Big day. Policy Gradient.New ConceptsApproximate Policy Evaluation and Roll-out utilityPolicy Optimization methods:Local Policy Search (aka Hooke-Jeeves Policy Search)Genetic Policy SearchCross Entropy MethodPolicy Gradient, Regression Gradient and Likelyhood Ratio GradientReward-to-GoImportant Results / Claimsmonte-carlo policy evaluationFinite-Difference Gradient EstimationLinear Regression Gradient EstimateQuestions for next office hour