Scholarlink

<p>Phishing is a cyber-attack in which the attacker redirects Internet users to fraudulent websites. Fake websites look very similar to legitimate ones, leading users to trust them and disclose sensitive information. Despite the available methods, these attacks have grown exponentially, emphasizing the need for advanced techniques. This study proposes an EGSO-CNN model to detect web phishing by integrating features and optimizing deep learning (DL) techniques. A novel dataset has been created to address the availability of existing updated phishing datasets. The StandardScaler and Variational Autoencoders (VAE) are employed for preprocessing and feature extraction. The Enhanced Grid Search Optimization (EGSO) technique optimizes the model's performance. The proposed model yields an accuracy of 99.44%, a recall of 99.21%, and an f1-score of 99.32% with low false positive and error rates. The presented model can assist management by selecting effective phishing detection strategies to enhance customer delight.</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

Web-based phishing URL detection model using deep learning optimization techniques

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月8日

查看全文
  • Kousik Barik,
  • Sanjay Misra,
  • Raghini Mohan

摘要

Phishing is a cyber-attack in which the attacker redirects Internet users to fraudulent websites. Fake websites look very similar to legitimate ones, leading users to trust them and disclose sensitive information. Despite the available methods, these attacks have grown exponentially, emphasizing the need for advanced techniques. This study proposes an EGSO-CNN model to detect web phishing by integrating features and optimizing deep learning (DL) techniques. A novel dataset has been created to address the availability of existing updated phishing datasets. The StandardScaler and Variational Autoencoders (VAE) are employed for preprocessing and feature extraction. The Enhanced Grid Search Optimization (EGSO) technique optimizes the model's performance. The proposed model yields an accuracy of 99.44%, a recall of 99.21%, and an f1-score of 99.32% with low false positive and error rates. The presented model can assist management by selecting effective phishing detection strategies to enhance customer delight.