Modern threat landscapes continue to evolve with increasing sophistication, challenging traditional detection methodologies and necessitating innovative solutions capable of addressing complex adversarial tactics. A novel framework was developed to identify ransomware activity through multimodal execution path analysis, integrating high-dimensional embeddings and dynamic heuristic derivation mechanisms to capture behavioral patterns across diverse attack variants. The approach demonstrated high adaptability, effectively mitigating obfuscation strategies and polymorphic characteristics often employed by ransomware families to evade detection. Comprehensive experimental evaluations revealed significant advancements in precision, recall, and accuracy metrics compared to baseline techniques, particularly under conditions of variable encryption speeds and obfuscated execution flows. The framework achieved scalable and computationally efficient performance, ensuring robust applicability across a range of system configurations, from resource-constrained environments to high-performance infrastructures. Notable findings included reduced false positive rates and enhanced detection latency, even for ransomware families employing sophisticated encryption mechanisms. The modular design allowed seamless integration of additional modalities, enabling extensibility and future-proofing against emerging threat vectors. Quantitative analyses further highlighted the system's energy efficiency, emphasizing its practicality for deployment in environments with stringent operational constraints. The results underline the importance of integrating advanced computational techniques and dynamic adaptability to safeguard digital ecosystems from increasingly complex threats.
Ransomware and Artificial Intelligence: A Comprehensive Systematic Review of Reviews
Rajamangala University of Technology Phra Nakhon(皇家理工大学叻那那空校区);Rajamangala University of Technology Rattanakosin Wang Klai Kangwon Campus(皇家理工大学拉达纳哥信哇集康旺校区);Phetchabun Rajabhat University(碧差汶皇家大学);King Mongkut's University of Technology North Bangkok(泰国国王科技大学北曼谷校区);Suranaree University of Technology(素攀那瑞科技大学)
Differential Area Analysis for Ransomware Attack Detection within Mixed File Datasets
勒索软件的威胁在受影响的受害者人数以及受成功攻击影响的个人和组织所产生的成本方面都在继续增长。在大多数情况下,一旦受害者受到攻击,他们只有两种行动方案:要么支付赎金,要么丢失数据。所有加密勒索软件菌株之间共享的一个共同行为是,在执行过程中的某个时候,它们会尝试加密用户的文件。Penrose et al.(2013); Zhao et al.(2011)之前的研究强调了使用香农熵区分压缩和加密文件的困难,因为两种文件类型表现出相似的值。本文描述的实验之一显示了加密文件头片段的香农熵的独特特性。这个特性被用来区分加密文件和其他高熵文件,如档案。这一发现在文件分类模型的开发中得到了利用,该模型使用了正在分析的文件的熵曲线与从随机数据生成的熵曲线之间的差异区域。当将被分析的文件的熵曲线值与由包含纯随机数的文件生成的熵曲线值进行比较时,曲线的相关性越大,被分析的文件包含加密数据的置信度就越高。
Explainable AI-Driven Cyber Risk Analytics and Model Reliability Assessment for Intelligent Governance of U.S. Critical Infrastructure: An XGBoost and SHAP-Based Intrusion Detection Framework
Department of Business Information Systems, Central Michigan University(中央密歇根大学商业信息系统系);Department of Information Studies, Trine University(特林大学信息学系);Department of Computer Science, Maharishi International University(Maharishi国际大学计算机科学系)