MachineLearning/dqn library

Deep Q-Network (DQN) - minimal numeric example

A compact DQN wrapper that uses the project's ANN MLP as a function approximator. This implementation is intentionally small and educational — it demonstrates replay memory, epsilon-greedy actions, and periodic updates so it can be used in examples and unit tests without GPU dependencies.

Notes:

• States are represented as List
• Actions are integer indices

Classes

DQN

DQN wrapper with optional target network and configurable replay buffer