AI Model Revision Tracking for Microscope Task Re-Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI models for classifying tasks in precision assembly processes under a microscope require frequent updates due to variations in work processes and skill levels, making manual review of vast video data inefficient and burdensome.
Innovation Solution
A re-learning support system that displays design information for each trained AI model, associating version information to facilitate easy updating and reduce the burden of re-training by providing historical data and performance metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If AI models are frequently updated to adapt to variations in work processes and skill levels, then adaptability improves, but the time and effort required for re-training increases
Solution Approach 1:
The system performs preliminary actions by automatically archiving design information (training data, parameters, configuration) of previous AI model versions before updates occur. This preliminary archiving eliminates the need for manual re-training documentation, allowing developers to quickly reference historical models during re-training processes and significantly reducing the time required for model updates while maintaining adaptability to work process variations
Solution Approach 2:
The system implements feedback mechanisms by providing automated retrieval and comparison capabilities that allow developers to review performance metrics and design information from previous model versions. This feedback loop enables informed decision-making during re-training, reducing unnecessary re-training cycles and optimizing the balance between adaptability and re-training time
2Measurement precision
If manual review of vast video data is performed to update AI models, then measurement precision improves, but productivity decreases
Solution Approach 1:
The system enables self-service by automatically archiving and managing design information including training data, parameters, and configuration settings for each AI model version. This automated self-service capability eliminates the need for manual review and documentation processes, allowing the system to maintain measurement precision through automated tracking while significantly improving productivity by removing manual bottlenecks in the model update process
Solution Approach 2:
The system creates automated copies of design information from previous model versions, storing training data, parameters, and configuration settings in an organized archive. These automated copies enable developers to efficiently reference historical models during updates, maintaining classification accuracy through consistent reproduction of successful model configurations while dramatically improving update efficiency by eliminating manual copying and documentation work
3Loss of information
If comprehensive design information is stored for each AI model version, then information completeness improves, but device complexity increases
Solution Approach 1:
The system applies segmentation by dividing design information into distinct, organized components including training data, model parameters, configuration settings, and performance metrics for each AI model version. This segmented organization maintains information completeness by ensuring all necessary elements are captured and stored separately, while reducing system complexity through structured categorization that simplifies retrieval and management compared to storing undifferentiated comprehensive data
Data Source
AI summary
A storage stores one or more trained models trained using at least one microscopic image, one or more revisions associated with each of the trained models and indicating versions of the trained model, and one or more pieces of design information for the trained model associated with the one or more revisions. The processor receives a selection of at least one of the trained models, acquires one or more revisions and one or more pieces of design information for the selected trained model from a storage, displays a name of the selected trained model in a first display area of a model design information screen, and displays the acquired revisions and the acquired design information in association with each other in a second display area of the model design information screen.