Combinatorial Semi-Bandits with Knapsacks
Machine Learning
2018-02-22 v3
Abstract
We unify two prominent lines of work on multi-armed bandits: bandits with knapsacks (BwK) and combinatorial semi-bandits. The former concerns limited "resources" consumed by the algorithm, e.g., limited supply in dynamic pricing. The latter allows a huge number of actions but assumes combinatorial structure and additional feedback to make the problem tractable. We define a common generalization, support it with several motivating examples, and design an algorithm for it. Our regret bounds are comparable with those for BwK and combinatorial semi- bandits.
Cite
@article{arxiv.1705.08110,
title = {Combinatorial Semi-Bandits with Knapsacks},
author = {Karthik Abinav Sankararaman and Aleksandrs Slivkins},
journal= {arXiv preprint arXiv:1705.08110},
year = {2018}
}