ABSTRACT
This chapter investigates the close connection between fuzzy set theory and neural networks. A new structure of the artificial neuron is proposed, taking into consideration the recent knowledge about its physiology provided by biology. This new structure is shown to be well correlated with the basic notions in fuzzy logic, such that both languages may be used as complementary tools in modeling intelligent systems. In this context, fuzzy logic may be used to craft neural modules representing psychological constructs to be inherited by this intelligent system, in the same way as humans inherit some pre-wired (brain) circuits. These modules can be used as building blocks of the complex neural network used as the reasoning engine of the intelligent system. The learning capabilities of neural networks are then used to refine the connectivity inside the building blocks and to specify the connectivity among these blocks. In this way, neural networks become programmable besides being trainable.
