Boolean Soft Computing by Non-linear Neural Networks With Hyperincursive Stack Memory

摘要: This paper is a review of new theoretical basis for modelling neural Boolean networks by non-linear digital equations. With real numbers, soft tables can be generated. integer these equations are Heaviside fixed functions in the framework threshold logic. These represent neurons which split very easily into set McCulloch and Pitts formal with hidden neurons. It demonstrated that any represented such where weights always either an activation weight +1 or inhibition -1, threshold. The parity problem fully solved fractal network based on XOR. From feedback to inputs XOR equations, it showed compete each other. Moreover, output neuron gives rise chaos. A model stack memory designed from chaos map Binary digits memorised folding variable anti-chaotic hyperincursive process. retrieval data computed incursive chaotic last value variable. Incursion extension recursion iterate function variables not only defined past present time also future. Hyperincursion incursion generating multiple iterates at step. basic PearlVerhulst one zone realises coding input binary message under form similar Gray code. exclusive OR equation mixing numbers.