So im just looking for as many examples of machine learning in commerical games as i can.

I have so far:

Neural Nets for Black and White, Fable 2, Fable 3, Creatures, Colin McCrae Rally 2

Drivatar for Forza 3. This learns from the user, dont know exactly what that technique is called.

Anyone know any more?

Cheers

There's mention of those and other stuff like the Halo3 Bayesian skill ranking TrueSkill system in this presentation:

http://research.microsoft.com/en-us/projects/mlgames2008/

It isn't a video game, but the mechanics of the commercial boardgame Yavalath were created by evolutionary algorithm:

http://www.cameronius.com/games/yavalath/

Its got a 7.26 average rating on BoardGameGeek so apparently it is pretty decent:

http://boardgamegeek.com/boardgame/33767/yavalath

Quote: Original post by discodowney
So im just looking for as many examples of machine learning in commerical games as i can.

I have so far:

Neural Nets for Black and White, Fable 2, Fable 3, Creatures, Colin McCrae Rally 2

Black & White was not a ANN but rather used reinforcement learning. Similar, but not the same. The main difference is that rather than generic "neurons", you are tweaking dials specific to a behavior or belief.

I would like to know where you got the information about Fable 2 and 3. Creatures did (and used a GA as well).

Edit: Sorry that was a mistake. the rest is right as far as i can tell

Its actually just an updated version of the Black And White one but the behaviour is context sensitive, so its prob not a true ANN in the same way you say Black and Whites isnt

Championship Manger used neural networks for analysis of football match replays. Basically a neural net was used to determine what portions of the replay could be removed as we degraded match histories over time without losing the vitals of the game.

Cheers Dave, just the kind of stuff im looking for.