ABSTRACT

To this point, we demonstrated that neural networks can be taught by teachers or learned independently. We will now talk about networks that teach themselves without teachers and agree based on the interactions to handle the functioning of all neurons so that their cumulative results take on some new quality. These systems automatically generate complex mapping by transforming a set of input signals into output signals. This mapping generally cannot be predicted or controlled. Thus these networks are far more self-reliant and independent than the networks described earlier in this book.