Comparative Analysis of Neural Network Models for Spam E-mail Detection

With increasing number of internet users and the fast growth of digital communication have made email one of the most important ways of communication. Along with the growing number of users, spam emails also start rising. It has become one of the common problems on the internet. Spam emails have a huge impact on users as spammers’ deceptive emails can trick people into sharing personal information or unknowingly downloading harmful software by pretending to be genuine messages. The consequences of falling for scams can vary from losses to the exposure of personal data. To address this issue, numerous types of machine learning and deep learning models have been proposed in various research papers. This paper presents a comparative study of Neural network models such as [deep learning models] Capsule Networks (CAPSNET), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN) and [Transformer model] RoBERTa to determine the best model. The capacity of each model was evaluated by assessing various performance metrics, including F1 score, recall, precision, accuracy, and support.