Home▸Machine Learning & Neural Networks▸Deep Q-Network: Learning to Play from Pixels

🎮 Deep Q-Network: Learning to Play from Pixels

Explore how DeepMind's Deep Q-Network combines Q-learning with neural networks, experience replay and a target network to master Atari games directly from raw pixels.

Machine Learning & Neural Networks3DModerate60 FPS

The 3D simulation visualises an agent's neural network estimating Q-values for each possible action as it explores a simple environment, alongside its experience replay buffer filling and its target network trailing the main network's weights.

🔬 What It Demonstrates

🎮 How to Use

Pick an environment and training speed, then press play to watch the agent explore, sample batches from replay memory, and periodically sync its target network; pause anytime to inspect the current Q-value estimates.

💡 Did You Know?

DeepMind's original DQN agent was trained on the exact same network architecture and hyperparameters across all 49 Atari games, with no game-specific tuning, yet still outperformed a professional human tester on the majority of them.

⚙ Under the hood

Interactive 3D lab where an agent's neural network estimates Q-values while exploring a grid world, filling an experience replay buffer and syncing a target network.

deep-q-networkreinforcement-learningexperience-replayq-learningdeep-learningatari

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install