#stability

Regularized Emphatic Temporal-Difference Learning: StabilityNEW⭐9

Regularized emphatic TD (RETD) restores stability of off-policy TD learning under constant stepsizes, with convergence guarantees and 10,000-run experiments val...