Anomaly detection has been extensively implemented in industry. The reality is that an application may have numerous scenarios where anomalies need to be monitored. However, the complete process of anomaly detection will take much time, including data acquisition, data processing, model training, and model deployment. In particular, some simple scenarios do not require building complex anomaly detection models. This results in a waste of resources. To solve these problems, we build an anomaly detection pipeline(ADOps) to modularize each step. For simple anomaly detection scenarios, no programming is required and new anomaly detection tasks can be created by simply modifying the configuration file. In addition, it can also improve the development efficiency of complex anomaly detection models. We show how users create anomaly detection tasks on the anomaly detection pipeline and how engineers use it to develop anomaly detection models.

Manifold alignment (MA) is a popular semi-supervised domain adaptation method. It requires the corresponding points across domains or the target labels, which may not exist in reality. In this paper, we suppose target domain has no prior information and investigate the unsupervised MA method. Instead of using target labels to construct the cross-domain relations, we utilize pseudo labels of target data to calculate the similarity and dissimilarity matrix between domains. However, the accuracy of pseudo labels may not be satisfactory, especially when the spectral drift is significant. Therefore, we propose a multi-classifiers consistency (MCC) approach to select a part of target data points that have more reliable predictions, and only these pseudo labels are utilized for aligning features across domains. This multi-classifiers consistency based unsupervised MA method is denoted as MCC_UMA. Its efficiency was demonstrated in experiments with three multi-temporal Hyperion hyperspectral remote sensing images.