Development of quadratic neural unit with applications to pattern classification

Abstract

The computational neural-network structures described in the literature are often based on the concept of linear neural units (LNUs). The biological neuron is a complex computing element, which performs more computations than just linear summation. The computational efficiency of the neural network depends on its structure and the training methods employed. Higher-order combinations of inputs and weights will yield higher neural performance. Here, a quadratic-neural unit (QNU) has been developed using a novel general matrix form of the quadratic operation. We have used the QNU for realizing different logic circuits