Adaptive refinement of fuzzy knowledge bases using trend rules and inverse inference

In this paper, an adaptive approach to refinement of fuzzy classification knowledge bases within the framework of fuzzy relational equations is proposed. The fuzzy classification knowledge base can be built using the system of trend fuzzy rules and inverse inference. The essence of the approach is in constructing and training the composite neuro-fuzzy network isomorphic to linguistic solutions of fuzzy relational equations. The composite network allows adaptive refinement of the expert rules while the bounds of decision classes are changing.