Confidence-Guided Label Augmentation for ML Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models face challenges in accurately predicting outcomes due to limited training data, as unknown data lacking known labels cannot be directly utilized, limiting their predictive accuracy.
Innovation Solution
A method and system that incorporate a confidence model to evaluate the accuracy of predicted labels for unknown data, selecting a subset with high confidence scores for training, thereby augmenting the available training data and improving model accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If unknown data without known labels is used for training, then the training dataset size increases, but the prediction accuracy deteriorates due to unreliable labels
Solution Approach 1:
The trained prediction model generates its own training data by predicting labels for unknown data and evaluating the confidence of these predictions. The model serves itself by identifying which unknown data points are suitable for training based on its own confidence scores, eliminating the need for external manual labeling while maintaining data quality
Solution Approach 2:
A confidence model provides feedback on the quality of predicted labels by generating confidence scores. This feedback mechanism allows the system to selectively use only those unknown data points with high confidence scores for training, creating a closed-loop system that continuously improves prediction accuracy while expanding the training dataset
2Measurement precision
If only accepted data with known labels is used for training, then the prediction accuracy is maintained, but the training dataset size remains limited
Solution Approach 1:
The system performs preliminary training on accepted data with known labels to establish an initial prediction model. This preliminary action creates a foundation that can then be used to evaluate and select unknown data for expansion, allowing the training dataset to grow while maintaining quality standards
Solution Approach 2:
The confidence model acts as an intermediary between the prediction model and the training data selection process. It mediates by evaluating predicted labels and determining which unknown data points are reliable enough to be added to the training set, bridging the gap between maintaining accuracy and expanding data volume
Data Source
AI summary
A method including training, using accepted data having known labels, an untrained prediction model to generate a trained prediction model. The method also includes generating, using the trained prediction model, predicted labels for unknown data. The method also includes generating, using a confidence model, confidence scores for the predicted labels. The method also includes selecting, using the confidence scores, a subset of the unknown data and confident labels. The confident labels include ones of the predicted labels for the subset of the unknown data. The method also includes training, using the subset of the unknown data and the confident labels, the trained prediction model to generate an updated prediction model.