Q-Learning Basics

Q-learning is an off-policy method that iteratively updates action-value estimates toward the max.

What does Q-learning aim to learn?