DRAM Failure Prediction in AIOps: Empirical Evaluation, Challenges and Opportunities

Wu, Zhiyue; Xu, Hongzuo; Pang, Guansong; Yu, Fengyuan; Wang, Yijie; Jian, Songlei; Wang, Yongjun

doi:10.48550/arxiv.2104.15052

Cited by 1 publication

(1 citation statement)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Over recent decades, with the burgeoning of informatization, a substantial amount of time series data have been continuously created. As the functioning status of various target systems such as large-scale data centers [49], cloud servers [19], space crafts [20], and even human bodies [30], these time series data are a source where we can monitor and alarm potential faults, threats, and risks of target systems by identifying their unusual states (i.e., anomalies). Anomaly detection, an important field in data mining and analytics, is to find exceptional data observations that deviate significantly from the majority [34], which is playing a critical role in achieving this goal.…”

Section: Introductionmentioning

confidence: 99%

Calibrated One-class Classification for Unsupervised Time Series Anomaly Detection

Xu¹,

Wang²,

Jian³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Unsupervised time series anomaly detection is instrumental in monitoring and alarming potential faults of target systems in various domains. Current state-of-the-art time series anomaly detectors mainly focus on devising advanced neural network structures and new reconstruction/prediction learning objectives to learn data normality (normal patterns and behaviors) as accurately as possible. However, these one-class learning methods can be deceived by unknown anomalies in the training data (i.e., anomaly contamination). Further, their normality learning also lacks knowledge about the anomalies of interest. Consequently, they often learn a biased, inaccurate normality boundary. This paper proposes a novel one-class learning approach, named calibrated one-class classification, to tackle this problem. Our one-class classifier is calibrated in two ways: (1) by adaptively penalizing uncertain predictions, which helps eliminate the impact of anomaly contamination while accentuating the predictions that the one-class model is confident in, and (2) by discriminating the normal samples from native anomaly examples that are generated to simulate genuine time series abnormal behaviors on the basis of original data. These two calibrations result in contamination-tolerant, anomaly-informed one-class learning, yielding a significantly improved normality modeling. Extensive experiments on six real-world datasets show that our model substantially outperforms twelve state-of-the-art competitors and obtains 6% -31% 𝐹 1 score improvement. The source code is available at https://github.com/xuhongzuo/couta.

show abstract