Skip to main content
GameDev.net gamedev.net
🔒 Locked

uncertainty in AI

Started by Seteropere Feb 23, 2010 at 10:13 PM 4 replies 2.4k views
Original Post
Seteropere
Seteropere
Hello , I'm looking for a way to apply uncertainty in a real application .. I know Bayesian aka belief nets and Java .. I just dont know how to combine them ,is there any book useful for this ?or at least online material ? any ideas ? thanx ..
Slather
Slather
I don't know anything about belief nets, but maybe you could use them to determine the fitness of a node in a neural network?
If you're new to genetic algorithms and neural nets, ai-junkie has a decent primer.
One highly-recommended book that I've been meaning to buy is Programming Game AI by Example. It supposed to talk about fuzzy logic, and might get you close to implementing belief nets.
IADaveMark
IADaveMark
(He did NOT just bust out NNs, did he?)

Yes, some form of Bayesian inference is usually the best way of handling uncertainty. The next problem is what you do with that information on the other end, though.

Perhaps a better description of your problem?
Dave Mark - President and Lead Designer of Intrinsic Algorithm LLC
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-founder and 10 year advisor of the GDC AI Summit<
Hodgman
Hodgman
Quote:
Original post by InnocuousFox
Perhaps a better description of your problem?
Seconded ;)

If you just want the AI to be a bit less logical with it's choices, rand is all you need.

Usually when making a decision, you use some heuristic to weigh up all the options, then pick the best one. Before picking the best one, add a random number (within a reasonable range - e.g. 10% of the weighting range) to each of the weights. Now if there's a few "good options", the AI will choose from any of them, instead of always taking the very best option.
Slather
Slather
yeah, it was my mistake to answer a topic I know nothing about, but I figure my answer is as good as any when someone asks how to "combine" belief nets and Java. :) How do I combine game design with Python? forgive me for the snarky remarks, im sleepy.
IADaveMark
IADaveMark
Quote:
Original post by Hodgman
Usually when making a decision, you use some heuristic to weigh up all the options, then pick the best one. Before picking the best one, add a random number (within a reasonable range - e.g. 10% of the weighting range) to each of the weights. Now if there's a few "good options", the AI will choose from any of them, instead of always taking the very best option.

Whoa... chapters 12-16 of my book! ;-)

Anyway, that's the thing we need to clarify. Is he asking about dealing with uncertain information for input or trying to create varied results?

Dave Mark - President and Lead Designer of Intrinsic Algorithm LLC
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-founder and 10 year advisor of the GDC AI Summit<

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.

Sign in to reply to this topic.