TensorLearn
Back to Course
Reinforcement Learning: Agents
Module 3 of 8

3. Deep Q-Networks (DQN)

1. The Table is too big

For a video game, the "State" is pixels. 256^ScreenSize. We cannot use a table. We use a Neural Network to approximate the function $Q(s,a)$.

Mark as Completed

TensorLearn - AI Engineering for Professionals