University of Cambridge > > Probability Theory and Statistics in High and Infinite Dimensions > A robust and adaptive estimator for regression I

A robust and adaptive estimator for regression I

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact clc32.

Our purpose is to present a new method for adaptively estimating a regression function when little is known about the shape and scale of the errors. For instance, it can cope with error distributions as different as Gaussian, Uniform, Cauchy or even with unimodal unbounded densities. In favorable cases and when the true distribution belongs to the model, the estimator is asymptotically equivalent to the M.L.E. and, nevertheless, still behaves reasonably well when the model is wrong, even in cases for which the least-squares do not work. The assumptions that are needed to get our results are rather weak, in particular no moment condition is required on the errors, and this is why the method can adapt to both the regression function, the shape of the errors and their scale. Moreover, it appears that the practical results obtained by simulation are surprisingly good as compared to more specific estimators. The corresponding paper is available on arXiv at Joint work with Mathieu Sart.

This talk is part of the Probability Theory and Statistics in High and Infinite Dimensions series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2023, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity