This is an old revision of the document!


http://www.cs.toronto.edu/~vmnih/docs/dqn.pdf
Name: General Game Playing
You will learn: Java, ALE, Atari 2600 games, AI
Difficulty: hard
Excitement: You will be creating AI that can learn to play any (well at least a few) Atari 2600 games!
Description: People are trying to develop a general AI that can learn to play arbitrary Atari 2600 game. They even have a Java framework for that and there are some cool papers about this topic in the wild already. And that's what this topic is about.

Combine the framework with some AI algorithm (HyperNEAT, Deep learning, Evolutionary programming) and try to learn how to play some games!

This thesis can be either exploratory (you will try to apply AI technique to many games) or very focused (you will try to work with games that were learned already and you will try to improve their result).

Links: Atari 2600 and Deep learning paper

theme/general_game_playing.1423754889.txt.gz · Last modified: 2015/02/12 16:28 by jakub.gemrot