Scholarlink

<p>The digital transformation of modern society has intensified cyber threats, demanding advanced, real-time defense mechanisms. Traditional lack the capability to sustain pace in light of the scope and multifaceted nature of attacks, making AI-driven solutions vital. Leveraging Big Data enables these AI models to analyze vast, diverse datasets, such as network logs and user activities, to identify patterns and enhance predictive cybersecurity. This manuscript proposes an innovative, optimized Deep learning (DL) assisted Spark architecture for analyzing various cyberattacks in a big data IoT environment. Generally, a set of slave nodes and a master node are used to train the classes. Initially, the attack data sourced from the original dataset is prepared through the execution of data cleaning and data normalization using Pareto scaling to enhance the data quality. After preprocessing, the features are extracted, and multiclass classification is performed using the stacked dilated Convolutional gated bidirectional recurrent unit (SDC-GBRU) model, which classifies various attacks like MITM-ARP-spoofing, SSH-brute force attack, FTP-brute force, DDOS-ICMP, DDOS-RAWIP, DDOS-UDP, DOS attack, exploiting-FTP attack, fuzzing, ICMP flood, SYN-flood, port scanning, remote code execution, SQL injection, and XSS accurately. enhanced through the implementation of the Lyrebird optimization (LBO) strategy that streamlines network computation during the learning process. A Python simulation tool is devised for experimentation, and various performance measures like accuracy, F-score, Matthew’s correlation coefficient (MCC), F-score, false negative rate (FNR), and Area Under the Receiver Operating Characteristic curve (AUC-ROC) are evaluated and compared with traditional techniques. The overall accuracy of 0.972, F-score of 0.96, MCC of 0.957, and FNR of 0.025 are obtained by the developed scheme on identifying various cyberattacks effectively.</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 effective cyberattack classification in big data analytics using optimized stacked dilated convolutional gated bidirectional recurrent unit on spark architecture

  • 期刊论文

  • 发表日期：

    2026年4月21日

查看全文
  • G. Tagore Sai Prasad,
  • P. S. G. Aruna Sri

摘要

The digital transformation of modern society has intensified cyber threats, demanding advanced, real-time defense mechanisms. Traditional lack the capability to sustain pace in light of the scope and multifaceted nature of attacks, making AI-driven solutions vital. Leveraging Big Data enables these AI models to analyze vast, diverse datasets, such as network logs and user activities, to identify patterns and enhance predictive cybersecurity. This manuscript proposes an innovative, optimized Deep learning (DL) assisted Spark architecture for analyzing various cyberattacks in a big data IoT environment. Generally, a set of slave nodes and a master node are used to train the classes. Initially, the attack data sourced from the original dataset is prepared through the execution of data cleaning and data normalization using Pareto scaling to enhance the data quality. After preprocessing, the features are extracted, and multiclass classification is performed using the stacked dilated Convolutional gated bidirectional recurrent unit (SDC-GBRU) model, which classifies various attacks like MITM-ARP-spoofing, SSH-brute force attack, FTP-brute force, DDOS-ICMP, DDOS-RAWIP, DDOS-UDP, DOS attack, exploiting-FTP attack, fuzzing, ICMP flood, SYN-flood, port scanning, remote code execution, SQL injection, and XSS accurately. enhanced through the implementation of the Lyrebird optimization (LBO) strategy that streamlines network computation during the learning process. A Python simulation tool is devised for experimentation, and various performance measures like accuracy, F-score, Matthew’s correlation coefficient (MCC), F-score, false negative rate (FNR), and Area Under the Receiver Operating Characteristic curve (AUC-ROC) are evaluated and compared with traditional techniques. The overall accuracy of 0.972, F-score of 0.96, MCC of 0.957, and FNR of 0.025 are obtained by the developed scheme on identifying various cyberattacks effectively.