Fractal Pattern Extraction for AI Classifier Boundary Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning models face challenges in accurately determining the cluster assignment of new input points, especially in high-dimensional spaces, leading to high error rates and the need for human intervention due to inefficiencies in hyperplane and convex hull methods.
Innovation Solution
The method involves mapping output values during the training phase, identifying dense and sparse data portions, extracting fractal function base sets from dense areas, and applying these patterns to sparse data portions to improve classification accuracy without increasing training data volume or causing overfitting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hyperplanes or hyper convex hulls are used to define cluster boundaries, then boundary separation is achieved, but the ability to determine paths between points on the cluster boundary is lost
Solution Approach 1:
The patent pre-computes and stores fractal patterns from dense data areas during the training phase before actual classification occurs. These pre-extracted patterns are then reused to guide path determination between boundary points, eliminating the need for complex real-time calculations while maintaining accurate boundary separation.
Solution Approach 2:
The patent creates simplified copies of complex boundary patterns by extracting fractal features from dense data regions and applying them to sparse regions. These pattern copies enable efficient path determination between cluster boundary points without requiring the full complexity of the original data, thus improving operational ease while preserving measurement precision.
2Extent of automation
If probabilistic conditions or distance computations are used to identify points within clusters, then cluster assignment can be determined, but efficiency is reduced and human intervention is required
Solution Approach 1:
The system performs self-service by automatically extracting fractal patterns from dense data areas and applying them to classify sparse data points without human intervention. The automated pattern matching process eliminates the need for manual verification while maintaining high classification accuracy, thus improving both automation extent and productivity simultaneously.
Solution Approach 2:
The patent transforms the classification approach by changing from probabilistic distance computations to fractal pattern matching. This parameter change in the classification methodology enables fully automated operation with improved efficiency, as fractal patterns provide deterministic classification rules that do not require human judgment or iterative probabilistic calculations.
3Reliability
If large training data sets are used to ensure adequate boundary definition, then separation across clusters is improved, but the complexity of determining output points increases
Solution Approach 1:
The patent extracts only the essential fractal patterns from large training data sets during the training phase, separating the critical boundary-defining features from the bulk data. This extraction process maintains reliable boundary definition while significantly reducing the complexity of output determination, as the compressed fractal patterns contain the necessary information without requiring processing of the entire original data set.
Solution Approach 2:
The patent performs preliminary extraction of fractal patterns from dense data areas during the training phase, preparing simplified representations before the actual classification task. This preliminary action reduces the complexity of subsequent output determination while preserving the reliability of boundary definition, as the pre-extracted patterns encapsulate the essential boundary characteristics.
Data Source
AI summary
The classifier of an artificial intelligence model is trained by mapping output values in a final run of a training phase, measuring dense data portion and sparse data portion of a data plot produced by the training phase by identifying outputs near edges of the data plot and extracting the curve patterns as linear functions in the dense areas of the output; obtaining a fractal function base set of patterns from the linear functions provided by the dense areas of the output; applying the fractal function base set of patterns from the dense areas of the output to the sparse data portion of the data plot; and training the artificial intelligence model using the data plot including the dense areas of the output and the sparse data portion of the data plot that has been fit to the data curve using the fractal function base set of patterns.