A Contextual Bandit Bake-off

Abstract : Contextual bandit algorithms are essential for solving many real-world interactive machine learning problems. Despite multiple recent successes on statistically and computationally efficient methods, the practical behavior of these algorithms is still poorly understood. We leverage the availability of large numbers of supervised learning datasets to compare and empirically optimize contextual bandit algorithms, focusing on practical methods that learn by relying on optimization oracles from supervised learning. We find that a recent method (Foster et al., 2018) using optimism under uncertainty works the best overall. A surprisingly close second is a simple greedy baseline that only explores implicitly through the diversity of contexts, followed by a variant of Online Cover (Agarwal et al., 2014) which tends to be more conservative but robust to problem specification by design. Along the way, we also evaluate and improve several internal components of contextual bandit algorithm design. Overall, this is a thorough study and review of contextual bandit methodology.
Document type :
Preprints, Working Papers, ...
Complete list of metadatas

Cited literature [33 references]  Display  Hide  Download

https://hal.inria.fr/hal-01708310
Contributor : Alberto Bietti <>
Submitted on : Wednesday, May 30, 2018 - 11:12:52 PM
Last modification on : Friday, January 4, 2019 - 4:49:19 PM
Long-term archiving on : Friday, August 31, 2018 - 6:57:49 PM

File

main.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01708310, version 2
  • ARXIV : 1802.04064

Citation

Alberto Bietti, Alekh Agarwal, John Langford. A Contextual Bandit Bake-off. 2018. ⟨hal-01708310v2⟩

Share

Metrics

Record views

253

Files downloads

713