Multivariate Anomaly Detection based on Prediction Intervals Constructed using Deep Learning

TL;DR

This paper presents a deep learning-based method for multivariate anomaly detection using prediction intervals, addressing challenges like computational complexity and false positives, and demonstrating competitive performance against traditional statistical models.

Contribution

The paper introduces a novel deep learning approach utilizing prediction intervals for anomaly detection that overcomes common issues of existing methods, such as false positives and the need for labeled data.

Findings

Deep learning models outperform traditional statistical methods in anomaly detection.

The proposed approach reduces false positives compared to existing deep learning techniques.

Benchmarking shows competitive or superior performance of deep learning architectures.

Abstract

It has been shown that deep learning models can under certain circumstances outperform traditional statistical methods at forecasting. Furthermore, various techniques have been developed for quantifying the forecast uncertainty (prediction intervals). In this paper, we utilize prediction intervals constructed with the aid of artificial neural networks to detect anomalies in the multivariate setting. Challenges with existing deep learning-based anomaly detection approaches include $(i)$ large sets of parameters that may be computationally intensive to tune, $(ii)$ returning too many false positives rendering the techniques impractical for use, $(iii)$ requiring labeled datasets for training which are often not prevalent in real life. Our approach overcomes these challenges. We benchmark our approach against the oft-preferred well-established statistical models. We focus on three deep learning architectures, namely, cascaded neural networks, reservoir computing and long short-term memory recurrent neural networks. Our finding is deep learning outperforms (or at the very least is competitive to) the latter.