A Deep Reinforcement Learningn agent implemented in python using Keras with TensorFlow as backend. This algorithm can be found in the Playing Atari with Deep Reinforcement Learning paper
The code runs on Python 3.6 and uses the following modules
Modules can be installed using the requirements.txt file
-
DQNagent.py-The file contains the Deep Q netwrok agent object.
-
Env.py-This file runs the OpenAI gym simulation and calls the DQN class.
agent.memory(100000)
batch = 64 #trainig batch size
eph = 0.9 #ephsilon
eph_min = 0.01
decay = 0.995 #decay for ephsilon
gamma = 0.99 # discount factor
Cartpole environment