AI Model Deployment via Capability-Based Encoder Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing communication technologies lack an efficient method for training and deploying AI/ML models in network devices and user equipment, particularly in determining and utilizing the AI and ML capabilities of user equipment.
Innovation Solution
A method and device for training and deploying AI/ML models, where a network device obtains capability information from user equipment, generates encoder and decoder models based on this information, and sends model information to the user equipment to deploy the models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If AI/ML models are deployed in network devices and user equipment, then communication technology is enhanced, but the complexity of determining and utilizing AI and ML capabilities increases
Solution Approach 1:
The network device obtains capability information from the user equipment in advance, before model training and deployment. This preliminary acquisition of capability information allows the system to prepare appropriate models according to the UE's AI/ML capabilities, avoiding complex capability determination during actual model deployment operations.
Solution Approach 2:
The capability information is segmented into specific components including AI supporting capability and ML supporting capability. This segmentation allows the network device to process and utilize different capability aspects independently, simplifying the overall complexity of capability determination while enabling targeted model selection and deployment.
2Productivity
If capability information is obtained and used to generate tailored encoder and decoder models, then model deployment efficiency is improved, but the time required for model training increases
Solution Approach 1:
The network device generates different encoder and decoder models by adjusting parameters based on the obtained capability information. Instead of training completely new models from scratch for each capability scenario, the system modifies existing model parameters to suit different UE capabilities, significantly reducing training time while maintaining deployment efficiency.
Solution Approach 2:
Base models are prepared and trained in advance at the network device. When capability information is obtained, the system performs faster adaptation by modifying these pre-trained models rather than training new models, thus reducing the time loss associated with model training while maintaining high deployment efficiency.
Data Source
AI summary
The present disclosure provides a method for training and deploying a model. The method may is performed by a network device and include: obtaining capability information reported by a user equipment (UE), the capability information being configured to indicate at least one of an artificial intelligence (AI) supporting capability or a machine learning (ML) supporting capability of the UE; generating an encoder model and a decoder model based on the capability information; and sending model information of the encoder model to the UE for the UE to deploy the encoder model.