Towards neuro-linguistic modelling

摘要

In neuro-fuzzy (or fuzzy-neural) models using unconstrained learning it is not possible to guarantee that the resulting membership functions represent human-interpretable linguistic terms. However one of the most interesting features of fuzzy systems is the insight provided on the linguistic relationship between their variables. This paper summarizes the requirements to build neuro-linguistic models. These are linguistically intrepretable due to the employement of a set of constraints that when used within an optimization scheme, such as backpropagation, obviate the subjective task of interpretating membership functions. Examples illustrating the performance of neuro-linguistic models are included.