Communication-efficient sparse regression: a one-shot approach
Machine Learning
2015-08-12 v3 Machine Learning
Abstract
We devise a one-shot approach to distributed sparse regression in the high-dimensional setting. The key idea is to average "debiased" or "desparsified" lasso estimators. We show the approach converges at the same rate as the lasso as long as the dataset is not split across too many machines. We also extend the approach to generalized linear models.
Cite
@article{arxiv.1503.04337,
title = {Communication-efficient sparse regression: a one-shot approach},
author = {Jason D. Lee and Yuekai Sun and Qiang Liu and Jonathan E. Taylor},
journal= {arXiv preprint arXiv:1503.04337},
year = {2015}
}
Comments
29 pages, 3 figures