ML Model Configuration for Consistent Tuning Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for developing machine learning models for testing and measurement of electronic devices face challenges in efficiently managing data consistency, training updates, and model deployment across manufacturing lines, particularly in environments like optical transmitter tuning systems.
Innovation Solution
The implementation of a structured data store within the OptaML™ ML Tools application, which interfaces with user automation software, maintains consistency between training and runtime requirements, supports quick updates, and enables the deployment of trained models across multiple devices, using a combination of local and global data stores to optimize data management and visualization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a structured data store is implemented to maintain data consistency between training and runtime, then data consistency is improved, but system complexity increases
Solution Approach 1:
The data store is segmented into multiple distinct stores including training data store, reference parameter store, trained models store, and communication store. Each store has a specific function and data structure optimized for its purpose. This segmentation maintains data consistency by ensuring that data flows through defined pathways between stores while reducing overall system complexity through modular organization.
Solution Approach 2:
The structured data store acts as an intermediary layer between the machine learning system and the test automation application. It provides standardized interfaces and data formats that mediate between training requirements and runtime requirements, ensuring consistency without requiring direct complex interactions between the ML system and automation software.
2Productivity
If local and global data stores are used to optimize data management, then data management efficiency is improved, but system complexity increases
Solution Approach 1:
The system implements both local data stores (on individual manufacturing devices) and global data stores (centralized). Local stores provide fast access to frequently used data like trained models and reference parameters for immediate tuning operations, while the global store manages training data and coordinates updates across the network. This hierarchical approach optimizes data management efficiency by placing data closer to where it is needed while maintaining centralized control.
3Adaptability or versatility
If quick updates to trained models are supported, then adaptability is improved, but system stability may worsen
Solution Approach 1:
The system performs preliminary actions by maintaining a structured repository of trained models, reference parameters, and training data that are pre-organized and validated. When updates are needed, the system can quickly retrieve and deploy pre-validated models from the trained models store without requiring complex real-time validation, thus enabling quick updates while maintaining stability through pre-established data quality controls.
Solution Approach 2:
The communication store and structured data architecture enable feedback mechanisms where model performance is monitored and update effectiveness is tracked. This feedback loop allows the system to quickly adapt by incorporating new training data and retrained models while maintaining stability through continuous validation against reference parameters and historical performance data stored in the structured data stores.
Data Source
AI summary
A manufacturing system has a machine learning (ML) system having one or more neural networks and a configuration file associated with a trained neural network (NN), a structured data store having interfaces to the ML system a test automation application, a training store, a reference parameter store, a communications store, a trained model store, and one or more processors to control the data store to receive and store training data, allow the ML system to access the training data to train the one or more NNs, receive and store reference parameters and to access the reference parameters, receive and store prediction requests for optimal tuning parameters and associated data within the communication store, to provide requests to the ML system, allow the ML system to store trained NNs in the trained models store, and to recall a selected trained NN and provide the prediction to the test automation application.