Introduces (α,β,δ,N)-resilience for bi-criteria approximations and a black-box offline-to-online reduction yielding Õ(δ^{2/3} N^{1/3} T^{2/3}) regret and cumulative constraint violation in combinatorial bandits without linearity, submodularity, or semi-bandit assumptions.
Title resolution pending
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
-
A Resilience Framework for Bi-Criteria Combinatorial Optimization with Bandit Feedback
Introduces (α,β,δ,N)-resilience for bi-criteria approximations and a black-box offline-to-online reduction yielding Õ(δ^{2/3} N^{1/3} T^{2/3}) regret and cumulative constraint violation in combinatorial bandits without linearity, submodularity, or semi-bandit assumptions.