Anomaly Detection in Cybersecurity: Evaluating Machine Learning Models on the BETH Dataset

Anomaly detection is a critical task in cybersecurity, aimed at safeguarding against cyber threats by identifying unusual or harmful activity within network frameworks. In this context, Machine Learning (ML) models are employed to detect irregularities and potential security risks efficiently. Herein, an experimental study for anomaly detection is carried out on the BETH dataset by employing ML models. Validation and test reports for these models are presented, along with computed AUROC values for comparison. The inclusion of additional features and the consideration of variations in dataset densities have significantly improved the performance of one of the ML models.