Proposes Slate-GLM-OFU and Slate-GLM-TS algorithms for logistic contextual slate bandits that achieve polynomial per-round time and sublinear regret under a diversity assumption.
Finally, following the same line of thought as Lemma B.8 and Lemma B.9, and using the fact that 1 κ ≤ 1 4, we obtain 3 4 U H t ≼ V H t ≼ 5 4 U H t Lemma B.12
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Efficient Algorithms for Logistic Contextual Slate Bandits with Bandit Feedback
Proposes Slate-GLM-OFU and Slate-GLM-TS algorithms for logistic contextual slate bandits that achieve polynomial per-round time and sublinear regret under a diversity assumption.