An Asynchronous Parallel Stochastic Coordinate Descent Algorithm

Abstract

We describe an asynchronous parallel stochastic coordinate descent algorithm for minimizing smooth unconstrained or separably constrained functions. The method achieves a linear convergence rate on functions that satisfy an essential strong convexity property and a sublinear rate (1/K) on general convex functions. Near-linear speedup on a multicore system can be expected if the number of processors is O(n^1/2) in unconstrained optimization and O(n^1/4) in the separable-constrained case, where n is the number of variables. We describe results from implementation on 40-core processors.

Cite

Text

Liu et al. "An Asynchronous Parallel Stochastic Coordinate Descent Algorithm." International Conference on Machine Learning, 2014.

Markdown

[Liu et al. "An Asynchronous Parallel Stochastic Coordinate Descent Algorithm." International Conference on Machine Learning, 2014.](https://mlanthology.org/icml/2014/liu2014icml-asynchronous/)

BibTeX

@inproceedings{liu2014icml-asynchronous,
  title     = {{An Asynchronous Parallel Stochastic Coordinate Descent Algorithm}},
  author    = {Liu, Ji and Wright, Steve and Re, Christopher and Bittorf, Victor and Sridhar, Srikrishna},
  booktitle = {International Conference on Machine Learning},
  year      = {2014},
  pages     = {469-477},
  volume    = {32},
  url       = {https://mlanthology.org/icml/2014/liu2014icml-asynchronous/}
}