Detecting Ransomware Threats Using Machine Learning Approach

The rise of ransomware-related cyber-attacks poses a severe threat to organizations across various sectors. This research addresses the urgent need for robust detection mechanisms using advanced machine learning (ML) models. Ransomware, characterized by irreversible encryption of user data, demands proactive strategies for identification and prevention. This study leverages publicly available ransomware and benign executable files from sources such as VirusTotal which are run in a sandboxed environment to obtain a dataset to train an ML model capable of differentiating between diverse ransomware samples and classifying them based on specific characteristics. The process begins with meticulous data preprocessing ensuring the dataset's quality. Feature extraction algorithms are employed to identify relevant characteristics, streamlining the dataset and optimizing model performance. The model is trained on the dataset to effectively learn patterns. Evaluation metrics, including accuracy, precision, recall, F1-score, and the ROC curve, provide comprehensive insights into the model's effectiveness.