Automatic variable selection and granular adaptation in fuzzy Boolean nets

In this work the problem of meta-learning, that is, perceiving what to learn (which variables, which granularity), is addressed in the context of Boolean nets with fuzzy behaviour. Fuzzy relational operators, embedded in those neural networks, are defined and the author shows how they can be used to establish the relevant antecedents as well as their topology of the network according these concepts and in order to efficiently learn a given set of rules from experiments is presented