A New Computational Approach for High Capacity Multiple Rule FAMs

Y.S. Boutalis, S. Chois, and T.L. Kottas (Greece)


Fuzzy Associative Memory (FAM), Multiple rule multivariable inference, Perfect Recall Theorem.


A new method for constructing multiple rule Fuzzy Associative Memories (FAM) is presented in this paper. The method is based on the appropriate definition of a conjunction operator, which allows a significant increase in the storage capacity of the FAM. At the same time, the results of the inference procedure are identical to those obtained when the traditional min of Mamdani is used. The resulting FAM can store more rules than any other similar method of the literature, especially when fuzzy rules with more than one antecedent variables are involved Moreover, the proposed method does not suffer from inaccuracies due to the existence of non-symmetric membership functions. The method is ideal for use in Fuzzy systems involving large multivariable rule bases.

Important Links:

Go Back