Unsupervised Clustering for Labeling Assistance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Labeling each piece of clustered data is a costly and time-consuming task, especially when dealing with large amounts of unlabeled data, such as video data or multiple sensor data, where determining common points becomes complex.
Innovation Solution
A labeling assistance system that uses unsupervised learning to generate clusters of data, searches for common points within each cluster, and outputs information on these common points to assist in labeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data is classified into clusters based on extracted features, then data classification is achieved, but labeling each piece of clustered data becomes highly costly
Solution Approach 1:
The patent extracts and outputs only the common points (representative features) from each cluster rather than requiring labeling of all individual data points. This extraction approach reduces the labeling workload from entire clusters to just their representative common points, significantly reducing time loss while maintaining classification accuracy.
Solution Approach 2:
The system creates a simplified representation (copy) of each cluster through common points that capture the essential characteristics. Instead of labeling original complex data points, the system works with these simplified copies that represent the cluster's core features, reducing labeling complexity and time.
2Loss of information
If video data or multiple sensor data is classified, then comprehensive data analysis is achieved, but determining which data to focus on becomes complex
Solution Approach 1:
The patent extracts common points that represent the essential characteristics of each cluster from multiple sensor data or video data. This extraction process identifies and isolates the most relevant features, reducing the complexity of determining which data to focus on while maintaining comprehensive analysis through the representative common points.
Solution Approach 2:
The system applies different processing approaches to different data types within clusters, focusing computational resources on identifying common points specific to each cluster's characteristics. This local quality approach simplifies the overall complexity by treating each cluster's data selection independently based on its specific features rather than applying a uniform complex selection process to all data.
Data Source
AI summary
The classification means 191 generates a plurality of clusters by classifying data to be labeled through unsupervised learning. The search means 192 searches for common points of the data included in each generated cluster. The output means 193 outputs information on the searched common points for each cluster.