`local' vs. `global' parameters -- breaking the gaussian complexity barrier
Machine Learning
2015-04-10 v1 Statistics Theory
Statistics Theory
Abstract
We show that if is a convex class of functions that is -subgaussian, the error rate of learning problems generated by independent noise is equivalent to a fixed point determined by `local' covering estimates of the class, rather than by the gaussian averages. To that end, we establish new sharp upper and lower estimates on the error rate for such problems.
Keywords
Cite
@article{arxiv.1504.02191,
title = {`local' vs. `global' parameters -- breaking the gaussian complexity barrier},
author = {Shahar Mendelson},
journal= {arXiv preprint arXiv:1504.02191},
year = {2015}
}