Building a Spam Detector Using Neural Networks Activation Functions

Abstract

This project explores the application of Artificial Neural Networks in the classification of electronic mail as either "Spam" or "Ham" (legitimate). By using activation functions, which serve as the mathematical "gates" that decide whether a neuron should fire a spam prediction. The experiment shows how automated filtering systems can greatly increase accuracy and reduce false positives by using tuning functions such as the Sigmoid and the Rectified Linear Unit (ReLU) [1]. Artificial neural networks are strong instruments that can make difficult decisions and learn from data. They are made up of layers of linked neurons that change with practice. Using the test data, the network updates its weights and biases. Activation functions facilitate this process by allowing the model to learn from errors and get better over time. This is the beginning of the creation of predictive generative artificial intelligence. Key Terms ⎯ Activation Function, Artificial Neural Network, Rectified Linear Unit, Sigmoid.