Bridging the Gap between Newton-Raphson Method and Regularized Policy Iteration