A0567
Title: Minimax geometric inference: Open and closed problems
Authors: Catherine Aaron - Universite Clermont Auvergne (France) [presenting]
Abstract: An overview of minimax set and manifold estimators are presented from convex sets to stratified manifold estimators. The convergence of the Hausdorff distance between the set and its estimator is first discussed. Another way is proposed to measure the error in set estimation, such as Wasserstein distance or integral probability metric. The more challenging problem of minimax volume estimation and the still open problem of minimax perimeter estimation are also discussed.
