SAD: An Unsupervised System for Subsequence Anomaly Detection

Boniol, Paul; Linardi, Michele; Roncallo, Federico; Palpanas, Themis

doi:10.1109/icde48307.2020.00168

Cited by 25 publications

(20 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, auto-regression based approaches are rarely used for anomaly detection in high-order multivariate time series due to their inability to efficiently capture volatile time-series [1]. Other methods like SAND [10], CPOD [47] and Elle [28] utilize clustering and database read-write history to detect outliers.…”

Section: Related Workmentioning

confidence: 99%

“…Researchers have developed reconstruction-based methods that predominantly aim to encapsulate the temporal trends and predict the time-series data in an unsupervised fashion, then use the deviation of the prediction with the ground-truth data as anomaly scores. Based on various extreme value analysis methods, such approaches classify timestamps with high anomaly scores as abnormal [4,10,14,20,28,29,45,60,62]. The way prior works generate a predicted time-series from a given one varies from one work to another.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data

Tuli¹,

Casale²,

Jennings³

2022

Preprint

View full text Add to dashboard Cite

Efficient anomaly detection and diagnosis in multivariate timeseries data is of great importance for modern industrial applications. However, building a system that is able to quickly and accurately pinpoint anomalous observations is a challenging problem. This is due to the lack of anomaly labels, high data volatility and the demands of ultra-low inference times in modern applications. Despite the recent developments of deep learning approaches for anomaly detection, only a few of them can address all of these challenges. In this paper, we propose TranAD, a deep transformer network based anomaly detection and diagnosis model which uses attentionbased sequence encoders to swiftly perform inference with the knowledge of the broader temporal trends in the data. TranAD uses focus score-based self-conditioning to enable robust multi-modal feature extraction and adversarial training to gain stability. Additionally, model-agnostic meta learning (MAML) allows us to train the model using limited data. Extensive empirical studies on six publicly available datasets demonstrate that TranAD can outperform state-of-the-art baseline methods in detection and diagnosis performance with data and time-efficient training. Specifically, TranAD increases F1 scores by up to 17%, reducing training times by up to 99% compared to the baselines.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data

Tuli¹,

Casale²,

Jennings³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Real-time outlier detection in data streams is a growing research area [ 15 , 92 ], but only three of the surveyed outlier explanation techniques are designed for streaming environments. However, those techniques do not address some of the characteristics of data streams, such as concept drift and data uncertainty.…”

Section: Conclusion and Research Directionsmentioning

confidence: 99%

A survey on outlier explanations

et al. 2022

View full text Add to dashboard Cite

While many techniques for outlier detection have been proposed in the literature, the interpretation of detected outliers is often left to users. As a result, it is difficult for users to promptly take appropriate actions concerning the detected outliers. To lessen this difficulty, when outliers are identified, they should be presented together with their explanations. There are survey papers on outlier detection, but none exists for outlier explanations. To fill this gap, in this paper, we present a survey on outlier explanations in which meaningful knowledge is mined from anomalous data to explain them. We define different types of outlier explanations and discuss the challenges in generating each type. We review the existing outlier explanation techniques and discuss how they address the challenges. We also discuss the applications of outlier explanations and review the existing methods used to evaluate outlier explanations. Furthermore, we discuss possible future research directions.

show abstract

“…Many of the above-discussed algorithms are implemented in the SAD anomaly detection tool [9]. SAD allows data scientists to interactively discover anomalies in time series and compare the results of different algorithms.…”

Section: Sequence Anomaly Detectionmentioning

confidence: 99%

Distributed detection of sequential anomalies in univariate time series

2021

View full text Add to dashboard Cite

The automated detection of sequential anomalies in time series is an essential task for many applications, such as the monitoring of technical systems, fraud detection in high-frequency trading, or the early detection of disease symptoms. All these applications require the detection to find all sequential anomalies possibly fast on potentially very large time series. In other words, the detection needs to be effective, efficient and scalable w.r.t. the input size. Series2Graph is an effective solution based on graph embeddings that are robust against re-occurring anomalies and can discover sequential anomalies of arbitrary length and works without training data. Yet, Series2Graph is no t scalable due to its single-threaded approach; it cannot, in particular, process arbitrarily large sequences due to the memory constraints of a single machine. In this paper, we propose our distributed anomaly detection system, short DADS, which is an efficient and scalable adaptation of Series2Graph. Based on the actor programming model, DADS distributes the input time sequence, intermediate state and the computation to all processors of a cluster in a way that minimizes communication costs and synchronization barriers. Our evaluation shows that DADS is orders of magnitude faster than S2G, scales almost linearly with the number of processors in the cluster and can process much larger input sequences due to its scale-out property.

show abstract

SAD: An Unsupervised System for Subsequence Anomaly Detection

Cited by 25 publications

References 12 publications

TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data

TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data

A survey on outlier explanations

Distributed detection of sequential anomalies in univariate time series

Contact Info

Product

Resources

About