Optimized LSTM–GWO Framework for Intelligent Email Spam Filtering in 5G Networks

This article is a comparison study revolving around machine learning and deep learning algorithms which have been applied in the spam email detection process with a particular focus on population based optimization algorithms. The comparison was also made between the traditional classifiers such as Logistic Regression, SVM, KNN, Decision Tree, Random Forest, Naive Bayes and K-Means with the deep learning networks such as LSTM, GRU, CNN and Dense networks. Systematic use of grey wolf optimizer (GWO), genetic algorithm (GA) and particle swarm optimization (PSO) optimization algorithms was applied to enhance the performance of models. The deep learning models have been determined to perform better than the traditional classifiers on a range of metrics, including the accuracy, recall, F1-score, MCC, Cohen Kappa and specificity. The performance of LSTM optimized by GA and GWO were the best and yielded a higher accuracy between 97.9, 96.7, and F1-score of 0.973. PSO not only improved CNN and Dense models but also gave the best traditional model, Accuracy of 97.5 percent and F1-score of 0.974. The findings reveal that deep learning and population-based optimizers. can be applied to design powerful and competent email spam detectors in the field.