FuseAD: Unsupervised Anomaly Detection in Streaming Sensors Data by Fusing Statistical and Deep Learning Models

Abstract

The need for robust unsupervised anomaly detection in streaming data is increasing rapidly in the current era of smart devices, where enormous data are gathered from numerous sensors. These sensors record the internal state of a machine, the external environment, and the interaction of machines with other machines and humans. It is of prime importance to leverage this information in order to minimize downtime of machines, or even avoid downtime completely by constant monitoring. Since each device generates a different type of streaming data, it is normally the case that a specific kind of anomaly detection technique performs better than the others depending on the data type. For some types of data and use-cases, statistical anomaly detection techniques work better, whereas for others, deep learning-based techniques are preferred. In this paper, we present a novel anomaly detection technique, FuseAD, which takes advantage of both statistical and deep-learning-based approaches by fusing them together in a residual fashion. The obtained results show an increase in area under the curve (AUC) as compared to state-of-the-art anomaly detection methods when FuseAD is tested on a publicly available dataset (Yahoo Webscope benchmark). The obtained results advocate that this fusion-based technique can obtain the best of both worlds by combining their strengths and complementing their weaknesses. We also perform an ablation study to quantify the contribution of the individual components in FuseAD, i.e., the statistical ARIMA model as well as the deep-learning-based convolutional neural network (CNN) model.

A Survey on Data-Driven Predictive Maintenance for the Railway Industry

“…Recent works have addressed anomaly detection for PdM supported by learning strategies on sequential data [ 2 , 39 , 106 , 115 , 116 , 117 , 118 ]. In the last few years, several papers were published approaching Anomaly Detection with Time-Series data applied to the most different domains, including industry, public water, and energy systems, among many others [ 1 , 109 , 112 , 114 , 118 , 119 , 120 , 121 , 122 , 123 , 124 , 125 , 126 , 127 , 128 , 129 , 130 , 131 , 132 , 133 , 134 , 135 , 136 , 137 , 138 , 139 , 140 ].…”

A Hybrid Deep Learning Framework for Unsupervised Anomaly Detection in Multivariate Spatio-Temporal Data

“…Munir at el. [31] propose a novel approach for anomaly detection in streaming sensors data which takes advantage of both statistical and deep learning-based techniques. They use statistical ARIMA model and a convolutional neural network (CNN) model to forecast the next timestamp in a given time-series.…”

A Novel Deep Learning Framework Based RNN-SAE for Fault Detection of Electrical Gas Generator

“…For industrial applications, early fault detection can enhance the identification of hazard patterns and achieve reliable maintenance for mechanical equipment [18]. The explosion of industrial data has emphasized the necessity for the DL technique in many applications, for instance, malware detection [19], multi-sensor fusion [20], anomaly detection [21], and soft sensor modeling [15].…”