LogBERT: Log Anomaly Detection via BERT

TL;DR

LogBERT introduces a BERT-based self-supervised framework for detecting anomalies in system logs, effectively learning normal patterns to identify deviations indicative of malfunctions or attacks.

Contribution

It presents a novel self-supervised training approach using BERT for log anomaly detection, outperforming existing methods on multiple datasets.

Findings

Outperforms state-of-the-art anomaly detection methods

Effective in identifying deviations in log sequences

Demonstrates robustness across different datasets

Abstract

Detecting anomalous events in online computer systems is crucial to protect the systems from malicious attacks or malfunctions. System logs, which record detailed information of computational events, are widely used for system status analysis. In this paper, we propose LogBERT, a self-supervised framework for log anomaly detection based on Bidirectional Encoder Representations from Transformers (BERT). LogBERT learns the patterns of normal log sequences by two novel self-supervised training tasks and is able to detect anomalies where the underlying patterns deviate from normal log sequences. The experimental results on three log datasets show that LogBERT outperforms state-of-the-art approaches for anomaly detection.