Kernel-WIS kernel-localizes weighted importance sampling and is claimed to improve off-policy evaluation under behavior-policy misspecification, though it degrades on continuous rewards.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Kernel weighted importance sampling for off-policy evaluation in contextual bandits
Kernel-WIS kernel-localizes weighted importance sampling and is claimed to improve off-policy evaluation under behavior-policy misspecification, though it degrades on continuous rewards.