Reward Once, Penalize Once: Rectifying Time Series Anomaly Detection
Reward Once, Penalize Once: Rectifying Time Series Anomaly Detection
复制标题
DOI：
10.1109/ijcnn55064.2022.9891913
复制
发表时间：
2022-03
期刊：
影响因子：
--
通讯作者：
Keval Doshi;Shatha Abudalou;Yasin Yılmaz
中科院分区：
文献类型：
--
作者：
Keval Doshi;Shatha Abudalou;Yasin Yılmaz
虽然时间序列中的异常检测已经是一个活跃的研究领域了好几年，但最近的方法采用了一个不充分的评估标准，导致F1分数膨胀。我们表明，一个基本的随机猜测方法可以优于国家的最先进的检测器在这个流行的，但错误的评价标准。在这项工作中，我们提出了一个适当的评价指标，衡量检测序列异常的及时性和精度。此外，大多数现有的方法无法从长序列中捕获时间特征。基于自注意力的方法，如transformers，已被证明在捕获远程依赖关系方面特别有效，同时在训练和推理期间具有计算效率。我们还提出了一个有效的Transformer方法在时间序列中的异常检测，并广泛评估我们提出的方法在几个流行的基准数据集。
While anomaly detection in time series has been an active area of research for several years, most recent approaches employ an inadequate evaluation criterion leading to an inflated F1 score. We show that a rudimentary Random Guess method can outperform state-of-the-art detectors in terms of this popular but faulty evaluation criterion. In this work, we propose a proper evaluation metric that measures the timeliness and precision of detecting sequential anomalies. Moreover, most existing approaches are unable to capture temporal features from long sequences. Self-attention based approaches, such as transformers, have been demonstrated to be particularly efficient in capturing long-range dependencies while being computationally efficient during training and inference. We also propose an efficient transformer approach for anomaly detection in time series and extensively evaluate our proposed approach on several popular benchmark datasets.