Adaptive local likelihood modeling with applications to imaging

Vladimir Spokoiny
Monday, 20.3.2006, 11:30
Taub 001

The talk discusses a novel method of adaptive nonparametric smoothing based on the likelihood modeling in adaptively selected local neighborhoods. The method applies in a unified way to the mean, binary and Poisson regression, density and volatility estimation, classification etc. The performance of the methods is illustrated by a number of numerical examples and applications to image de-noising, Positron Emission Tomography, Functional and Dynamic MRI.

Back to the index of events