Abstract:
One gate corresponds to linear threshold function (LTF) arity n. The algorithm of work of the gate is
based on research of the internal device of the LTF-functions inducing splitting of numerical space of
scales on final number of classes, being in one to one conformity with realized logic functions. The gate it
is adaptive increases weight in the form of a n-bit vector (on one bit on each argument). (1) Algorithm
doesn't limit number accessible linear function. (2) Algorithm is based on prescheduled calculation of a
sign. The gate is intended for use in traditional neural systems. The gate can be realized at hardware level
in system of architecture FPGA.