Learning Augmented Energy Minimization via Speed Scaling
Abstract
As power management has become a primary concern in modern data centers, computing resources are being scaled dynamically to minimize energy consumption. We initiate the study of a variant of the classic online speed scaling problem, in which machine learning predictions about the future can be integrated naturally. Inspired by recent work on learning-augmented online algorithms, we propose an algorithm which incorporates predictions in a black-box manner and outperforms any online algorithm if the accuracy is high, yet maintains provable guarantees if the prediction is very inaccurate. We provide both theoretical and experimental evidence to support our claims.
Cite
Text
Bamas et al. "Learning Augmented Energy Minimization via Speed Scaling." Neural Information Processing Systems, 2020.Markdown
[Bamas et al. "Learning Augmented Energy Minimization via Speed Scaling." Neural Information Processing Systems, 2020.](https://mlanthology.org/neurips/2020/bamas2020neurips-learning/)BibTeX
@inproceedings{bamas2020neurips-learning,
title = {{Learning Augmented Energy Minimization via Speed Scaling}},
author = {Bamas, Etienne and Maggiori, Andreas and Rohwedder, Lars and Svensson, Ola},
booktitle = {Neural Information Processing Systems},
year = {2020},
url = {https://mlanthology.org/neurips/2020/bamas2020neurips-learning/}
}