Ternary decision diagrams with inverted edges and cofactoring-an application to discrete neural networks synthesis

A central issue in multiple valued logic design is the definition of a complete methodology that starting from a raw input/output description of the desired behaviour could give a detailed netlist of a circuit implementing it. The use of Decision Diagram techniques to represent multiple-valued functions, together with new learning algorithms derived from Neural Network theory may help in building such a methodology. It is possible to integrate all steps in a complete CAD synthesis technique.