Q-MMR introduces recursive reweighting and moment matching for off-policy evaluation, delivering dimension-free error bounds under Q^π realizability alone.
In their bound, the coverage term is supf∈F |E(s,a)∼dπ h [fh(s, a)]|/ q ED[ 1 H PH h=1 fh(s, a)2]
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Q-MMR: Off-Policy Evaluation via Recursive Reweighting and Moment Matching
Q-MMR introduces recursive reweighting and moment matching for off-policy evaluation, delivering dimension-free error bounds under Q^π realizability alone.