摘要
时间序列异常检测作为时间序列研究的重要组成部分,已经引起学术界和工业界的广泛关注和研究。针对时间序列数据中蕴含的深层局部特征和复杂的前后依赖关系,提出一种融合双重注意力机制的异常检测模型。该模型采用自编码器结构,由挤压激励注意力模块(SEAB)和概率稀疏自注意力模块(PSAB)组成编码器。SEAB通过利用动态加权窗口划分,为具有强可辨识性的子序列片段赋予更大的权重,使模型能够更加有效地挖掘出具有重要信息的局部特征。PSAB则采用稀疏自注意力机制,保留具有较高权重的点积,去除冗余的时序特征,降低了时间复杂度,从而捕获时间序列的长期依赖关系。实验结果表明,该模型在9个对比模型中取得了最高的F1分数0.97,并在14个测试数据集中有8个F1分数超过其他所有对比模型,因此可有效地识别时间序列数据中的异常情况,并具备先进的异常检测性能。
As an important part of time series research,time series anomaly detection has attracted extensive atten-tion and research in academia and industry.In view of the deep local features and complex dependency in time se-ries data,an anomaly detection model with dual attention mechanism is proposed.The model adopts autoencoder structure.The encoder is composed of a squeeze excitation attention block(SEAB)and a probsparse self-attention block(PSAB).SEAB mines local features containing important information by assigning greater weights to se-quence segments with strong discriminability using dynamic weighted window partitioning.PSAB adopts sparse self-attention mechanism to retain dot products with higher weights,eliminate redundant timing features,and reduce time complexity,so as to capture the long-term dependence of time series.Experimental results show that the pro-posed model achieves the highest F1 score of 0.97 among 9 comparison models and outperforms all other compari-son models in 8 of 14 tested datasets in terms of F1 score,which can effectively identify abnormal situation in time series data and achieve advanced anomaly detection performance.
作者
杨超城
严宣辉
陈容均
李汉章
YANG Chaocheng;YAN Xuanhui;CHEN Rongjun;LI Hanzhang(School of Computer and Cyberspace Security,Fujian Normal University,Fuzhou 350117,China;Fujian Environmental Monitoring Internet of Things Laboratory,Fujian Normal University,Fuzhou 350117,China)
出处
《计算机科学与探索》
CSCD
北大核心
2024年第3期740-754,共15页
Journal of Frontiers of Computer Science and Technology
基金
国家自然科学基金面上项目(61976053)
福建省科技厅引导性项目(2020H0011,2023Y0012)。
