Safe-Private Lin-UCB balances local differential privacy against safety in coupled linear bandits, with a provably unilaterally unimprovable privacy allocation for simplex safety sets under a regret budget.
Correlational dueling bandits with application to clinical treatment in large decision spaces,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Safety-Privacy Tradeoff in Linear Bandits
Safe-Private Lin-UCB balances local differential privacy against safety in coupled linear bandits, with a provably unilaterally unimprovable privacy allocation for simplex safety sets under a regret budget.