Structure of oscillating neural network for artificial neural circuit and pattern separation and recognition

摘要

The present invention relates to an artificial neural network to which an artificial neural pulse coding principle is applied and a neural network constructed by using the artificial neural pulse coding principle and performing separation and recognition of a pattern using phase information of a neural circuit, A transistor M2 provided with a predetermined signal Vr3 and a transistor M2 connected between the two transistors M1 and M2 and provided with a predetermined input signal Sj A synapse circuit 10 constituted by a combination of the input signal Sj (M3) and emitting or absorbing a current 13 proportional to the weighting voltage when a predetermined input signal Sj (t) is provided, And a colorimetric projection circuit 30 for converting an output signal of the nerve body circuit 20 into a predetermined pulse string by a predetermined reference signal It is characterized by being composed