A novel neural network half adder

This paper focuses upon the design of neural network to produce good solution to multiple-valued logic circuits. The theoretical basis for applying neural networks to multiple-valued logic algebra called neuro-algebra is proposed. This research also studies the design of a single artificial neural network model for half adders of binary, ternary, quaternary and quinary systems. The model has proven its efficiency with these four different radices. The advantages of the proposed multiple-valued logic algebra, neuro-algebra, are: supervised learning capability, simplicity of the neural network design, high performance, suitability for digital applications, straightforwardness of hardware implementations. The results demonstrate that it is possible to employ a systematic approach in designing neural networks for digital systems and that large-scale neural networks are capable of yielding high-quality solutions to complex problems.