InterpretableSAD: Interpretable Anomaly Detection in Sequential Log Data
Abstract
We propose InterpretableSAD, a framework for anomaly detection in sequential log data with built-in interpretability. The model detects abnormal subsequences while providing explanations through interpretable prototypes. Experiments on system log datasets demonstrate that InterpretableSAD achieves competitive detection accuracy while offering human-understandable explanations.
Type
Publication
In IEEE International Conference on Big Data (Big Data 2021)
Citation
@inproceedings{han2021interpretablesad,
title={InterpretableSAD: Interpretable anomaly detection in sequential log data},
author={Han, Xiao and Cheng, He and Xu, Depeng and Yuan, Shuhan},
booktitle={2021 IEEE International Conference on Big Data (Big Data)},
pages={1183--1192},
year={2021},
organization={IEEE}
}
