一句话:通过实证研究对比深度学习与经典机器学习在基于日志的异常检测中的性能与计算成本,揭示实际部署取舍。
本文开展实证研究,系统比较深度学习与经典机器学习在基于日志的异常检测中的表现与计算开销。问题在于:尽管深度学习方法检测性能强,其日志预处理、模型训练和推理耗时较长,难以满足在线分布式云系统快速部署日志异常检测服务的需求。研究强调需权衡检测效益与计算成本。由于摘要未披露具体数据集、模型与结果,本文的核心贡献在于以实证方式揭示这一权衡,并为实际系统选择方法提供依据。
While deep learning (DL) has emerged as a powerful technique, its benefits must be carefully considered in relation to computational costs. Specifically, although DL methods have achieved strong performance in log anomaly detection, they often require extended time for log preprocessing, model training, and model inference, hindering their adoption in online distributed cloud systems that require rapid deployment of log anomaly detection service.
DOI 原文 ·
@inproceedings{paperbot11416,
title = {Deep Learning or Classical Machine Learning? An Empirical Study on Log-Based Anomaly Detection},
author = {Boxi Yu and Jiayi Yao and Qiuai Fu and Zhiqing Zhong and Haotian Xie and Yaoliang Wu and Yuchi Ma and Pinjia He},
booktitle = {ICSE},
year = {2024},
doi = {10.1145/3597503.3623308}
}