Approximability of Probability Distributions
Abstract
We consider the question of how well a given distribution can be approx- imated with probabilistic graphical models. We introduce a new param- eter, effective treewidth, that captures the degree of approximability as a tradeoff between the accuracy and the complexity of approximation. We present a simple approach to analyzing achievable tradeoffs that ex- ploits the threshold behavior of monotone graph properties, and provide experimental results that support the approach.
Cite
Text
Beygelzimer and Rish. "Approximability of Probability Distributions." Neural Information Processing Systems, 2003.Markdown
[Beygelzimer and Rish. "Approximability of Probability Distributions." Neural Information Processing Systems, 2003.](https://mlanthology.org/neurips/2003/beygelzimer2003neurips-approximability/)BibTeX
@inproceedings{beygelzimer2003neurips-approximability,
title = {{Approximability of Probability Distributions}},
author = {Beygelzimer, Alina and Rish, Irina},
booktitle = {Neural Information Processing Systems},
year = {2003},
pages = {377-384},
url = {https://mlanthology.org/neurips/2003/beygelzimer2003neurips-approximability/}
}