A COMPREHENSIVE REVIEW OF LEARNING RULES AND ARCHITECTURE OF PERCEPTRON IN ARTIFICIAL NEURAL NETWORKS (ANNS) (Book Chapter)

Abstract

The complicated neural networks of the human mind have acted as a significant model for creating artificial neural networks (ANNs) of computational intelligence. ANNs can recognize patterns in data, make decisions, and perform other functions. The study provides a comprehensive review that explores ANNs by analysing the crucial elements of learning rules and perceptron architectures. This chapter clarifies the foundational learning rules underlying ANNs’ ability to adapt and generalize from data. The investigation comprehensively inspects the vital elements of learning rules and perceptron architectures in Artificial Neural Networks (ANNs) inspired by the detailed neuronal networks of the human brain. This chapter subsequently explores the dynamic realm of perceptron architectures within ANNs. Single-layer perceptrons are examined for their inability to handle intricate relationships. In contrast, multilayer perceptrons (MLPs) emerge as formidable solutions. The complex composition of MLPs, characterized by input, hidden, and output layers, is deconstructed, highlighting their potential to capture intricate non-linear patterns through the strategic deployment of activation functions. This analysis showcases a merging of academic notions and actionable effects. The combined effect between learning rules and perceptron architectures forms the foundation of ANNs’ expertise in pattern recognition, prediction, and decision-making tasks. By comprehensively understanding these underpinnings, researchers and practitioners can connect the potential of ANNs across diverse domains.