Ransomware Attack Detection using Intelligent Machine Learning Algorithms: A Systematic Review of Challenges and Future Directions

Abstract

Cybersecurity is a major concern in the IT sector these days. As a direct result of the increasing number of ransomware attacks and other types of cyberattacks, the protection of user data has become a primary focus. This systematic review explores the use of smart machine learning (ML) algorithms to identify ransomware. It demonstrates how these algorithms, towards their flexibility and precision, can be used for malware detection. A thorough survey of 30 primary studies from 2018 to 2025 reveals that advanced machine learning models like Random Forest, SVM, CNN, LSTM, XGBoost, and ensemble methods are the main contributors to achieving high detection accuracy results, very often beyond 99%, particularly when ransomware datasets are used. These models are able to detect the presence of ransomware in various platforms by utilizing static, dynamic, and hybrid features sets. The review also takes into account issues such as dataset imbalance, limitations of generalization, and detection strategies. The next goals are to create models that are resistant to adversarial attacks, employing generative augmentation (GANs), developing light machine learning architectures for real-time edge deployment, and using Explainable AI (XAI) to make AI more transparent. The findings demonstrate that ML models with high accuracy that are well-optimized and contextualized can significantly enhance the efficacy of cutting-edge ransomware defense systems. In sum, this study serves as a detailed and organized guide from the perspective of researchers and practitioners for the production of high-precision, robust, and scalable ransomware detection systems.