https://github.com/alvinwan/bots-for-atari-games
Bots for Atari Games using Reinforcement Learning
https://github.com/alvinwan/bots-for-atari-games
atari-games deep-learning deep-reinforcement-learning frozenlake q-learning reinforcement-learning spaceinvaders
Last synced: 7 months ago
JSON representation
Bots for Atari Games using Reinforcement Learning
- Host: GitHub
- URL: https://github.com/alvinwan/bots-for-atari-games
- Owner: alvinwan
- License: apache-2.0
- Created: 2018-01-26T23:21:06.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2019-04-06T23:26:28.000Z (about 7 years ago)
- Last Synced: 2025-06-11T20:56:04.270Z (12 months ago)
- Topics: atari-games, deep-learning, deep-reinforcement-learning, frozenlake, q-learning, reinforcement-learning, spaceinvaders
- Language: Python
- Homepage:
- Size: 25.4 KB
- Stars: 12
- Watchers: 1
- Forks: 9
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# [How to Build a Bot for Atari with OpenAI Gym](https://www.digitalocean.com/community/tutorials/how-to-build-atari-bot-with-openai-gym)
**Want an in-person tutorial with step-by-step walkthroughs and explanations? See the corresponding AirBnb experience for both beginner and experienced coders alike, at ["Build a Dog Filter with Computer Vision"](https://abnb.me/UunXrPyqVO)** ([See the 45+ 5-star reviews](https://www.airbnb.com/users/show/87172280))
This repository includes all source code for the [tutorial on DigitalOcean](https://www.digitalocean.com/community/tutorials/how-to-build-atari-bot-with-openai-gym) with the same title, including:
- Q-table based agent for FrozenLake
- Simple neural network q-learning agent for FrozenLake
- Least squares q-learning agent for FrozenLake
- Code to use fully pretrained Deep Q-learning Network (DQN) agent on Space Invaders
> Each of these agents solve FrozenLake in 5000 episodes or fewer; whereas not in record time or even close to it, the agents are written with minimal tuning
created by [Alvin Wan](http://alvinwan.com), January 2018
# Getting Started
For complete step-by-step instructions, see the [tutorial on DigitalOcean](https://www.digitalocean.com/community/tutorials/how-to-build-atari-bot-with-openai-gym). This codebase was developed and tested using `Python 3.6`. If you're familiar with Python, then see the below to skip the tutorial and get started quickly:
> (Optional) [Setup a Python virtual environment](https://www.digitalocean.com/community/tutorials/common-python-tools-using-virtualenv-installing-with-pip-and-managing-packages#a-thorough-virtualenv-how-to) with Python 3.6.
1. Navigate to the repository root, and install all Python dependencies.
```
pip install -r requirements.txt
```
2. Navigate into `src`.
```
cd src
```
3. Download the Tensorflow model for SpaceInvaders, from Tensorpack's A3C-Gym sample.
```
mkdir models
wget http://models.tensorpack.com/OpenAIGym/SpaceInvaders-v0.tfmodel -O models/SpaceInvaders-v0.tfmodel
```
4. Launch the script to see the Space Invaders agent in action.
```
python bot_6_dqn.py --visual
```
# How it Works
See the below resources for explanations of related concepts:
- ["Understanding Deep Q-Learning"](http://alvinwan.com/understanding-deep-q-learning)
- ["Understanding the Bias-Variance Tradeoff"](http://alvinwan.com/understanding-the-bias-variance-tradeoff)