PDA

Click to See Complete Forum and Search --> : math paper on artificial itnelligence


MXAlPhA
Dec 10th, 2002, 07:41 PM
I'm writing a math paper on artificial intelligence. What topics and/or concepts you think i can include in this? Preferably ones that have some math in it.

alkatran
Dec 10th, 2002, 08:50 PM
I'm not really an expert, but I read a article on it once..

Apparently you should:

-Give the AI a random property to keep the user amused
-Learning is good

Wow, I know I helped

Don't forget to do what Blizzard did to make their Insane, Normal, and Easy AIs in Warcraft 3 did, make them get more gold per trip tp teh gold mine as difficulty increases, wow, so complicated.

MXAlPhA
Dec 10th, 2002, 09:04 PM
Originally posted by alkatran
I'm not really an expert, but I read a article on it once..

Apparently you should:

-Give the AI a random property to keep the user amused
-Learning is good

Wow, I know I helped

Don't forget to do what Blizzard did to make their Insane, Normal, and Easy AIs in Warcraft 3 did, make them get more gold per trip tp teh gold mine as difficulty increases, wow, so complicated.

LOL!!!

SteveCRM
Dec 10th, 2002, 09:33 PM
www.aidepot.com should have some good stuff :cool: i think its .com, iif not try some other extensions :)

Spooner
Dec 11th, 2002, 01:24 AM
I would certainly include genetic algorithms and neural networks. Probably agents, too. Fuzzy logic is way cool.

You could look at programming languages which are used in AI? Like Lisp.

You can't really escape the philosophical issues of what intelligence is in the first place. Once you've grappled with that, you then have to decide what artificial intelligence is, which implies there is some kind of real intelligence too. That's the part that makes my brain hurt.

To give your paper some practicality, how about looking into the application of ai, like pattern recognition for ocr, medical diagnoses, financial dealings and so on.

Here're some links that might help:


Links to more resources (http://ai.iit.nrc.ca/ai_point.html)

UMBC Agent Web (http://agents.umbc.edu/)

vd Smagt and Krose from the robotics area at German Aerospace have made this Intro to Neural Networks (http://www.robotic.dlr.de/Smagt/books/neuro-intro.ps.gz) available. Browse the rest of the site.

For info on Fuzzy Logic, look for anything by Lotfi Zadeh who invented it or Bart Kosko who really took it to heart.

DiGiTaIErRoR
Dec 11th, 2002, 01:10 PM
I've had my own thoughts about rationalizing and such.

Perhaps talks about probability, statistics, relativity, and so forth.

I've thought about developing an AI.

There is symbol recongnition, phrase, macro phrase, micro phrase, and so forth.

For instance.

a phrase being, hello, how are you today, similar to a sentence. A macro phrase, for example, this entire message, micro phrases, or in other terms words. And symbols being the letters. All with relations.

The issue of artificial intelligence is really, is it artificial? The point would to make such an entity think for itself. So where do you draw the lines between making it think and letting it think.

And once intelligent is it really artificial, if you can grasp that which is intelligence into a program, then it's not really artificial at all. It would not have life, of course, but thought would be there.

I'd thought about relational realization. You can't have intelligence without knowledge. Thus input is the key. Also, time is good to have. Realizing things aren't singular but multi dimensional.

Also, there are thoughts that are more common than others. All thoughts trigger by the preceding thought or inputed stimuli.

There are so many considerations. And there is also favoring, forming opinion, and so much more.

Math is only a small part of it. I think you should focus on other points as well.

MXAlPhA
Dec 11th, 2002, 02:21 PM
fuzzy logic i was definitely including, but i need more math! any relatively simple topics about ai that work with a good deal of like math?

alkatran
Dec 11th, 2002, 02:23 PM
What about pathing? That's mathy isn't it?:confused: