Dominion Strategy Forum

Dominion => Simulation => Topic started by: jaybh on October 26, 2011, 01:53:28 pm

Title: AI project - where to start?
Post by: jaybh on October 26, 2011, 01:53:28 pm

Hi,
I am a graduate student in an Artificial Intelligence class, and I am working on a project that involves using genetic algorithms to develop strategies for playing Dominion.  This forum seems to have a great deal of information, and I am going to work my way through it, but I was wondering if anyone could point me to some good examples of previous work that has been done in this area, so I have an idea of where I should focus my energies in order to produce something new and useful.
Thanks so much!

Title: Re: AI project - where to start?
Post by: DStu on October 26, 2011, 02:07:18 pm

So from what I know:
 There is a Master thesis on the BaseGame  (http://boardgamegeek.com/thread/647682/masters-thesis-on-dominion-ai)
There are the two Simutors I know, which have a lot of standard strategies and heuristics
 rspeer  (http://forum.dominionstrategy.com/index.php?topic=619.0) (with code on github) and
 Geronimoos (with a lot of bots)  (http://forum.dominionstrategy.com/index.php?topic=144.0)

Title: Re: AI project - where to start?
Post by: jonts26 on October 26, 2011, 03:43:03 pm

There's also this thread.

http://forum.dominionstrategy.com/index.php?topic=632.0

It brings up the topic of topic of genetic algorithms. It doesnt really get into it too deeply but it's probably a decent place to start.

Title: Re: AI project - where to start?
Post by: DG on October 26, 2011, 05:00:30 pm

One difficulty you might find is the lack of separation between the decisions of an AI. If you try to separate a high level strategy, card play, card purchasing, and deck composition, you might find that they all affect each other in different ways each turn. This is even before any consideration is made of the kingdom supply and an opponent's deck.