Machine Learning Model Generation Support Apparatus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models require individual tuning for each device, making it difficult to generate a related model when the algorithm changes, especially in scenarios where a common algorithm is used across multiple devices.
Innovation Solution
A generation support apparatus that stores component information and management data for machine learning models, allowing for the recording of updates and generation of configuration information for a new model with the same component versions as the previous model, facilitating the creation of a related machine learning model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If individual tuning is performed for each machine learning model to monitor abnormalities of multiple devices, then the model accuracy for each device is improved, but the time and effort required for model generation increases significantly
Solution Approach 1:
The system performs preliminary actions by automatically generating initial machine learning models using historical component information and change histories before individual tuning is needed. This preliminary model generation reduces the time required for subsequent individual tuning while maintaining accuracy requirements.
Solution Approach 2:
The system creates copies of existing machine learning model configurations and component information to generate new models for different devices. By copying proven configurations and adapting them with minimal tuning, the system maintains model accuracy while significantly reducing generation time compared to creating models from scratch for each device.
2Ease of operation
If machine learning model configurations are managed without version control, then the management process is simpler, but it becomes difficult to track changes and generate related models when algorithms are updated
Solution Approach 1:
The system implements feedback mechanisms by automatically recording change histories of machine learning model components and using this information to generate updated models. The version control system provides feedback about what has changed, enabling automated model regeneration while maintaining simple operation through automation.
Solution Approach 2:
The system introduces an intermediary version control mechanism that manages the complexity of tracking changes without directly complicating the user interface. This intermediary layer automatically handles versioning, change detection, and model regeneration, maintaining operational simplicity while preventing information loss about model evolution.
3Reliability
If complete individual tuning is performed for each device's machine learning model, then the model performance is optimized, but the complexity of the overall system increases
Solution Approach 1:
The system segments machine learning model management into distinct components: common configuration elements that can be shared across devices, and device-specific tuning parameters. This segmentation allows standardized model generation for common functionality while maintaining optimized performance for individual devices, reducing overall system complexity.
Solution Approach 2:
The system implements universality by creating a framework where machine learning model configurations can serve multiple devices through shared component information and change histories. This multi-functional approach maintains optimized performance across different devices while reducing system complexity through standardized processes and reusable configurations.
Data Source
AI summary
Provided is a generation support apparatus including: a storage device that stores component tables for storing a change history of a plurality of component groups for a machine learning model, and a model component management table for storing versions for the plurality of component groups; a model change recorder that inputs configuration information on an updated first machine learning model, adds the change history to the component tables, and records a version of an updated component group in the model component management table; and a derived model parameter synthesizer that generates a model parameter including a component corresponding to a version of an updated component group of the first machine learning model, for a second machine learning model including a component group of the same version as a component group before updating of the first machine learning model.