A Learning Network Using Adaptive Threshold Elements

The synthesis of Boolean functions is a process for converting some specification for circuit behavior into an expression (optimal in some sense) for a Boolean function. It has been conventional to specify this behavior by means of a table of combinations (or truth table) in which for each system of values for the input variables (i.e., input state) there is associated either the value 1 or the value 0 or the symbol d [1]. These are called, respectively, ONE states, ZERO states and DON'T CARE states. This specification then identifies a nonempty class of Boolean functions and the process of Boolean function synthesis is to select from this class of functions one capable of being represented most simply.