Scholarlink

<p>In recent years, malware aimed at Android has become more common. Because of code obfuscation, antivirus solutions and outdated detection algorithms have a hard time keeping up. Researchers have begun using deep learning (DL) algorithms for malware detection, according to current research. Presented a DL model that combines Temporal Convolutional Networks (TCN) with Shapley Additive Explanations (SHAP) model as the basis for an algorithm for Android malware detection. Start by taking a look at the Android virus; not only will static characteristics be extracted, but dynamic behavioral elements with good anti obfuscation ability will as well. The TCN is implemented for processing the static characteristics due to their relative independence. When processing a series of dynamic features, the gate recurrent unit (GRU) is used because of the features’ temporal correlation. Hence, Understandable Artificial Intelligence- Android Malware Detection -Deep Learning (XAI-AMD-DL), a hybrid Android malware detection system employing DL models, is proposed in this research. It combines XAI with TCNs and finally classified using Logistic Twin based Support Vector Machine (SVM) model. Test the suggested model via the CICMalDroid 2020 dataset. The suggested XAI-AMD-DL model surpasses the current DL models with values of 99.98% accuracy, 99.75% precision, 99.75% recall, and f1-score, respectively.</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

Hybrid Temporal Convolutional Networks with LSTSVM Model for Android Malware Detection Using Explainable AI

  • 期刊论文

  • 发表日期：

    2025年9月5日

查看全文
  • A. Mohanraj,
  • K. Sivasankari

摘要

In recent years, malware aimed at Android has become more common. Because of code obfuscation, antivirus solutions and outdated detection algorithms have a hard time keeping up. Researchers have begun using deep learning (DL) algorithms for malware detection, according to current research. Presented a DL model that combines Temporal Convolutional Networks (TCN) with Shapley Additive Explanations (SHAP) model as the basis for an algorithm for Android malware detection. Start by taking a look at the Android virus; not only will static characteristics be extracted, but dynamic behavioral elements with good anti obfuscation ability will as well. The TCN is implemented for processing the static characteristics due to their relative independence. When processing a series of dynamic features, the gate recurrent unit (GRU) is used because of the features’ temporal correlation. Hence, Understandable Artificial Intelligence- Android Malware Detection -Deep Learning (XAI-AMD-DL), a hybrid Android malware detection system employing DL models, is proposed in this research. It combines XAI with TCNs and finally classified using Logistic Twin based Support Vector Machine (SVM) model. Test the suggested model via the CICMalDroid 2020 dataset. The suggested XAI-AMD-DL model surpasses the current DL models with values of 99.98% accuracy, 99.75% precision, 99.75% recall, and f1-score, respectively.