Hybrid AI Model Selection by Removing Redundant Predictors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hybrid model creation methods combine all AI models, including redundant ones that are not complementary, leading to reduced accuracy.
Innovation Solution
A method that pools multiple AI models, selects and combines two or more models to create hybrid model candidates, and chooses the most accurate candidate based on comparison, excluding models with low prediction accuracy or strong correlations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If all AI models are combined to create a hybrid model, then the number of models used increases, but prediction accuracy decreases due to inclusion of redundant models
Solution Approach 1:
The patent extracts and removes redundant models from the pool of AI models before creating the hybrid model. By calculating similarity between models and identifying those with high similarity ratios, the system extracts only the necessary complementary models, thereby maintaining high prediction accuracy while reducing the number of models used in the final hybrid model.
2Adaptability or versatility
If redundant models are included in the hybrid model, then more models are available for prediction, but the model combination becomes less efficient and accuracy is reduced
Solution Approach 1:
The patent changes the parameter of model selection by introducing similarity calculation and threshold-based filtering. Instead of using all available models, the system calculates similarity metrics between models and applies parameter thresholds to select only those models that meet the complementary criteria, thereby improving prediction efficiency while maintaining adaptability through careful model selection.
3Adaptability or versatility
If all models are used without selection, then model coverage is maximized, but computational resources are wasted on processing redundant models
Solution Approach 1:
The patent performs preliminary actions by calculating similarity between models and identifying redundant models before the actual hybrid model creation process. This preliminary filtering step prevents unnecessary computational resources from being wasted on processing redundant models during the main prediction task, while still maintaining comprehensive model coverage through the systematic selection process.
Data Source
AI summary
First, a plurality of models that predict categories of input data are pooled. At least one of the plurality of models is a model trained by machine learning. Next, each of a plurality of hybrid model candidates that judge the categories are created by selecting and combining two or more models from among the plurality of pooled models. Then, by comparing the plurality of hybrid model candidates, one of the plurality of hybrid model candidates is selected as a hybrid model.