Email spam detection using gated recurrent unit (GRU) optimized by improved triangulation topology aggregation optimizer (ITTAO)

Résumé fourni par la source

Email spam detection is essential for preserving the security and integrity of digital communication systems by eliminating potentially dangerous and unwanted messages. This research investigates the effectiveness of advanced machine learning methods, with particular emphasis on a gated recurrent unit (GRU) neural network optimized using the improved triangulation topology aggregation optimizer (ITTAO). Through a thorough preprocessing approach, including stop word removal, normalization, lemmatization, and stemming, combined with the Bag of Words (BoW) model, the proposed system significantly enhanced email content classification. When the proposed model was evaluated on established datasets, such as Ling-Spam, the SMS Spam Collection v.1, and Enron, the model demonstrated remarkable precision, recall, and F1-score values of 96.87%, 96.98%, and 96.92%, respectively. The findings indicated that the proposed GRU-ITTAO model surpassed traditional algorithms and other neural network architectures, including GRU, Long Short‑Term Memory (LSTM), and Bidirectional Long Short‑Term Memory (BiLSTM). This study emphasized the potential of deep learning methods and machine learning in improving spam detection systems and pointed out that the model was capable of identifying the majority of spam emails.