Date Added: Jul 2012
Predicting, considering and designing all of the possible states of real world problems that are justified in artificial intelligence domain is relatively difficult or rather impossible for the experts. One reason is that the real world problems are highly complicated and they depend on a lot of variables. Furthermore, they do not practice their behavior as similar as their past comportments and it is rare that the authors find a linear behavior from such mentioned problems. They propose to apply data-driven data mining approaches to learn non-linear systems' behaviors (rather than expert's knowledge driven ones), so they could define delicate fuzzy states to indicate the system behavior.