Asymptotic exponential bounds for MLE deviation under minimal conditions via classical and generic chaining methods
Probability
2009-03-25 v1
Abstract
In this paper non-asymptotic exact exponential estimates are derived (under minimal conditions) for the tail of deviation of the MLE distribution in the so-called natural terms: natural function, natural distance, metric entropy, Banach spaces of random variables, contrast function, majorizing measures or, equally, generic chaining.
Cite
@article{arxiv.0903.4062,
title = {Asymptotic exponential bounds for MLE deviation under minimal conditions via classical and generic chaining methods},
author = {E. Ostrovsky and E. Rogover},
journal= {arXiv preprint arXiv:0903.4062},
year = {2009}
}