Statistical Sufficiency for Classes in Empirical L2 Spaces
Abstract
We explore the notion of " sufficient linear statistics for a class of real valued functions. We show that for function classes with a polynomial rate of the Parametric Pollard dimension one can find a set of linear empirical functionals of polynomial size in the dimension that are sufficient for " approximation of any function in the class. We also present a probabilistic scheme for producing those functionals.
Cite
Text
Mendelson and Tishby. "Statistical Sufficiency for Classes in Empirical L2 Spaces." Annual Conference on Computational Learning Theory, 2000.Markdown
[Mendelson and Tishby. "Statistical Sufficiency for Classes in Empirical L2 Spaces." Annual Conference on Computational Learning Theory, 2000.](https://mlanthology.org/colt/2000/mendelson2000colt-statistical/)BibTeX
@inproceedings{mendelson2000colt-statistical,
title = {{Statistical Sufficiency for Classes in Empirical L2 Spaces}},
author = {Mendelson, Shahar and Tishby, Naftali},
booktitle = {Annual Conference on Computational Learning Theory},
year = {2000},
pages = {81-89},
url = {https://mlanthology.org/colt/2000/mendelson2000colt-statistical/}
}