Auto Tuning Anomaly Detection via Dynamic Contextual Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection algorithms fail to account for dynamic thresholds and contextual variations, leading to false anomalies and inability to handle sustained or recurring outlier values, which can result in missed issues or unnecessary alarms.
Innovation Solution
A computing device establishes context metric keys based on location, time, and device type, and dynamically updates metric ranges and anomaly windows to adapt to changing data trends, distinguishing contextual variations and recurring outliers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If static thresholds are used in anomaly detection algorithms, then the algorithm is simple to implement, but it produces false alarms and cannot handle contextual variations and recurring outliers
Solution Approach 1:
The patent transforms static anomaly detection thresholds into dynamic thresholds that automatically adapt to contextual variations. The system learns normal behavioral patterns for different contexts (time periods, locations, devices) and adjusts thresholds accordingly, allowing the same algorithm to handle both simplicity and contextual accuracy without requiring manual recalibration for each scenario.
Solution Approach 2:
The system changes the parameter of thresholds from fixed static values to dynamic values that vary based on contextual parameters such as time of day, location, and device type. This allows the anomaly detection to adapt to recurring patterns like nighttime data transmissions or location-specific temperature ranges, reducing false alarms while maintaining detection sensitivity.
2Reliability
If dynamic thresholds and contextual analysis are implemented, then anomaly detection accuracy improves, but algorithm complexity increases
Solution Approach 1:
The anomaly detection system performs self-learning and self-adjustment by automatically analyzing historical data to establish contextual patterns and threshold ranges. Rather than requiring manual configuration of complex parameters, the system autonomously adapts to different contexts and updates its detection criteria, reducing the operational complexity despite the enhanced analytical capabilities.
Solution Approach 2:
The patent segments the anomaly detection process into distinct phases: a training phase where contextual patterns are learned from historical data, and an evaluation phase where anomalies are detected using the established patterns. This segmentation simplifies the overall algorithm by breaking down the complex task of contextual adaptation into manageable, sequential steps that can be implemented efficiently.
3Reliability
If contextual metric keys are established for different locations, times, and device types, then false alarms are reduced, but data processing complexity increases
Solution Approach 1:
The patent creates a universal contextual metric key system that handles multiple contexts (locations, times, device types) through a unified framework. Rather than implementing separate detection algorithms for each context, the system uses a single multi-functional approach that automatically identifies and applies the appropriate contextual patterns, reducing processing complexity while maintaining high reliability across diverse scenarios.
Data Source
AI summary
Automatic tuning anomaly detection is described. The context metric keys are established during a training phase based on the surrounding context of data received from devices over time. Anomaly and tuning windows are also established for metric ranges of the context metric keys. After the training phase, incoming data is correlated against the keys to identify sets of the data associated with certain context metric keys. For any given context metric key, metric data values in the associated set of data fall either within or outside the metric range of the context metric key. If they fall outside the range for longer than the anomaly window, an alarm is raised. If they fall outside the range for longer than the tuning window, boundaries for the metric range are updated. Additionally, the contextual parameters of the context metric keys are also updated over time, as new data contexts appear.