An optimal multilayer neural interpolating (OMNI) net in a generalized Fock space setting

As a generalization and extension of the OI (optimal interpolative) artificial neural network, an OMNI (optimal multilayer neural interpolative) net is proposed. An OMNI net is an interconnected system of OI nets in either feedforward, feedback, or recurrent configuration. The synaptic weights for the OMNI net can be calculated by closed-form expressions. For the feedforward configuration a forward-propagation algorithm is described. OMNI nets can be used for classification and interpretation of complex patterns. As an example, an OMNI net model for the six-layer cortical column of the human brain is described.<>