Transparent Ransomware Detection in Bitcoin Transactions: Leveraging Machine Learning and Explainable AI

Ransomware presents a significant threat within the Bitcoin ecosystem, necessitating robust detection methods. This study leverages machine learning techniques, particularly Random Forest Model, to identify ransomware activities in Bitcoin transactions using the BitcoinHeist dataset. We integrate explainable AI (XAI) tools, specifically LIME, to enhance model transparency and provide insights into predictions. Innovative feature engineering techniques are employed to capture critical transaction patterns, such as loops and volume changes, which improve detection accuracy. Our ensemble models achieve high performance metrics, including accuracy, precision, recall, and ROC-AUC scores. The use of XAI mitigates challenges like false positives and adapts to evolving ransomware tactics by elucidating the factors influencing model decisions. This research underscores the necessity of interpretability in ransomware detection systems, laying the groundwork for future studies focused on real-time detection and adaptive learning in cybersecurity.