Time sequence anomaly detection method and device

Abstract

The embodiment of the invention provides a time sequence anomaly detection method and device. The method comprises the steps that a labeled data set containing labeled data is acquired; Obtaining a feature combination set corresponding to each piece of annotation data in the annotation data set; according to all the feature combination sets, all the annotation types corresponding to all the annotation data and preset feature combinations, acquiring candidate data similar to all the annotation data; extracting target data which can be used for optimizing a preset model from all the candidate data; determining an optimization processing strategy for the preset model according to all the target data; and carrying out abnormity detection on the to-be-detected time sequence through the optimized preset model. The device executes the method. According to the method and the device provided by the embodiment of the invention, the applicability of time sequence anomaly detection can be improved.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of time series, and in particular to a time series anomaly detection method and device. Background technique [0002] Time series refers to the numerical sequence obtained by arranging the values ​​of the same statistical index in chronological order (for example: the network throughput of a certain server per hour, the number of visits per minute of a certain website, etc.). Time series anomaly detection refers to detecting whether a value or a subsequence of values ​​in a time series deviates from the normal pattern. [0003] Existing methods for detecting whether time series are abnormal, mainly using statistical or machine learning algorithms for abnormal detection, generally have the following defects: (1) Models are established for specific scenarios or a certain type of detection objects with similar abnormal patterns, and the generality Not strong; (2) The false positive rate a...

AI Technical Summary

Problems solved by technology

[0003] Existing methods for detecting whether time series are abnormal, mainly using statistical or machine learning algorithms for abnormal detection, generally have the following defects: (1) Models are established for specific scenarios or a certain type of detection objects with similar abnormal patterns, and the generality Not strong; (2) The false positive rate and false negative rate are high; (3) It is necessary to manually label more time series to be detected, which is inefficient

Images