Jump to content
  • Advertisement
Sign in to follow this  
kfboelter

machine learning c++ project - tip

This topic is 2908 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hello. Once again, I come in search of guidance.

So.. I am supposed to get a AI project going (yes, for school),,, one that makes use of machine learning... in a game,,, it could be some sort of enemy, or some manager for a given entity, or something like that... turns out I am really out on ideas of what to do... so coud anyone one point me the way?
what could possibly be a good idea,,, something realistic, given that I am a newbie to machine learning (and general AI)...
I have chosen to use c++ for the implementation, so I would need to know an adequate library to use as well..
So basically, If anyone could just say "well, you could try to do xxxxx using xxx" and some additional insight on the project, I would be very thankful...

Share this post


Link to post
Share on other sites
Advertisement
sorry for the bump...
but can't anyone really just tell me what a good machine learning project would be a nice one for a newbie at the subject to get started?

Share this post


Link to post
Share on other sites
Currently, the number one application for machine learning in game development is car racing AI.

Share this post


Link to post
Share on other sites
See the parallel discussion on why, generally speaking, I don't favor learning for Game AI. :)

With that said... nothing wrong with it for a school project. Even if it doesn't work fantastically, you'll still learn something.

I believe Ian Millington's book on Game AI discusses the use of simple pattern recognizers (not sure if that really counts as learning or not) for fighting games. There may be some other examples of using machine learning as well - I don't recall for sure and don't have the book handy.

Controlling a vehicle in a racing game is definitely an option.

You could try learning for an FPS bot. I don't expect it to go well, but it might give you some good insight into why it's hard (which is probably as valuable or more valuable than insight into cases where it's easy).

Play selection in a sports game (e.g. for football - by which I don't mean soccer). I don't expect that online learning would work very well there either, because it's probably too easy for the player to manipulate the learning, but it might be interesting to try offline learning and see what you come up with. Then again, the sort of pattern recognizers that Millington talks about for fighting games might work better than I expect.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!