Reconstruct Anomaly to Normal: Adversarial Learned and Latent Vector-constrained Autoencoder for Time-series Anomaly Detection
This addresses a critical limitation in unsupervised time-series anomaly detection for applications like medical monitoring, though it appears incremental as it builds on existing adversarial and autoencoder approaches.
The paper tackles the problem of time-series anomaly detection where existing methods often fail to detect anomalies because they reconstruct both normal and anomalous data well, leading to low reconstruction errors for anomalies. They propose RAN, which reconstructs anomalies to match normal data distribution, achieving higher AUC-ROC scores than other algorithms in experiments on datasets like ECG diagnosis.
Anomaly detection in time series has been widely researched and has important practical applications. In recent years, anomaly detection algorithms are mostly based on deep-learning generative models and use the reconstruction error to detect anomalies. They try to capture the distribution of normal data by reconstructing normal data in the training phase, then calculate the reconstruction error of test data to do anomaly detection. However, most of them only use the normal data in the training phase and can not ensure the reconstruction process of anomaly data. So, anomaly data can also be well reconstructed sometimes and gets low reconstruction error, which leads to the omission of anomalies. What's more, the neighbor information of data points in time series data has not been fully utilized in these algorithms. In this paper, we propose RAN based on the idea of Reconstruct Anomalies to Normal and apply it for unsupervised time series anomaly detection. To minimize the reconstruction error of normal data and maximize this of anomaly data, we do not just ensure normal data to reconstruct well, but also try to make the reconstruction of anomaly data consistent with the distribution of normal data, then anomalies will get higher reconstruction errors. We implement this idea by introducing the "imitated anomaly data" and combining a specially designed latent vector-constrained Autoencoder with the discriminator to construct an adversary network. Extensive experiments on time-series datasets from different scenes such as ECG diagnosis also show that RAN can detect meaningful anomalies, and it outperforms other algorithms in terms of AUC-ROC.