Anomaly Detection Model Using Graph Laplacian Label Propagation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current anomaly detection techniques in industrial machinery and vehicles lack reliability due to insufficient utilization of past anomaly patterns and require arbitrary preprocessing, failing to effectively incorporate label information from both labeled and unlabeled samples.
Innovation Solution
The technique introduces the concept of similarity-based label information by expressing samples as multi-dimensional vectors with varying observation noise levels, using a graph Laplacian to determine an optimal linear transformation matrix and calculate anomaly scores, thereby reducing arbitrariness and enhancing detection reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If past anomaly patterns are incorporated into anomaly detection models, then detection reliability should improve, but arbitrary preprocessing is required which reduces effectiveness
Solution Approach 1:
The patent extracts and utilizes only the label information (normal/anomalous classification) from past anomaly patterns without requiring complex preprocessing of the underlying data. The method takes the essential categorical information and integrates it directly into the anomaly detection model through label propagation mechanisms, eliminating the need for arbitrary preprocessing while maintaining detection reliability.
Solution Approach 2:
The patent performs preliminary classification of data into labeled (normal/anomalous) and unlabeled samples before the actual anomaly detection process. This preliminary labeling action enables the model to learn from past patterns without requiring complex preprocessing, as the classification information is prepared in advance and can be directly utilized during model training and inference.
2Measurement precision
If label information from labeled and unlabeled samples is utilized, then detection accuracy improves, but the method becomes more complex
Solution Approach 1:
The patent creates a unified anomaly detection model that can handle multiple types of input data (labeled normal samples, labeled anomalous samples, and unlabeled samples) through a single integrated framework. The model uses universal label propagation mechanisms that work across all sample types without requiring separate processing procedures, thereby improving detection accuracy while avoiding increased methodological complexity.
Solution Approach 2:
The patent incorporates feedback mechanisms where the model continuously refines its anomaly detection capabilities by learning from both labeled and unlabeled samples. The label information from past anomalies and normal operations is fed back into the model during training, enabling iterative improvement of detection accuracy without requiring complex manual intervention or multiple separate methods.
Data Source
AI summary
A method providing an analytical technique introducing label information into an anomaly detection model. Effective utilization of label information is based on introducing the degree of similarity between samples. Assuming, for example, there is a degree of similarity between normally labeled samples and no similarity between normally labeled and abnormally labeled samples. Also each sensor value is generated by the linear sum of a latent variable and a coefficient vector specific to each sensor. However, the magnitude of observation noise is formulated to vary according to the label information for the sensor values, and set so that normal label≤unlabeled≤anomalously labeled. A graph Laplacian is created based on the degree of similarity between samples, and determines the optimal linear transformation matrix according to a gradient method. A optimal linear transformation matrix is used to calculate an anomaly score for each sensor in the test samples.