presence of outliers. Besides, we provide an algorithm for calculating when is a VC-class of functions of low or moderate dimension and we carry out a simulation study to compare the performance of to that of the MLE and median-based estimators.
View on arXiv