RETD regularizes the auxiliary centering recursion in emphatic TD to fix the positive-definiteness loss from naive centering, derives the core matrix, proves convergence under a sufficient condition on the regularization parameter, and shows stable performance on linear off-policy diagnostics.
Gradient temporal-difference learning with regularized corrections
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Regularized Centered Emphatic Temporal Difference Learning
RETD regularizes the auxiliary centering recursion in emphatic TD to fix the positive-definiteness loss from naive centering, derives the core matrix, proves convergence under a sufficient condition on the regularization parameter, and shows stable performance on linear off-policy diagnostics.