Exercise 6.3.2
On-policy vs off-policy control: data efficiency and exploration
Reinforcement Learning › Exercises › Lecture 6 · Function Approximation › 6.3 Control with value-function approximation
Contrast the learning process of an on-policy control method (e.g., Semi-gradient SARSA) with an off-policy one (e.g., Q-learning with function approximation). Discuss the implications of this difference for data efficiency and exploration.