Model-centered Ensemble for Anomaly Detection in Time Series

Time-series anomalies detection is a fast-growing area of study, due to the exponential growth of new data produced by sensors in many different contexts as the Internet of Things (IOT). Many predictive models have been proposed, and they provide promising results in differentiating normal and anomalous points in a timeseries. In this paper, we aim to find and combine the best models on detecting anomalous time series, so that their different strategies or parameters can contribute to the time series analysis. We propose TSPME-AD (stands for Time Series Prediction Model Ensemble for Anomaly Detection). TSPME-AD is a model-centered based ensemble that trains some of the state-of-the-art predictive models with different hyper-parameters and combines their anomaly scores with a weighted function. The efficacy of our proposal was demonstrated in two real-world time-series datasets, power demand, and electrocardiogram.

[1]  Lovekesh Vig,et al.  LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection , 2016, ArXiv.

[2]  Eamonn J. Keogh,et al.  Finding the most unusual time series subsequence: algorithms and applications , 2006, Knowledge and Information Systems.

[3]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[4]  VARUN CHANDOLA,et al.  Anomaly detection: A survey , 2009, CSUR.

[5]  Lovekesh Vig,et al.  Long Short Term Memory Networks for Anomaly Detection in Time Series , 2015, ESANN.

[6]  Simon S. Woo,et al.  Detecting Anomalies in Space using Multivariate Convolutional LSTM with Mixtures of Probabilistic PCA , 2019, KDD.

[7]  Minqiang Xu,et al.  An Unsupervised Anomaly Detection Approach for Spacecraft Based on Normal Behavior Clustering , 2012, 2012 Fifth International Conference on Intelligent Computation Technology and Automation.

[8]  Jing Gao,et al.  Converting Output Scores from Outlier Detection Algorithms into Probability Estimates , 2006, Sixth International Conference on Data Mining (ICDM'06).

[9]  Xing Wang,et al.  Exact variable-length anomaly detection algorithm for univariate and multivariate time series , 2018, Data Mining and Knowledge Discovery.

[10]  David L. Iverson.  Inductive System Health Monitoring , 2004, IC-AI.

[11]  Jiri Matas,et al.  On Combining Classifiers , 1998, IEEE Trans. Pattern Anal. Mach. Intell..

[12]  Christian S. Jensen,et al.  Outlier Detection for Multidimensional Time Series Using Deep Neural Networks , 2018, 2018 19th IEEE International Conference on Mobile Data Management (MDM).

[13]  Charu C. Aggarwal,et al.  Outlier ensembles: position paper , 2013, SKDD.

[14]  Feng Xia,et al.  LoTAD: long-term traffic anomaly detection based on crowdsourced bus trajectory data , 2017, World Wide Web.

[15]  Zhixiao Wang,et al.  An overview on trajectory outlier detection , 2019, Artificial Intelligence Review.