Function Approximation
摘要
For algorithms in Chaps. 3–5, each update can only update the value estimate for one state or one state–action pair. Unfortunately, state spaces or action spaces can be very large in some tasks, even can be infinitely large.