... differene learning DYNA Q eligibility trace Monte Carlo tree search algorithm primitives exploration actor critic reinforcement learning gradient policy gradients Nabla operator sensorimotor interactions estimation finite difference estimation covariance matrix adaptation evolution strategy...