DeepRL1.4A, Eligibility traces for policy gradient and actor-critic

views comments

Policy-gradient learning and actor-critic architectures can also be combined with eligibiligy traces. This leads to an elegant online learning rule.