Scholarlink

<p>Data security assurance is essential owing to the improving popularity of cloud computing and its extensive usage through several industries, particularly in light of the increasing number of cyber-security attacks. Ransomware-as-a-service (RaaS) attacks are prominent and widespread, allowing uniform individuals with minimum technology to perform ransomware processes. While RaaS methods have declined the access barriers for cyber threats, generative artificial intelligence (AI) growth might result in new possibilities for offenders. The high prevalence of RaaS-based cyberattacks poses essential challenges to cybersecurity, requiring progressive and understandable defensive mechanisms. Furthermore, deep or machine learning (ML) methods mainly provide a black box, giving no data about how it functions. Understanding the details of a classification model’s decision can be beneficial for understanding the work way to be identified. This study presents a novel Two-Tier Metaheuristic Algorithm for Cyberattack Defense Analysis using Explainable Artificial Intelligence based Bayesian Deep Learning (TTMCDA-XAIBDL) method. The main intention of the TTMCDA-XAIBDL method is to detect and mitigate ransomware cyber threats. Initially, the TTMCDA-XAIBDL method performs data preprocessing using Z-score normalization to ensure standardization and scalability of features. Next, the improved sand cat swarm optimization (ISCSO) technique is used for the feature selection. The Bayesian neural network (BNN) is employed to classify cyberattack defence. Moreover, the BNN’s hyperparameters are fine-tuned using the whale optimization algorithm (WOA) model, optimizing its performance for effective detection of ransomware threats. Finally, the XAI using SHAP is integrated to provide explainability, offering perceptions of the model’s decision-making procedure and adopting trust in the system. To demonstrate the effectiveness of the TTMCDA-XAIBDL technique, a series of simulations are conducted using a ransomware detection dataset to evaluate its classification performance. The performance validation of the TTMCDA-XAIBDL technique portrayed a superior accuracy value of 99.29% over the recent methods.</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

Two-Tier heuristic search for ransomware-as-a-service based cyberattack défense analysis using explainable Bayesian deep learning model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月5日

查看全文
  • Ali Saeed Almuflih

摘要

Data security assurance is essential owing to the improving popularity of cloud computing and its extensive usage through several industries, particularly in light of the increasing number of cyber-security attacks. Ransomware-as-a-service (RaaS) attacks are prominent and widespread, allowing uniform individuals with minimum technology to perform ransomware processes. While RaaS methods have declined the access barriers for cyber threats, generative artificial intelligence (AI) growth might result in new possibilities for offenders. The high prevalence of RaaS-based cyberattacks poses essential challenges to cybersecurity, requiring progressive and understandable defensive mechanisms. Furthermore, deep or machine learning (ML) methods mainly provide a black box, giving no data about how it functions. Understanding the details of a classification model’s decision can be beneficial for understanding the work way to be identified. This study presents a novel Two-Tier Metaheuristic Algorithm for Cyberattack Defense Analysis using Explainable Artificial Intelligence based Bayesian Deep Learning (TTMCDA-XAIBDL) method. The main intention of the TTMCDA-XAIBDL method is to detect and mitigate ransomware cyber threats. Initially, the TTMCDA-XAIBDL method performs data preprocessing using Z-score normalization to ensure standardization and scalability of features. Next, the improved sand cat swarm optimization (ISCSO) technique is used for the feature selection. The Bayesian neural network (BNN) is employed to classify cyberattack defence. Moreover, the BNN’s hyperparameters are fine-tuned using the whale optimization algorithm (WOA) model, optimizing its performance for effective detection of ransomware threats. Finally, the XAI using SHAP is integrated to provide explainability, offering perceptions of the model’s decision-making procedure and adopting trust in the system. To demonstrate the effectiveness of the TTMCDA-XAIBDL technique, a series of simulations are conducted using a ransomware detection dataset to evaluate its classification performance. The performance validation of the TTMCDA-XAIBDL technique portrayed a superior accuracy value of 99.29% over the recent methods.