Transformer-based Alarm Context-Vectorization Representation for Reliable Alarm Root Cause Identification in Optical Networks
Jinwei Jia, Danshi Wang, Chunyu Zhang, Hui Yang, Luyao Guan, Xue Chen, Min Zhang · 2021
A Transformer-based alarm context-vectorization representation technique is proposed for alarm root cause identification and correlation analysis. Three common root alarms are identified with an accuracy of 94.47%, and other correlated alarms are obtained with quantified correlation degrees.