In the eld of data{based fuzzy modeling two approaches are predominately applied: Firstly, the optimization of an entire rule base by minimizing the modeling error and secondly to incrementally set up the rule base by individual tested rules. Following the second approach the search space increases exponentially with the number of input variables. On the other hand, due to the limited amount of available data, the search space becomes more and more sparsely populated. Therefore, part of the possible rules are no longer supported by any data set and can be neglected in the search process. The tree{oriented rule search concept, presented in this paper, takes advantage of this fact and leads to a drastically reduced computational e ort.
[1]
Brian R. Gaines,et al.
Fuzzy reasoning and its applications
,
1981
.
[2]
A. Krone,et al.
A hybrid evolutionary search concept for data-based generation of relevant fuzzy rules in high dimensional spaces
,
1999,
FUZZ-IEEE'99. 1999 IEEE International Fuzzy Systems. Conference Proceedings (Cat. No.99CH36315).
[3]
Timo Slawinski,et al.
Test- and Rating Strategies for Data Based Rule Generation
,
1998
.
[4]
A. Krone,et al.
Relevance Test for Fuzzy Rules
,
1998
.
[5]
H. Kiendl.
Decision Analysis by Advanced Fuzzy Systems
,
1999
.