Multi-Model Fusion Training for 3GPP Service Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing 3GPP service-oriented management architecture in 5G networks faces poor management effectiveness due to the lack of support for joint training and testing of multiple AI/ML models, leading to inadequate service management capabilities.
Innovation Solution
A method and apparatus for joint training and testing of multiple models by incorporating training and testing indication information to establish a fusion model, utilizing multi-model fusion algorithms and strategies, enhancing management service capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional single-model management method is used in 3GPP service-oriented architecture, then the system structure remains simple and easy to manage, but the management effectiveness and service management capability are poor
Solution Approach 1:
The patent merges multiple AI/ML models into a unified management framework within the 3GPP service-oriented architecture. The network element manages and coordinates multiple models (first model, second model, etc.) that work together to provide comprehensive service management capabilities, transforming the traditional single-model approach into a multi-model collaborative system that improves management effectiveness while maintaining architectural coherence
Solution Approach 2:
The patent implements a universal management framework that can handle multiple types of AI/ML models simultaneously. The network element provides multi-functional capabilities including model training, testing, deployment, and coordination across different models, allowing the system to manage diverse model types (classification, regression, clustering, etc.) through a single standardized interface, thereby improving management effectiveness without proportionally increasing complexity
2Adaptability or versatility
If multiple AI/ML models are introduced to improve service management capability, then the management ability and intelligence are enhanced, but the system complexity and difficulty of coordination increase
Solution Approach 1:
The patent segments the multi-model management system into distinct functional components: individual model training modules, model testing modules, and a central coordination mechanism. Each model can be trained and tested independently according to its specific requirements, while the network element provides centralized coordination to manage interactions between models. This segmentation allows high service management capability through model diversity while controlling complexity through modular organization
Solution Approach 2:
The patent introduces a network element as an intermediary that mediates between multiple AI/ML models and the 3GPP service-oriented architecture. This intermediary manages model training requests, coordinates model interactions, handles testing procedures, and standardizes communication between different models and external systems. The mediator role reduces coordination complexity by providing a unified interface and management layer, allowing multiple models to work together efficiently without direct peer-to-peer complexity
3Measurement precision
If joint training and testing of multiple models is implemented, then model performance and generalization are improved, but the training and testing process becomes more complex and resource-intensive
Solution Approach 1:
The patent implements preliminary action by enabling independent model training before joint deployment. Each AI/ML model can be trained separately on its specific dataset and optimized for its particular function first. The training indication information and model identifiers are prepared in advance, allowing models to reach individual maturity before being integrated into the joint multi-model system. This preliminary training approach improves final model performance while reducing the complexity of simultaneous joint training
Solution Approach 2:
The patent implements feedback mechanisms in the model training and testing process. The network element monitors training progress, model performance metrics, and testing results, using this feedback to adjust training parameters, coordinate model interactions, and optimize the overall system performance. Testing results feed back into the training process to improve model accuracy and generalization. This feedback-driven approach systematically manages training complexity while achieving high model performance through iterative optimization
Data Source
AI summary
The present disclosure provides a model training method and apparatus, a model testing method and apparatus, and a storage medium. The model training method includes: receiving a training request of a second device of a management service consumer, where the training request carries a training attribute, the training attribute includes: training indication information, the training indication information is used to indicate an association relationship between multiple models and a model generation strategy; and establishing a configuration instance in a first device, and jointly training the multiple models to obtain a training result according to the training indication information and multiple model identifiers of the training attribute, where the training result includes a trained fusion model. The present disclosure can realize joint training of the multiple models, obtain better model performance, enhance the intelligence of network operation and maintenance, and improve a management ability of devices in 3GPP.