ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment

Zhao, Xiaowei; Guo, Kaiwei; Huang, Mingting; Qiu, Shaojian; Lu, Lu

doi:10.3390/computers14070272

Open AccessArticle

ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment

by

Xiaowei Zhao

^1,2,

Kaiwei Guo

³,

Mingting Huang

³,

Shaojian Qiu

^3,* and

Lu Lu

^1,*

¹

School of Computer Science and Engineering, South China University of Technology, Guangzhou 510006, China

²

Digital Research Department, China Southern Power Grid, Guangzhou 510663, China

³

College of Mathematics and Informatics, South China Agricultural University, Guangzhou 510642, China

^*

Authors to whom correspondence should be addressed.

Computers 2025, 14(7), 272; https://doi.org/10.3390/computers14070272

Submission received: 2 June 2025 / Revised: 29 June 2025 / Accepted: 3 July 2025 / Published: 10 July 2025

(This article belongs to the Special Issue Machine Learning Applications in Pattern Recognition)

Download

Browse Figures

Versions Notes

Abstract

Existing log-based anomaly detection methods typically require large volumes of labeled data for training, presenting significant challenges when applied to new systems with limited labeled data. This limitation has spurred the need for cross-system log anomaly detection (CSLAD) methods. However, current CSLAD approaches often face challenges in effectively handling distributional differences in log data across systems. To address this issue, we propose ELFA-Log, a transfer learning-based approach for cross-system log anomaly detection. By enhancing pseudo-label generation with uncertainty estimation and feature alignment, ELFA-Log improves detection performance even in the presence of data distribution shifts. It uses entropy-based metrics to generate high-confidence pseudo-labels, minimizing reliance on labeled data. Additionally, a distance-based loss function optimizes the shared representation of cross-system log features. Experimental results on benchmark datasets demonstrate that ELFA-Log enhances the performance of CSLAD, offering a practical solution to the challenge of high labeling costs in real-world applications.

Keywords: log anomaly detection; transfer learning; pseudo-labeling; uncertainty estimation

Share and Cite

MDPI and ACS Style

Zhao, X.; Guo, K.; Huang, M.; Qiu, S.; Lu, L. ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment. Computers 2025, 14, 272. https://doi.org/10.3390/computers14070272

AMA Style

Zhao X, Guo K, Huang M, Qiu S, Lu L. ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment. Computers. 2025; 14(7):272. https://doi.org/10.3390/computers14070272

Chicago/Turabian Style

Zhao, Xiaowei, Kaiwei Guo, Mingting Huang, Shaojian Qiu, and Lu Lu. 2025. "ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment" Computers 14, no. 7: 272. https://doi.org/10.3390/computers14070272

APA Style

Zhao, X., Guo, K., Huang, M., Qiu, S., & Lu, L. (2025). ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment. Computers, 14(7), 272. https://doi.org/10.3390/computers14070272

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Menu

ELFA-Log: Cross-System Log Anomaly Detection via Enhanced Pseudo-Labeling and Feature Alignment

Abstract

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI