Dimensional Visualization Module for Interactive Machine Learning Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current supervised machine learning methods either rely solely on automated classification of unstructured data, leading to inappropriate results if the data does not fit the algorithm, or require laborious human intervention, which can be impractical.
Innovation Solution
A computer-implemented method and system that uses a dimensional visualization module with a global objective function to apply unsupervised dimension reduction and user-enhanced clustering, allowing users to input categorizations and visualize data for more effective clustering and classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If machine learning only is used to classify unstructured data, then speed is improved, but classification accuracy deteriorates
Solution Approach 1:
The patent introduces dimensional visualization as an intermediary between automated machine learning classification and human judgment. The system projects high-dimensional data into lower dimensions for human inspection, allowing users to identify misclassifications and provide feedback without manually examining all data points, thus maintaining speed while improving accuracy
Solution Approach 2:
The system implements a feedback loop where human users review visualized data points, provide correction feedback on misclassifications, and this feedback is used to iteratively improve the machine learning model's classification accuracy while maintaining efficient automated processing
2Measurement precision
If human intervention is used to classify unstructured data, then classification accuracy is improved, but productivity deteriorates
Solution Approach 1:
The patent extracts only the most ambiguous or problematic data points for human review by projecting them into lower-dimensional visual space, rather than requiring humans to manually classify all data points. This selective extraction maintains high accuracy for difficult cases while preserving overall productivity
Solution Approach 2:
The system creates a visual copy or representation of high-dimensional data in lower dimensions that humans can easily inspect. This visual copy allows rapid human judgment without requiring humans to process the full complexity of the original high-dimensional data, maintaining both accuracy and efficiency
3Ease of operation
If dimensionality reduction is applied to visualize data, then ease of operation is improved, but information loss increases
Solution Approach 1:
The patent transforms high-dimensional data into lower-dimensional visual representations that are easier for humans to interpret. By changing the dimensional representation rather than simplifying the data content, the system maintains essential information while improving visualizability and user interaction
Data Source
AI summary
A computer-implemented method and computer system for supervised machine learning and classification comprises applying, by a dimensional visualization module incorporating a global objective function, an unsupervised dimension reduction algorithm to a dataset including a plurality of data points capable of visual representation to produce a dimensionally reduced dataset and parsing, by the dimensional visualization module, the dimensionally reduced dataset to a user interface module for visual display of the dimensionally reduced dataset. The method comprises receiving a user input indicative of a categorization of at least one of the data points within the dataset, applying, by the dimensional visualization module, the global objective function weighted in accordance with the indicative categorization, to the dimensionally reduced dataset to produce a user-enhanced dimensionally reduced dataset, and parsing, by the dimensional visualization module, the user-enhanced dimensionally reduced dataset to the user interface module for visual display of the dimensionally reduced dataset.