AI/ML Model Adjustment for Deployment Node Compatibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
AI/ML models generated by one node may not adapt to the software and hardware environment of the node on which they are deployed, leading to low running efficiency or failure.
Innovation Solution
A method and apparatus for adjusting AI/ML models by considering the capability and environment of the deployment node, involving devices that send and adjust the model based on capability information to enhance adaptation and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If an AI/ML model is generated by one node and deployed on another node without adjustment, then the model generation process is simple and fast, but the model does not adapt to the software and hardware environment of the deployment node, resulting in low running efficiency or failure
Solution Approach 1:
The patent applies preliminary action by collecting capability information about the deployment node's software and hardware environment before generating the AI/ML model. The model is then adjusted based on this pre-collected information, ensuring it is optimized for the specific deployment environment before actual deployment occurs. This resolves the contradiction by preparing the model in advance with environment-specific optimizations.
Solution Approach 2:
The patent utilizes parameter changes by modifying the AI/ML model's parameters and configuration based on the capability information of the deployment node. Different hardware specifications, software versions, and environmental conditions trigger corresponding parameter adjustments in the model, allowing the same base model to adapt to various deployment scenarios while maintaining generation efficiency.
2Adaptability or versatility
If an AI/ML model is adjusted based on deployment node capability information, then the model adaptation to the deployment environment is improved, but the model adjustment process becomes more complex and time-consuming
Solution Approach 1:
The patent manages complexity by systematically organizing parameter changes based on capability information categories. Instead of arbitrary adjustments, the model modification follows structured parameter changes determined by specific hardware and software characteristics, making the adjustment process more predictable and manageable despite the increased complexity.
Solution Approach 2:
The patent applies local quality by making targeted adjustments to specific parts of the model based on local environmental factors. Rather than completely redesigning the model, only the necessary components are modified according to the deployment node's specific capabilities, reducing unnecessary complexity while maintaining essential adaptability.
3Productivity
If an AI/ML model is adjusted based on deployment node capability information, then the execution performance of the model is improved, but the time required for model adjustment increases
Solution Approach 1:
The patent recovers time loss by performing capability information collection and model adjustment preparation in advance, before the actual deployment is needed. This preliminary action allows the model to be pre-optimized for the target environment, reducing the time required at deployment while maintaining high execution performance.
Solution Approach 2:
The patent optimizes the balance between adjustment time and performance by implementing selective parameter changes based on the most critical capability factors. Not all parameters are adjusted equally - only those that have the greatest impact on execution performance are modified, reducing unnecessary adjustment time while preserving performance benefits.
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
This application relates to a method for adjusting an AI/ML model and an apparatus. A first device sends first information to a second device, where the first information includes information for requesting to adjust a first AI/ML model and/or capability information of a third device. The second device adjusts the first AI/ML model based on the first information, and sends second information, where the second information is information about an adjusted first AI/ML model. The third device receives the second information, and runs the adjusted first AI/ML model based on the second information. In embodiments of this application, when adjusting the first AI/ML model, the second device can consider an actual case of the device on which the first AI/ML model is deployed, so that an adjustment result can adapt to a software and hardware environment of the third device, to improve adaptation between the AI/ML model and the device on which the AI/ML model is deployed. In this way, execution performance of the AI/ML model can be improved.