Learning In Mean-Field Games And Continuous-Time Stochastic Control Problems