Scholarlink

<p>Maintaining data security in the cloud is crucial for safeguarding user data. Cloud data management efficiency can be enhanced by tackling essential strategies that mitigate cyber-attacks and vulnerabilities within the cloud infrastructure. A significant risk in cloud computing involves data threats, including unauthorized access to data or leakage from the cloud environment. Efficient data management and leak prevention techniques should be applied and regularly assessed to address these risks. The cloud system can protect data by implementing strong security protocols while facilitating secure technological functions. This study examines the efficiency of the isolation forest (iForest) algorithm for detecting anomalies in cloud settings, addressing the vital concern of data security against cyber-attacks. We suggest a hybrid method that fuses anomaly detection with Machine learning-driven behavioural analysis to improve threat intelligence. The iForest algorithm is assessed alongside conventional techniques like support vector machine, random forest, and local outlier factor, showcasing better performance indicators, such as increased detection accuracy and reduced false positive and false negative rates. In particular, the iForest algorithm obtained an average accuracy of 0.842 and a false negative rate of 0.06, demonstrating its effectiveness in detecting anomalies. Furthermore, the Algorithm showed effective processing times, varying from 10 to 150&#xa0;ms, and utilized little network bandwidth. The results indicate that the iForest algorithm enhances threat detection precision and seamlessly merges with current security frameworks, offering a scalable answer for real-time threat identification in cloud settings. Future research will investigate the real-world difficulties of executing this Algorithm in operational environments, particularly concerning latency and its integration with existing security systems.</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 anomaly-based cloud security and machine learning-based behavioral analysis for threat intelligence detection in cloud environment

  • 期刊论文

  • 发表日期：

    2025年8月19日

查看全文
  • Vijaykumar Mamidala,
  • Akhil Raj Gaius Yallamelli,
  • Mohanarangan Veerappermal Devarajan,
  • Thirusubramanian Ganesan,
  • Rama Krishna Mani Kanta Yalla,
  • Aceng Sambas

摘要

Maintaining data security in the cloud is crucial for safeguarding user data. Cloud data management efficiency can be enhanced by tackling essential strategies that mitigate cyber-attacks and vulnerabilities within the cloud infrastructure. A significant risk in cloud computing involves data threats, including unauthorized access to data or leakage from the cloud environment. Efficient data management and leak prevention techniques should be applied and regularly assessed to address these risks. The cloud system can protect data by implementing strong security protocols while facilitating secure technological functions. This study examines the efficiency of the isolation forest (iForest) algorithm for detecting anomalies in cloud settings, addressing the vital concern of data security against cyber-attacks. We suggest a hybrid method that fuses anomaly detection with Machine learning-driven behavioural analysis to improve threat intelligence. The iForest algorithm is assessed alongside conventional techniques like support vector machine, random forest, and local outlier factor, showcasing better performance indicators, such as increased detection accuracy and reduced false positive and false negative rates. In particular, the iForest algorithm obtained an average accuracy of 0.842 and a false negative rate of 0.06, demonstrating its effectiveness in detecting anomalies. Furthermore, the Algorithm showed effective processing times, varying from 10 to 150 ms, and utilized little network bandwidth. The results indicate that the iForest algorithm enhances threat detection precision and seamlessly merges with current security frameworks, offering a scalable answer for real-time threat identification in cloud settings. Future research will investigate the real-world difficulties of executing this Algorithm in operational environments, particularly concerning latency and its integration with existing security systems.