Anomaly Factor Estimation Device Using Per-Metric Z-Score Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection algorithms struggle to identify which observation metric is causing an anomaly, as they output a single dimension of 'anomalies' rather than the degree of anomaly per metric, making it difficult to determine the cause of detected anomalies.
Innovation Solution
A device comprising a detection unit to identify anomalies based on trained numerical vectors and a computing unit that calculates the deviation of each metric from its average value divided by its standard deviation to estimate the cause of the anomaly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If anomaly detection is performed using a single aggregated anomaly score, then the detection accuracy is improved, but the ability to identify the cause of the anomaly is lost
Solution Approach 1:
The patent segments the single aggregated anomaly score into multiple per-metric anomaly scores. Instead of treating all metrics uniformly with one anomaly score, the system calculates separate anomaly scores for each metric by comparing individual metric values against their respective normal ranges. This segmentation preserves the information about which specific metrics are anomalous, enabling cause identification while maintaining detection accuracy.
2Adaptability or versatility
If the number of observed metrics is increased, then the comprehensiveness of monitoring is improved, but the probability of false anomaly detection increases
Solution Approach 1:
The patent applies local quality by establishing individual normal ranges for each metric based on its historical data, rather than using a uniform threshold for all metrics. Each metric is evaluated against its own customized normal range, allowing the system to comprehensively monitor multiple metrics while reducing false anomalies. This local customization ensures that each metric's unique characteristics are respected in the anomaly detection process.
Data Source
AI summary
A device for estimating a cause of an anomaly comprises: a detection unit to detect an anomaly in a detection target based on a learner trained on first numerical vectors obtained from a detection target when the detection target is under a normal condition and second numerical vectors to be obtained from the detection target at multiple time; and a first computing unit to compute, for each metric of a second numerical vector from which an anomaly has been detected, as information for estimating a metric of cause of the anomaly, a value obtained by subtracting, from a value of the metric, an average of the metric in the first numerical vectors, and dividing a result of the subtracting by standard deviation of the metric in the first numerical vectors. This device supports estimation of the cause of an anomaly detected in a target object for detecting an anomaly.