🔒 Locked

Expert Systems

Started by VVoltz Jan 26, 2004 at 9:19 AM 3 replies 700+ views

Original Post

Jan 26, 2004 at 9:19 AM

I''m using a turn based squared grid to simulate on historic battle fought on 1825, both armies (there are only 2) are as follows: 10 Infantry troops (melee and ranged attacks [1 hit point]) 5 Cavalry troops (melee attacks [2 hit points]) 8 Cannon units (ranged attacks [3 hit points]) where evey one of them is a batallion of let''s say: 10 soldiers (referred as Hit Points or Life Points) Each turn each opponent army can move up to all his troops. Let''s say there are no modiffications to the hit points. The scenario is about 15x15 squares. My problem is this: I really don''t know what kind of AI to use, I opened a thread about using just Neural Networks (impossible to use), so mu options are: - Using Expert Systems - Using Decitional Trees (like the one used on chess games) - Using Expert Systems + Fuzzy Logic - Using Expert Systems + Expert Systems + Fuzzy Logic - Using Trees + some other algorithm that I actually don''t know and probably will have to design. What do you think?

Syntax

Jan 26, 2004 at 9:59 AM

I''ve done a few expert systems, and with all honesty whether you use some type of probabilistic reasoning or not, you still have a rule-based system.

Elaborate with your question, give us a few rules that you were thinking of, and maybe then we may answer with better accuracy and figure out if dectrees or fuzzy logic tacked on top of the RBS is a good idea or not.

-Lucas

Timkin

Jan 26, 2004 at 7:50 PM

quote:
Original post by Syntax
I''ve done a few expert systems, and with all honesty whether you use some type of probabilistic reasoning or not, you still have a rule-based system.

That''s only marginally true and only if you''re talking about discreet systems. Probability density functions over continuous variables cannot be reduced to a set of FOL rules over those variables. That''s the whole point of using density functions... because they describe more complex systems/relationships than rule sets do.

Timkin

VVoltz

Jan 27, 2004 at 5:10 PM

One of the few things I know about Expert Systems is that it needs some knowledge source.

In this case some battle tactics would be reasonable, for example:

If a large mass of soldiers advances from the center of the map, the best solution would probably be a both side flank attack, this battle tactics would represent the knowledge source.

I''m I correct?, is this possible??

Syntax

Jan 28, 2004 at 4:04 PM

Kinda, the knowledge of the expert is stored in the rules, so one rule may be

[If Enemy.Approaching & Enemy.Location.Center]
->
[Army.Flank.Left & Marines.Flank.Right]

The knowledge of the "general" is stored in a rule, like the one above...

-Lucas

[edited by - Syntax on January 28, 2004 5:04:50 PM]

Topic Locked

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