Single-Layer Perceptron Capable Of Classifying 2N+1 Distinct Input Patterns

Abstract

A new multi-threshold Perceptron capable of handling both binary and analog input is presented and discussed. The modified Perceptron replaces the sigmoid function with sum of Gaussian functions with different mean values. A computer program was developed to simulate behavior of a network utilizing the modified Perceptron. Both XOR and parity check problems were solved using a single-layer network utilizing this modified Perceptron. Based on the results obtained from our simulation, the modified Perceptron is capable of solving problems (such as. XOR) that cannot be solved using the classical Perceptrotl. Also, a network utilizing this modified Perceptron reqUIres fewer iterations to converge to a solution than that of a multi-layer network using back propagation.

Related Papers

• → The generalized sigmoid activation function: Competitive supervised learning(1997)220 cited
• → Weighted sigmoid gate unit for an activation function of deep neural network(2020)52 cited
• → A Novel Activation Function in Convolutional Neural Network for Image Classification in Deep Learning(2020)2 cited
• → Weighted Sigmoid Gate Unit for an Activation Function of Deep Neural Network(2018)8 cited
• → Evaluation of Sigmoid and ReLU Activation Functions Using Asymptotic Method(2022)