Back to Search
Start Over
LogBD: A Log Anomaly Detection Method Based on Pretrained Models and Domain Adaptation.
- Source :
- Applied Sciences (2076-3417); Jul2023, Vol. 13 Issue 13, p7739, 20p
- Publication Year :
- 2023
-
Abstract
- The log data generated during operation of a software system contain information about the system, and using logs for anomaly detection can detect system failures in a timely manner. Most existing log anomaly detection methods are specific to a particular system, have cold-start problems, and are sensitive to updates in log format. In this paper, we propose a log anomaly detection method LogBD based on pretrained models and domain adaptation, which uses the pretraining model BERT to learn the semantic information of logs. This method can solve problems caused by the multiple meaning of words and log statement updates. The distance to determine anomalies in LogBD is constructed on the basis of domain adaptation, using TCNs to extract common features of different system logs and mapping them to the same hypersphere space. Lastly, experiments were conducted on two publicly available datasets to evaluate the method. The experimental results showed that the method can better solve the log instability problem and exhibits some improvement in the cross-system log anomaly detection effect. [ABSTRACT FROM AUTHOR]
- Subjects :
- SYSTEM failures
POLYSEMY
SEMANTICS
SYSTEMS software
Subjects
Details
- Language :
- English
- ISSN :
- 20763417
- Volume :
- 13
- Issue :
- 13
- Database :
- Complementary Index
- Journal :
- Applied Sciences (2076-3417)
- Publication Type :
- Academic Journal
- Accession number :
- 164921463
- Full Text :
- https://doi.org/10.3390/app13137739