Scholarlink

<p>In the modern era, cyber-attacks have drawn international attention owing to the wide gamut of interferences they cause, embracing data breaches, identity theft, and the violation of user privacy. Defending systems against such growing threats and ensuring the protection of sensitive information continues to be a daunting challenge. Attack Detection Systems (ADS) are designed to identify malicious behavior in network environments; however, most existing systems rely on predefined attack patterns and lack the flexibility to detect the ever-evolving and complex nature of modern cyber threats. To overcome these limitations, this research introduces a fine-grained hybrid learning framework that combines a robust feature selection mechanism with an efficient classification strategy. The proposed model integrates Chaotic Theory with Tasmanian Devil Optimization (TDO) to perform effective feature selection, which is followed by a Multi-Layered Extreme Learning Machine (ML-ELM) for enhanced classification. The overall system is based on Python 3.19 and Scikit-Learn V2.0 and tested on the NSL-KDD (National Security Language-Knowledge Discovery in Databases) dataset. To demonstrate the merit of the recommended schema, performance measures such as accuracy, precision, recall, specificity, and F1-score are examined against the residing meta-heuristic optimized framework. The proposed model has a high accuracy of 98.78%, a precision of 98.56%, and an F1-score of 98.7%, confirming its effectiveness. Additionally, statistical validation is performed using the Shapiro-Wilk and Wilcoxon Signed-Rank tests, proving the reliability and stability of the proposed system. The ensemble framework demonstrates strong adaptability and efficiency in managing complex cyber threats, making it a potentially viable option for ongoing 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

A novel fine-grained intelligent framework for mitigating cyber-attacks through hybrid chaos-triggered tasmanian devil feature optimization and feedforward learning networks

  • 期刊论文

  • 发表日期：

    2025年9月19日

查看全文
  • Vandana Dharmapuri,
  • Sushama Rani Dutta

摘要

In the modern era, cyber-attacks have drawn international attention owing to the wide gamut of interferences they cause, embracing data breaches, identity theft, and the violation of user privacy. Defending systems against such growing threats and ensuring the protection of sensitive information continues to be a daunting challenge. Attack Detection Systems (ADS) are designed to identify malicious behavior in network environments; however, most existing systems rely on predefined attack patterns and lack the flexibility to detect the ever-evolving and complex nature of modern cyber threats. To overcome these limitations, this research introduces a fine-grained hybrid learning framework that combines a robust feature selection mechanism with an efficient classification strategy. The proposed model integrates Chaotic Theory with Tasmanian Devil Optimization (TDO) to perform effective feature selection, which is followed by a Multi-Layered Extreme Learning Machine (ML-ELM) for enhanced classification. The overall system is based on Python 3.19 and Scikit-Learn V2.0 and tested on the NSL-KDD (National Security Language-Knowledge Discovery in Databases) dataset. To demonstrate the merit of the recommended schema, performance measures such as accuracy, precision, recall, specificity, and F1-score are examined against the residing meta-heuristic optimized framework. The proposed model has a high accuracy of 98.78%, a precision of 98.56%, and an F1-score of 98.7%, confirming its effectiveness. Additionally, statistical validation is performed using the Shapiro-Wilk and Wilcoxon Signed-Rank tests, proving the reliability and stability of the proposed system. The ensemble framework demonstrates strong adaptability and efficiency in managing complex cyber threats, making it a potentially viable option for ongoing security systems.