Grouped AI Models for Multi-Dimensional Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional anomaly detection methods in computing systems often result in high false positives due to one-dimensional data analysis, necessitating a more comprehensive and multi-dimensional approach to accurately identify system anomalies.
Innovation Solution
A method involving grouping of multiple artificial intelligence models based on common identification fields, setting combination conditions, and determining weights to accurately detect anomalies by combining detection results, with the option to transmit detailed alarm messages to administrators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If one-dimensional data analysis is used for anomaly detection, then the detection process is simple, but false positives occur frequently
Solution Approach 1:
The patent transitions from one-dimensional data analysis to multi-dimensional analysis by grouping multiple AI models that analyze different dimensions of system data. Each model processes specific monitoring items (CPU usage, memory usage, network traffic, etc.) and their results are combined to form a comprehensive anomaly detection decision, thereby improving detection accuracy while maintaining operational simplicity through automated grouping and result aggregation.
2Reliability
If multiple AI models are used for comprehensive anomaly detection, then detection accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the anomaly detection system into multiple independent AI models, each specialized in detecting specific types of anomalies or analyzing particular monitoring items. These segmented models are then grouped based on their detection targets and combination conditions, allowing the system to achieve comprehensive coverage while maintaining modular simplicity and ease of management.
Solution Approach 2:
The patent creates a universal model grouping mechanism that can accommodate multiple different AI models with various detection capabilities. The grouping system serves multiple functions: organizing models by detection target, determining combination conditions, aggregating detection results, and generating comprehensive anomaly decisions. This multi-functional approach allows the system to handle diverse anomaly detection scenarios through a single unified framework.
3Reliability
If detection results from multiple models are combined, then comprehensive determination is achieved, but processing time increases
Solution Approach 1:
The patent performs preliminary actions by pre-defining model groups and their combination conditions before actual anomaly detection occurs. Models are pre-grouped based on their detection targets and relationships, and combination rules are established in advance. During runtime, the system only needs to execute the pre-planned aggregation logic, significantly reducing processing time while still achieving comprehensive anomaly determination through multi-model result combination.
Data Source
AI summary
Provided is a method for detecting an anomaly and a system, to which the method is applied. The anomaly detection method according to the embodiment of the present disclosure is a method performed by at least one computing device and comprises obtaining a plurality of models trained to detect an anomaly for different monitoring items, wherein input data of the models include at least one identification field for identifying an anomaly detection target, forming at least one model group by grouping models having a common identification field in the input data among the plurality of models, and detecting an anomaly of a detection target identified by a common identification field of the model group based on a detection result of a model group, wherein the input data of the models may include at least one identification field for identifying an anomaly detection target.