HearthStone Machine Learning

General presentation

This project is a major part of our studies [MSC Advanced Computer Science (Computational Intelligence)] at the University of Kent (United Kingdom).

Subject title: An application of Machine Learning to the Card Game Hearthstone.

It is about an AI (Machine Learning system) for card oriented game, applied to the game HearthStone (Blizzard company).

Contributors:

• Maigret Aurélien, original author of the project
• Zajda Florent
• Colin Julien

Supervisor:

• Marek Grzes

Project presentation

This is an AI project (Machine Learning system) for a card game, applied to the game HearthStone (Blizzard company). Hearthstone is a digital collectible card game (and free-to-play) that revolves around turn-based matches between two opponents. All played with cards that can summon creatures, weapons, spells, secrets, ...

We are interested by both artificial intelligence and the game HearthStone, that is why we would like to make our project of studies about a machine learning system, which will have the ultimate goal of learning decks by itself, and be able to advise the player on what to do against the predicted deck.

More information on docs/Project.md.

University deadlines

What you have to do	Deadline
Project Registrations Open	10am 18th January 2016
Submitted a Project Acceptance Form	4pm 29th January 2016
Early deliverable	2pm 27th June 2016
Submission of the corpus	2pm 25th August 2016
Submission of the dissertation	2pm 12th September 2016

Installation

Python 3 and NodeJS are required.

Application

In the Application directory:

npm install -g electron-prebuilt
npm install

Run the application (Application directory):

electron .

Update cards data

In the Application directory:

pip3 install Pillow
python3 fetch_data.py

Decks-Scraping

Python 2 is required.

pip2 install lxml
pip2 install requests

python2 scraping.py