The Monte Carlo algorithm is used in RL when the model of the environment is not known.
Refer section Estimating the value of pi using Monte Carlo.
In Monte Carlo prediction, we approximate the value function by taking the mean return instead of the expected return.
In every visit Monte Carlo, we average the return every time the state is visited in an episode. But in the first visit MC method, we average the return only the first time the state is visited in an episode.
Refer section Monte Carlo control.
Refer section On-policy Monte Carlo control and Off-policy Monte Carlo control
Refer section Let's play Blackjack with Monte Carlo.
