A model of analogue K-winners-take-all neural circuit.

Abstract

A model of analogue K-winners-take-all (KWTA) neural circuit which can identify the K largest from N unknown wide range inputs, where 1≤K<N, is presented and analysed. The model is described by one state equation with discontinuous right-hand side and output equation. An existence and uniqueness, stability and convergence to the KWTA operation of the model states are analysed. A corresponding functional block diagram of the circuit is presented as N feedforward and one feedback hard-limiting neurons, which is used to determine the dynamic shift of inputs. The model combines such properties as high accuracy and convergence speed, low computational and hardware implementation complexity, and independency on initial states.