Data Classification System Using Meta-Feature Clustering for Model Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated classification model selection techniques are time-consuming and resource-intensive, making it difficult to accurately classify large datasets in real-time, especially with the increasing volume of data.
Innovation Solution
The method involves generating meta-features from labeled datasets using clustering indices, correlating them with a prebuilt model to estimate classification performance scores, and selecting top models to build an ensemble classification model for efficient classification of unlabeled datasets, utilizing techniques like majority voting or model stacking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional classification model selection approach is used, then classification performance can be evaluated, but the process is time-consuming and resource-intensive
Solution Approach 1:
The patent pre-calculates and stores classification performance scores for multiple classification models across different datasets in a database before actual classification tasks. When a new classification task arises, the system retrieves pre-computed performance scores instead of training and evaluating models from scratch, significantly reducing model selection time while maintaining reliable performance evaluation
Solution Approach 2:
The patent creates a database that stores replicated performance score data from previous model evaluations. Instead of re-evaluating all classification models on all datasets, the system copies and retrieves relevant pre-computed performance scores, transforming a computationally intensive process into a quick data retrieval operation that maintains evaluation accuracy
2Reliability
If traditional classification model selection approach is used, then classification accuracy can be achieved, but computational resources are consumed excessively
Solution Approach 1:
The system performs computationally intensive model training and performance evaluation in advance during data preprocessing phases, storing the results in a database. During actual classification tasks, the system retrieves pre-computed performance scores, shifting the computational burden from real-time classification to offline preprocessing, thereby reducing energy consumption during critical classification operations
Solution Approach 2:
The patent stores replicated performance score data in a database that can be quickly accessed during classification tasks. This copying of performance data eliminates the need to re-compute performance metrics for each classification task, significantly reducing computational resource requirements while maintaining accurate model selection
3Extent of automation
If automated classification model selection techniques are used, then user intervention is reduced, but the process remains time-consuming for large datasets
Solution Approach 1:
The patent automatically pre-calculates and stores classification performance scores for multiple models during data preprocessing. When classification tasks arrive, the system automatically retrieves pre-computed scores from the database, enabling both automated model selection and high classification speed without requiring real-time computational intensive operations
Solution Approach 2:
The system copies pre-computed performance scores from a database into memory for quick retrieval during classification tasks. This copying approach enables automated model selection to proceed at high speed by replacing computationally intensive real-time evaluation with rapid data retrieval operations
Data Source
AI summary
The present disclosure describes a method and system (120) for data classification. The system (120) comprises at least one processor (230) coupled to memory (240) and configured to receive at least one first dataset comprising at least one labeled dataset and at least one unlabeled dataset and process the received labeled dataset to generate at least one first meta feature comprising cluster indices. The processor (230) is further configured to estimate a classification performance score of each of a plurality of classification models for the at least one labeled dataset by correlating the generated meta features with a prebuilt model. The processor (230) is further configured to generate a list comprising the classification models arranged in descending order of the estimated performance scores and select a top N classification models from the list to build an ensemble classification model for classifying the at least one unlabeled dataset.