Contextualized Anomaly Analyzer Using Reference Waveform Spaces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Anomaly detection in high-dimensional data is challenging due to the high dimensionality causing each data point to be considered an outlier, leading to inefficiencies in identifying true anomalies.
Innovation Solution
A method involving remapping high-dimensional input data into the frequency domain using the Goertzel algorithm, followed by wave model expansion and hypercomplex feature vectors to identify anomalies based on context spaces defined by reference waveforms, utilizing a complex and hypercomplex feature vector transformation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-dimensional data is analyzed directly, then data aggregation in space can be performed, but each data point is considered an outlier leading to false anomaly detection
Solution Approach 1:
The patent transforms data from high-dimensional space to a different representation space using wave model expansion and hypercomplex feature vectors. This dimensionality transformation allows the system to capture contextual patterns while avoiding the curse of dimensionality where every point becomes an outlier.
Solution Approach 2:
The system changes the parameters of data representation by applying Goertzel algorithm for frequency domain transformation and using wave model expansion. These parameter changes convert raw high-dimensional data into a form that preserves anomaly characteristics while reducing false positives.
2Productivity
If traditional anomaly detection methods are used, then simple outlier identification can be performed, but efficiency in identifying true anomalies deteriorates due to high dimensionality
Solution Approach 1:
The system performs preliminary transformation of data into wave model representation and hypercomplex feature vectors before anomaly detection. This preliminary action prepares the data in a form that enables more efficient and reliable anomaly identification by capturing contextual patterns upfront.
Solution Approach 2:
The patent introduces wave model expansion and hypercomplex feature vectors as intermediary representations between raw data and anomaly detection. These intermediaries transform the data into a space where true anomalies can be efficiently identified without being overwhelmed by high dimensionality.
3Measurement precision
If contextual information is incorporated for anomaly detection, then detection accuracy improves, but computational complexity increases
Solution Approach 1:
The patent segments contextual information extraction by using wave model expansion that decomposes signals into frequency components. This segmentation allows contextual patterns to be captured in an organized manner, reducing computational complexity compared to analyzing all contextual relationships simultaneously.
Data Source
AI summary
Methods, systems and apparatuses may provide for technology that transforms input data into a set of reference waveforms, defines a context space for the set of reference waveforms, and determines whether sample data is an anomaly based on the context space.