An Anomaly Detection Framework for System Logs Based on Ensemble Learning

Logs offer vital insights into system states and contextual details, crucial for identifying anomalies. Numerous machine learning and deep learning approaches have been proposed for log anomaly detection. Recent studies reveal that distinct software systems tend to generate a substantial volume of complexity and diversity of logs that exhibit considerable discrepancies in class distribution. In this paper, we introduce IELog, a framework for anomaly detection. IELog employs DSS (Denoise Selection Sampling) to oversample the minority class, mitigating imbalanced data impact. Subsequently, IELog proposes the AW (Anomaly Weighting) ensemble rule to effectively combine the prediction outcomes of individual base models, leveraging their distinct strengths. Extensive experiments have been performed on four different public log datasets, which demonstrate the validity of the proposed framework IELog.