Neural computing: theory and practice
Neural computing: theory and practice
Hi-index | 0.00 |
Principles of operation, structure, and theoretical characteristics of an artificial neuron without activating function, which consists of series-connected memory elements based on multilayer thin-film anisotropic, spin-valve, and spin-tunneling magnetoresistive structures, are investigated. Examples of such neurons based on memory elements with optimal parameters are given. Prospects for three types of neurons are outlined.