TD variance is asymptotically upper-bounded by Monte Carlo estimators with shorter horizons lowering variance, while Direct Advantage Estimation achieves tighter bounds as a regression-adjusted control variate.
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
UNVERDICTED 1representative citing papers
citing papers explorer
-
On the Variance of Temporal Difference Learning and its Reduction Using Control Variates
TD variance is asymptotically upper-bounded by Monte Carlo estimators with shorter horizons lowering variance, while Direct Advantage Estimation achieves tighter bounds as a regression-adjusted control variate.