Developing contemporary mathematical statistics
The objective of the group is to develop new robust substitutes to classical statistical methods such as maximum likelihood, least squares and Bayesian methods.
Robust strategies
Classical statistical methods are designed to provide good parameter estimates when the statistical model is exact. When it is approximate, these estimators may perform poorly. Unfortunately, what we can expect is that the model we use only provides a suitable approximation of reality. Our group aims at designing reliable procedures in the real situation where the model may only be approximate.
