Exercise 5.5.1
Maximization bias: why Q-learning overestimates
Reinforcement Learning › Exercises › Lecture 5 · Model-Free Control › 5.5 Maximization bias and double Q-learning
Q-learning updates its Q-values using an update rule that can sometimes lead to an overestimation of Q-values. What is the primary reason why Q-learning suffers from this overestimation problem?