Scholarlink

<p>Short Message Service (SMS) remains a widely used communication medium and a growing target for phishing attacks (smishing). This paper presents a hybrid deep learning framework that integrates Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and a multi-head self-attention mechanism to detect phishing SMS messages. To address the severe class imbalance in the University of California, Irvine (UCI) SMS Spam Collection dataset, we apply the Synthetic Minority Oversampling Technique (SMOTE) and optimize the model using Bayesian hyperparameter tuning. Experimental results demonstrate state-of-the-art performance, achieving 99.17% accuracy and an Area Under the Curve Receiver Operating Characteristic (AUC-ROC) of 0.9996, surpassing comparable baselines. The proposed approach combines high detection accuracy with computational efficiency, offering a practical solution for modern SMS phishing mitigation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Combating modern smishing: A deep learning framework leveraging hybrid CNN-LSTM and attention mechanisms

  • 期刊论文

  • 发表日期：

    2025年12月27日

查看全文
  • Rachid Bourigue,
  • Driss Ait Omar,
  • Mohamed Khayati,
  • Hicham Zougagh

摘要

Short Message Service (SMS) remains a widely used communication medium and a growing target for phishing attacks (smishing). This paper presents a hybrid deep learning framework that integrates Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and a multi-head self-attention mechanism to detect phishing SMS messages. To address the severe class imbalance in the University of California, Irvine (UCI) SMS Spam Collection dataset, we apply the Synthetic Minority Oversampling Technique (SMOTE) and optimize the model using Bayesian hyperparameter tuning. Experimental results demonstrate state-of-the-art performance, achieving 99.17% accuracy and an Area Under the Curve Receiver Operating Characteristic (AUC-ROC) of 0.9996, surpassing comparable baselines. The proposed approach combines high detection accuracy with computational efficiency, offering a practical solution for modern SMS phishing mitigation.