TiSAT: Time Series Anomaly Transformer
This work addresses evaluation flaws and long-sequence handling in time series anomaly detection, offering improvements for researchers and practitioners in data analysis.
The paper tackled the problem of inflated evaluation in time series anomaly detection by proposing a new metric for timeliness and precision, and introduced an efficient transformer-based method, achieving competitive results on benchmark datasets.
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.