حسين علي منجي النصراوي

ABSTRACT: In the contemporary digital landscape, safeguarding email communication against the omnipresent threat of spam remains a critical concern. This study introduces a novel approach, the Crow-Optimized Feedforward Neural Network with Long Short-Term Memory (C-FFNN-LSTM), designed to enhance spam detection capabilities. By integrating the collaborative behavior of crows through Crow Search Optimization (CSO) into the fine-tuning process of neural network parameters, the model aims to fortify email security. The combination of Feedforward Neural Network (FFNN) and Long Short-Term Memory (LSTM) architecture ensures a robust system for accurate spam detection. Efficiency is evaluated through established standards, including accuracy rates and false positive reduction. Experimental results demonstrate the efficacy of the C-FFNN-LSTM framework, showcasing exceptional accuracy levels and a notable decrease in false positives during testing. The proposed algorithm not only contributes to strengthening email security but also offers a promising avenue for refining spam detection algorithms in diverse domains. In the face of evolving cyber threats, this innovative approach represents an improved email security paradigm, emphasizing its robustness and reliability with an outstanding accuracy level of 99.1% during testing. Keywords: Unwanted emails, Cybersecurity threat, Crow Search Optimization (CSO), Advanced Neural Network (ANN), Memory-Augmented Neural Network (MANN), Unwanted communication detection, and Communication privacy.