Active Context Selection Improves Simple Regret in Contextual Bandits

Mohammad Shahverdikondori, Jalal Etesami, Negar Kiyavash · 2026

Read Active Context Selection Improves Simple Regret in Contextual Bandits by Mohammad Shahverdikondori, Jalal Etesami, Negar Kiyavash (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-20040.txt.

Description

Mohammad Shahverdikondori, Jalal Etesami and Negar Kiyavash study contextual bandits where the learner may choose which to sample. With a context distribution, passive sampling gives slower regret than active sampling with allocation by p to the two-thirds power. The improvement can reach order k to the one-quarter power, where k counts contexts. They extend the analysis to budgeted active sampling, identify when a limited budget recovers the active rate, and propose an Explore-Explore-Then-Commit algorithm for unknown distributions matching the active rate for large horizons.

Catalog record

CREATORMohammad Shahverdikondori, Jalal Etesami, Negar Kiyavash
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE94632
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.20040

Browse the full library · About the Tower of Babel