value-functions

Temporal-difference learning: learning from a better prediction before the outcome arrives Lesson

Temporal-difference learning updates a value estimate using the next reward and next prediction, allowing an agent to learn before an episode ends.