LogOW: A semi-supervised log anomaly detection model in open-world setting

Jingwei Ye, Chunbo Liu, Zhaojun Gu, Zhikai Zhang, Xuying Meng, Weiyao Zhang, Yujun Zhang

Abstract: Highlights•Proposed LogOW model combining uncertainty estimation and incremental pre-training for enhanced log anomaly detection.•Demonstrated LogOW’s superior performance over state-of-the-art log anomaly detection models.•The practical impact of each module on enhancing anomaly detection performance was validated by ablation experiments.