Multivariate Event Detection via Segmented Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in efficiently detecting and classifying anomalies, change-points, patterns, and outliers in large-scale, multivariate datasets, limiting their ability to diagnose root causes and patterns effectively.
Innovation Solution
A method and system for anomaly and event analysis that utilize a detection model to identify types of event observations, generate anomaly records, and automatically link event observations to create event records, enabling efficient detection and visualization of anomalies, change-points, patterns, and outliers through a centralized platform and recommendation engine.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional event detection methods are used to represent data as either an event or not, then the system is simple to operate, but the ability to diagnose root causes and patterns is limited
Solution Approach 1:
The patent segments event detection into multiple specialized detection models, each trained to identify specific types of events (anomalies, change-points, patterns, outliers). This segmentation allows the system to capture diverse event types without creating a single overly complex system, preserving diagnostic information while managing complexity through modular architecture.
Solution Approach 2:
The patent adds a classification dimension to event detection by not only identifying whether data represents an event but also categorizing the type of event. This dimensional expansion transforms binary event detection into multi-class classification, enabling root cause diagnosis and pattern recognition while maintaining operational simplicity through automated classification.
2Measurement precision
If multiple independent event training data sets are used to train detection models, then the accuracy of event identification is improved, but the computational complexity increases
Solution Approach 1:
The patent divides the training process into multiple independent models, each trained on specific event types. This segmentation allows parallel training and processing, reducing the computational burden on any single model while collectively achieving high accuracy across all event types through the ensemble of specialized detectors.
Solution Approach 2:
The patent creates multiple copies of detection models, each specialized for different event types. These model copies operate independently and can be processed in parallel, distributing computational complexity across multiple instances while maintaining high identification accuracy through the collective capability of the model ensemble.
3Productivity
If event observations are automatically linked to create event records, then the efficiency of anomaly analysis is improved, but the complexity of data processing increases
Solution Approach 1:
The patent merges multiple event observations into unified event records through automatic linking. This consolidation combines scattered anomaly data points into coherent event narratives, improving analysis efficiency by presenting integrated information while managing processing complexity through automated association algorithms that link observations based on temporal and contextual relationships.
Data Source
AI summary
Systems and methods of the present disclosure include at least one processor that receives a data set of a data stream from a data source, where the data set includes a time-varying data points. The processor determines event observations associated with data points of the time-varying data points based on a detection model to identify types of the event observations, including: i) anomalies, ii) change-points, iii) patterns, or iv) outliers. The processor generates anomaly records in an event data store based on the event observations and automatically generates event records for at least one of the anomaly records based on variables of at least one dimension of the time-varying data points, where the event record links one or more event observations. The processor automatically applies changes in the event record to each event observation of the one or more event observations based on the linking by the event record.