Computational limits to nonparametric estimation for ergodic processes
Information Theory
2016-11-18 v2 math.IT
Abstract
A new negative result for nonparametric estimation of binary ergodic processes is shown. I The problem of estimation of distribution with any degree of accuracy is studied. Then it is shown that for any countable class of estimators there is a zero-entropy binary ergodic process that is inconsistent with the class of estimators. Our result is different from other negative results for universal forecasting scheme of ergodic processes.
Keywords
Cite
@article{arxiv.1002.1559,
title = {Computational limits to nonparametric estimation for ergodic processes},
author = {Hayato Takahashi},
journal= {arXiv preprint arXiv:1002.1559},
year = {2016}
}
Comments
submitted to IEEE trans IT