Monte Carlo Methods in RL

Trajectory Sampling

Process of generating complete episodes by following a given policy until reaching a terminal state. The collected trajectories serve as the basis for Monte Carlo estimates of state or action values.

← Quay lại