Skip to Main content Skip to Navigation
Conference papers

Asymptotically Optimal Strategies for Combinatorial Semi-Bandits in Polynomial Time

Thibaut Cuvelier 1 Richard Combes 1 Eric Gourdin 2
1 CentraleSupélec
2 Orange Gardens
Abstract : We consider combinatorial semi-bandits with uncorrelated Gaussian rewards. In this article, we propose the first method, to the best of our knowledge, that enables to compute the solution of the Graves-Lai optimization problem in polynomial time for many combinatorial structures of interest. In turn, this immediately yields the first known approach to implement asymptotically optimal algorithms in polynomial time for combinatorial semi-bandits.
Keywords : Reinforcement learning Bandit algorithms Combinatorial optimisation Approximation algorithms
Document type :
Conference papers
Complete list of metadata
https://hal.archives-ouvertes.fr/hal-03162140
Contributor : Thibaut Cuvelier Connect in order to contact the contributor
Submitted on : Monday, March 8, 2021 - 12:12:31 PM
Last modification on : Friday, March 12, 2021 - 3:23:33 AM
Long-term archiving on: : Wednesday, June 9, 2021 - 6:55:22 PM

File

Vignette du document
2102.07254.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-03162140, version 1
  • ARXIV : 2102.07254

Collections

INSMI | TDS-MACS | CENTRALESUPELEC

Citation

Thibaut Cuvelier, Richard Combes, Eric Gourdin. Asymptotically Optimal Strategies for Combinatorial Semi-Bandits in Polynomial Time. Algorithmic Learning Theory, Mar 2021, Paris, France. ⟨hal-03162140⟩

Export

BibTeX TEI DC DCterms EndNote Datacite

Share

Metrics

Record views

22

Files downloads

11