Anomaly Detection in Intranet File Distribution Systems Using Machine Learning Algorithms

This study investigates the application of machine learning (ML) algorithms for anomaly detection in such environments, addressing the limitations of traditional rule-based methods that often fail to adapt to evolving threats.In the absence of publicly available datasets, a synthetic dataset was created to simulate both normal and anomalous file transfer activities.Three ML algorithms were evaluated: Isolation Forest, One-Class Support Vector Machine (SVM), and Random Forest, using metrics such as accuracy, precision, recall, F1-score, and AUC-ROC.Results show that Random Forest achieved the highest performance with 97.5% accuracy and strong generalization, making it ideal for environments with labeled data.Isolation Forest demonstrated robust detection of unseen anomalies in unlabeled or privacy-sensitive contexts.One-Class SVM underperformed, largely due to parameter sensitivity and lower accuracy.These findings underscore the viability of both supervised and unsupervised ML models in detecting intranet-based file anomalies.The study concludes with recommendations for model selection based on data availability and privacy considerations, and proposes future work on real-time detection and deployment in live enterprise networks.