Learning-Augmented Algorithms for Online Steiner Tree

Authors

  • Chenyang Xu Zhejiang University
  • Benjamin Moseley Carnegie Mellon University

DOI:

https://doi.org/10.1609/aaai.v36i8.20854

Keywords:

Machine Learning (ML)

Abstract

This paper considers the recently popular beyond-worst-case algorithm analysis model which integrates machine-learned predictions with online algorithm design. We consider the online Steiner tree problem in this model for both directed and undirected graphs. Steiner tree is known to have strong lower bounds in the online setting and any algorithm’s worst-case guarantee is far from desirable. This paper considers algorithms that predict which terminal arrives online. The predictions may be incorrect and the algorithms’ performance is parameterized by the number of incorrectly predicted terminals. These guarantees ensure that algorithms break through the online lower bounds with good predictions and the competitive ratio gracefully degrades as the prediction error grows. We then observe that the theory is predictive of what will occur empirically. We show on graphs where terminals are drawn from a distribution, the new online algorithms have strong performance even with modestly correct predictions.

Downloads

Published

2022-06-28

How to Cite

Xu, C., & Moseley, B. (2022). Learning-Augmented Algorithms for Online Steiner Tree. Proceedings of the AAAI Conference on Artificial Intelligence, 36(8), 8744-8752. https://doi.org/10.1609/aaai.v36i8.20854

Issue

Section

AAAI Technical Track on Machine Learning III