M24.5 CONNECT THE MECHANISM

Approximate action values with a neural network

Give Pip a camera and its Q-table explodes. Swap the table for a neural network, and learn the three tricks that let DQN master Atari games from pixels.

LESSON OVERVIEW14 min lesson

Lesson overview

Give Pip a camera and its Q-table explodes. Swap the table for a neural network, and learn the three tricks that let DQN master Atari games from pixels.

What you’ll explore

DQN combines neural action-value prediction with replay and delayed target networks; these mechanisms reduce some instability while leaving exploration, distribution shift, and approximation errors to evaluate.

GO TO THE SOURCE

Original explanations, connected to the research.

Human-level control through deep reinforcement learning (Mnih et al., Nature 2015)Playing Atari with Deep Reinforcement Learning Deep Reinforcement Learning with Double Q-learning

Suggest a correction

A precise note can make an explanation better.

Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.