---
title: "Deep Q-Learning (DQN)"
description: "Predict what each action is worth in future reward, build a training target from one transition, and see why DQN keeps a target network and replay buffer."
canonical_url: "https://fanout.sh/ai/lessons/reinforcement-learning-deep-q-learning"
md_url: "https://fanout.sh/ai/lessons/reinforcement-learning-deep-q-learning.md"
access: "pro-preview"
---

# Deep Q-Learning (DQN)

Predict what each action is worth in future reward, build a training target from one transition, and see why DQN keeps a target network and replay buffer.

- Access: Fanout Pro (public preview)
- Module: Reinforcement Learning

## A glimpse of Deep Q-Learning (DQN)

Suppose a game offers two actions. Instead of directly storing their probabilities, an agent predicts how much future reward each action is worth.

Those predictions are Q-values. Q(s, a) means the estimated total of future rewards for taking action a in state s, then choosing later actions according to a particular rule, or policy. That reward total is a return.

## Continue learning

Pro unlocks guided learning, complete notes, practical walkthroughs, exercises, and every AI Research lesson.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md