sinangncgl / deep-q-network-ataribreakoutgame

Playing Atari Breakout Game with Reinforcement Learning (DQN , Deep Q Learning)

Deep-Q-Network-AtariBreakoutGame

Playing Atari Breakout Game with Reinforcement Learning ( Deep Q Learning )

Overview

This project follows the description of the Deep Q Learning algorithm described in this paper.

Installation Dependencies:

• Python 3.x
• Numpy
• OpenCV-Python
• PyGame
• PyTorch

How To Run

• git clone https://github.com/SnnGnc/Deep-Q-Network-AtariBreakoutGame.git
• cd brekout
• To train the game;
• python dqn.py train,
• To test the pre-trained version,
• python dqn.py test

What is Deep Q Learning and How does it works ?

I highly recommend to read this Demystifying Deep Reinforcement Learning who are curious about reinforcement learning.

DQN Algorithm

Future Reward Function = Q

Loss Function

Network Architecture

"Working directly with raw Atari frames, which are 210 × 160 (in our case it depends on pygame screen) pixel images with a 128 color palette, can be computationally demanding, so we apply a basic preprocessing step aimed at reducing the input dimensionality. The raw frames are preprocessed by first converting their RGB representation to gray-scale and down-sampling it to a 84×84 image.As input Q-Network is preprocessing to the last 4 frames of a history and stacks them to produce the input to the Q-function.This process can be visualized as the following figure:

And convert these images to gray scale...

And send these into the Q-Network.

So what we have done;

• Take last 4 frames
• Resize images to 84x84
• Convert frames to gray-scale
• Stack them 84x84x4 input array and send them into the Q-Network.

The input to the neural network consists is an 84 × 84 × 4 image produced by φ. The first hidden layer convolves 32 8 × 8 filters with stride 4 with the input image and applies a rectifier nonlinearity. The second hidden layer convolves 64 4 × 4 filters with stride 2, again followed by a rectifier nonlinearity.The third hidden layer is fully-connected and consists of 7x7x64 input with 512 output,followed by a rectifier nonlinearity(input tensor is flattened). The final hidden layer is fully-connected and consists of 512 rectifier units. The output layer is a fully-connected linear layer with a single output for each valid action. The number of valid actions are 1 for left and 0 for right action.The architecture of the network is shown in the figure below:(Coming...)

Any contribution is welcome.