Machine Learning Zero-Value Segmentation with Binary Behavior Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning models fail to accurately account for zero values, leading to inconsistent and unreliable predictions due to skewed statistics and high false positive rates, particularly in scenarios involving behaviors like medication diversion and financial crimes.
Innovation Solution
A method that converts zero and non-zero values into binary feature maps, segments these maps into behavior profiles, and merges segments to ensure a minimum size for robust statistical analysis, using unsupervised learning to predict specific outcomes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional machine learning models use zero values in feature vectors, then the models can represent absence of behavior, but the zero values heavily shift relevant statistics and cause inconsistent predictions
Solution Approach 1:
The patent extracts zero values from the original feature vectors and creates a separate binary feature map that explicitly tracks the presence or absence of each behavior. This separation allows the model to preserve the information about absent behaviors without allowing zero values to skew the statistical distributions of non-zero values, thereby maintaining both information representation and prediction reliability.
Solution Approach 2:
The patent transforms the continuous or categorical feature values into binary parameters (0 or 1) that indicate presence or absence of behaviors. This parameter transformation changes the nature of the data from values that can heavily influence statistical calculations to binary indicators that preserve information without distorting the distributions of actual behavior measurements.
2Reliability
If zero values are neglected in feature vectors, then statistics are not skewed, but information about absence of behavior is lost
Solution Approach 1:
The patent segments the feature representation into two distinct components: the original feature values (which maintain statistical accuracy) and a binary feature map (which preserves information about behavior absence). This segmentation allows both statistical reliability and information completeness to be maintained simultaneously by assigning different roles to different parts of the feature representation.
3Device complexity
If conventional models process all feature values uniformly, then the processing is simple, but predictions are inconsistent and false positive rates are high
Solution Approach 1:
The patent segments the feature processing into uniform processing of original feature values and separate processing of binary behavior indicators. This segmentation allows the model to handle different types of information appropriately, improving prediction consistency while maintaining reasonable processing complexity through systematic rather than ad-hoc handling of different feature types.
Data Source
AI summary
A method includes generating a plurality of binary feature maps containing a set of feature map values including a first binary value and/or a second binary value, by at least converting each input value of a set of input values of a plurality of input feature vectors to the first binary value when the corresponding input value is the zero value or the second binary value when the corresponding input value is the non-zero value. The method includes segmenting the plurality of binary feature maps into a plurality of segments representing behavior profiles. Each segment includes at least one subsegment in which the set of feature map values is the same for all binary feature maps in the at least one subsegment. The method includes predicting, based on a segment of the plurality of segments, a specific outcome. Related methods and articles of manufacture are also disclosed.