Hierarchical Trained Model Selection for Medical Image Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information processing systems for medical image data analysis lack an efficient method to select and utilize a plurality of trained models with hierarchical relationships, which hinders accurate and efficient inference processing.
Innovation Solution
An information processing apparatus is designed to store and select a plurality of trained models with hierarchical relationships, allowing for the acquisition of inference information, selection of appropriate models based on this information, and notification of the selected models for enhanced inference processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple trained models are stored for different class hierarchies, then inference accuracy is improved, but device complexity increases
Solution Approach 1:
The system segments the model selection process by dividing models into different class hierarchies (first class hierarchy and second class hierarchy). The selection unit selectively applies models based on the specific inference target, using the hierarchical structure to organize and manage multiple models efficiently without overwhelming complexity.
Solution Approach 2:
The system performs preliminary organization of models into hierarchical classes before inference. The storage unit pre-arranges multiple trained models according to their class hierarchies, and the selection unit pre-identifies the appropriate model based on the inference target information, reducing decision complexity during actual inference.
2Productivity
If model selection is performed based on inference information, then inference efficiency is improved, but processing time increases
Solution Approach 1:
The selection unit applies different selection strategies based on the local characteristics of the inference target. By analyzing the specific inference information and matching it with the appropriate class hierarchy level, the system selects the most suitable model for that specific task, optimizing the balance between selection time and inference efficiency.
Data Source
AI summary
An information processing apparatus according to an exemplary embodiment of the present disclosure includes a storage unit configured to store a plurality of trained models including a first trained model for classifying medical image data into a class belonging to a first class hierarchy, and a second trained model for classifying the medical image data into a class belonging to a second class hierarchy lower than the first class hierarchy, an acquisition unit configured to acquire information about inference, a selection unit configured to select a plurality of trained models having a hierarchical relationship from among the stored plurality of trained models, based on the acquired information about inference, and a notification unit configured to provide a notification of the selected plurality of trained models.