Scholarlink

<p>Short Messaging Services (SMS) are plagued with spam most of the time, which creates several problems such as privacy violations, security risks, and information overload. Therefore, enhancing the performance of the SMS spam detection model is necessary. For more advanced analysis of SMS content and sender behaviour, many Machine Learning, Deep Learning models, and ensemble models are being used. The major drawback of existing models is the computational complexity of deep learning models and machine learning models. In this paper, we propose a new SMS Spam detection model to perform the spam detection effectively. Moreover, this model uses the existing classifiers such as Random Forest, Multi-nominal Naïve Bayes (MNB) and the newly developed Bayesian Optimized Convolutional Neural Network (BOCNN) for effective classification with the incorporation of Bayesian optimization technique on CNN. In addition, the standard data pre-processing tasks like Data cleaning, stop word removal, stemming, removal of numeric characters and punctuation, tokenization, N-Gram analysis, word cloud generation, vectorization, handling data imbalance and SMOTE. Finally, the proposed model is evaluated by conducting experiments by using spam or ham-labelled SMS message collection dataset that are taken from Kaggle and achieved 98.7% as detection accuracy which is the highest when compared to other SMS spam detection models. The performance of the proposed spam detection model is cross validated by conducting 10-fold cross validation.</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

An intelligent SMS spam detection model using Bayesian optimized deep learning classification

  • 期刊论文

  • 发表日期：

    2026年4月19日

查看全文
  • Sannasi Ganapathy,
  • Anthea Suzanne,
  • R. Hemamalini,
  • R. Kaarthika

摘要

Short Messaging Services (SMS) are plagued with spam most of the time, which creates several problems such as privacy violations, security risks, and information overload. Therefore, enhancing the performance of the SMS spam detection model is necessary. For more advanced analysis of SMS content and sender behaviour, many Machine Learning, Deep Learning models, and ensemble models are being used. The major drawback of existing models is the computational complexity of deep learning models and machine learning models. In this paper, we propose a new SMS Spam detection model to perform the spam detection effectively. Moreover, this model uses the existing classifiers such as Random Forest, Multi-nominal Naïve Bayes (MNB) and the newly developed Bayesian Optimized Convolutional Neural Network (BOCNN) for effective classification with the incorporation of Bayesian optimization technique on CNN. In addition, the standard data pre-processing tasks like Data cleaning, stop word removal, stemming, removal of numeric characters and punctuation, tokenization, N-Gram analysis, word cloud generation, vectorization, handling data imbalance and SMOTE. Finally, the proposed model is evaluated by conducting experiments by using spam or ham-labelled SMS message collection dataset that are taken from Kaggle and achieved 98.7% as detection accuracy which is the highest when compared to other SMS spam detection models. The performance of the proposed spam detection model is cross validated by conducting 10-fold cross validation.