Network Training Platform for ML Resource Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Beyond 5G networks, the lack of a platform to manage and share context among multiple Machine Learning (ML) applications leads to wastage of ML resources, network performance degradation, and increased costs due to repetitive training on the same data.
Innovation Solution
A network training platform that allows for the registration of ML applications, identifies similar applications based on parameters, and shares predicted data between them, thereby reducing redundant training and optimizing resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If each ML application is trained separately on the same data, then each application can be trained independently, but ML resources are wasted and network performance degrades
Solution Approach 1:
The patent merges multiple ML applications that require training on the same data into a single training process. The network training platform identifies applications with identical training requirements and combines them, so that one training operation serves multiple applications simultaneously, eliminating redundant resource consumption while maintaining independent application functionality
Solution Approach 2:
The patent creates a universal training mechanism where a single training process can serve multiple ML applications. The network training platform acts as a multi-functional system that handles registration, identification, comparison, and data sharing across different applications, allowing one training operation to fulfill multiple application needs
2Reliability
If repetitive training is performed on the same data, then each ML application receives dedicated training, but network performance degrades and operational costs increase
Solution Approach 1:
The patent implements preliminary identification and grouping of ML applications based on their training requirements before training execution. The network training platform pre-processes application registrations, compares their data requirements, and groups them in advance, so that when training is needed, it can be performed once for multiple applications rather than repeatedly for each one
3Adaptability or versatility
If no platform exists to manage multiple ML applications, then each application can be deployed freely, but context sharing is impossible and resources are wasted
Solution Approach 1:
The patent introduces a network training platform as an intermediary between multiple ML applications. This platform receives application registrations, identifies similarities in training requirements, and facilitates context sharing by providing trained data to multiple applications that need it, enabling information exchange without restricting application deployment flexibility
Data Source
AI summary
A method for sharing data between machine learning (ML) applications with a network training platform. The method includes: receiving a request to register a first ML application with the network training platform, wherein the request comprises first one or more parameters related to the first ML application; identifying at least one second ML application registered with the network training platform based on the first one or more parameters; identifying second one or more parameters related to the at least one second ML application; comparing the first one or more parameters with the second one or more parameters related to the at least one second ML application; and sharing, with the first ML application, predicted data corresponding to the at least one second ML application based on the comparing.