Object Classification Using Prediction Consistency for Class Imbalance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image object classification models face challenges due to class imbalance, where unequal amounts of training data for different classes lead to degraded classification performance, particularly for classes with insufficient data.
Innovation Solution
A computing device uses a prediction consistency value to classify objects as majority or minority classes based on the consistency of class prediction values from multiple models, then applies suitable class predictors for each class type to improve accuracy, addressing class imbalance without additional training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single classification model is used for all classes, then the device complexity is reduced, but the classification accuracy for minority classes deteriorates due to class imbalance
Solution Approach 1:
The patent divides the classification task into two separate models: a majority class prediction model and a minority class prediction model. This segmentation allows each model to be optimized for its specific class type, improving classification accuracy for minority classes without significantly increasing overall system complexity.
Solution Approach 2:
The patent applies different prediction strategies locally based on the class type. For majority classes, ensemble models are used; for minority classes, target models with optimized training are applied. This local differentiation ensures that each class receives the most appropriate prediction approach, enhancing overall accuracy.
2Measurement precision
If multiple classification models are trained separately for different classes, then the classification accuracy for minority classes is improved, but the device complexity increases
Solution Approach 1:
The patent introduces a dynamic class type determination mechanism that selects between majority and minority class prediction models based on the input data characteristics. This dynamic approach allows the system to adapt to different class types without requiring a separate model for every possible class, thereby managing complexity while maintaining accuracy.
Solution Approach 2:
The patent creates a universal prediction framework that handles both majority and minority classes through a unified architecture. The system uses a single prediction consistency value calculation mechanism and a unified class type determination process that works for all classes, reducing the need for completely separate processing pipelines.
3Reliability
If ensemble models are used for all classes, then the prediction consistency is improved, but the classification performance for minority classes deteriorates due to insufficient training data
Solution Approach 1:
The patent applies different prediction strategies locally based on the class type. For majority classes, ensemble models are used; for minority classes, target models with optimized training are applied. This local differentiation ensures that each class receives the most appropriate prediction approach, enhancing overall accuracy.
Solution Approach 2:
The patent segments the prediction process into two distinct pathways: one for majority classes using ensemble models and another for minority classes using target models. This segmentation allows minority classes to receive focused attention and optimized training without being diluted by the ensemble approach.
4Measurement precision
If target models are trained with optimized parameters for specific classes, then the classification accuracy for those classes is improved, but the overall system complexity increases
Solution Approach 1:
The patent introduces a dynamic class type determination mechanism that selects between majority and minority class prediction models based on the input data characteristics. This dynamic approach allows the system to adapt to different class types without requiring a separate model for every possible class, thereby managing complexity while maintaining accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed is a method that includes generating a prediction consistency value that indicates a consistency of prediction of an object in an input image with respect to class prediction values for the object in an input image from classification models to which the input image is input, and determining a class of the object. Determining the class of the object includes, in response to a class type being determined, based on the prediction consistency value, of the object being determined to correspond to a majority class, determining a class of the object based on a corresponding class prediction value output for the object from a majority class prediction model, and in response to the class type of the object being determined to correspond to a minority class, determining the class of the object based on another corresponding class prediction value output for the object from a minority class prediction model.