Configuration Information for Two-Sided AI Model Training

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The dynamic and proprietary architecture of wireless networks often results in essential information being lost or modified during two-sided AI/ML model training, leading to inefficiencies in dataset sharing and model performance between network elements.

Innovation Solution

A method and apparatus that utilize configuration information to determine and transmit specific data sets for training between network elements, ensuring efficient and flexible two-sided model training by indicating what data to share, how to share it, and when, using machine learning algorithms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If dataset sharing is performed between network elements for two-sided AI/ML model training, then model training efficiency is improved, but information loss or modification occurs due to dynamic and proprietary architecture

Engineering Contradiction:
Improvemodel training efficiencyVSAvoidinformation loss or modification
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The patent introduces configuration information as an intermediary mechanism that mediates the dataset sharing process between network elements. This configuration information contains instructions about what data to share, how to share it, and when to share it, ensuring that the sharing process maintains data compatibility and prevents information loss while enabling efficient model training.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent applies preliminary action by establishing configuration information before dataset sharing occurs. The configuration information is sent in advance to instruct network elements on the proper sharing parameters, ensuring that when sharing happens, the correct data is shared in the correct format at the correct time, preventing information loss.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If configuration information is sent to instruct data sharing, then data compatibility and privacy considerations are improved, but communication overhead increases

Engineering Contradiction:
Improvedata compatibility and privacyVSAvoidcommunication overhead
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The configuration information is tailored to each specific network element and sharing scenario, providing localized instructions rather than generic ones. This allows the system to send only the necessary configuration details relevant to each element's specific needs, reducing overall communication overhead while maintaining data compatibility and privacy for each local context.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The configuration information dynamically adjusts parameters such as what data to share, how to share it, and when to share it based on specific requirements. By changing these parameters adaptively rather than using fixed configurations, the system minimizes communication overhead while ensuring data compatibility and privacy protection for each specific scenario.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4369253A1Dataset sharing transmission instructions for separated two-sided ai/ML based model training
Publication Date: 2024.05.15 NOKIA TECHNOLOGIES OY
  • EP4369253A1 patent drawingFigure 1
  • EP4369253A1 patent drawingFigure 2
  • EP4369253A1 patent drawingFigure 3

AI summary

The present disclosure relates to a method and a system for improving the establishment of a two-sided AI/ML based model implemented in an apparatus API - APX, such as an user equipment, UE, and a device DEI - DEY, such as a server or a gNodeB, gNB, in a wireless network system in which, by sending configuration information providing instructions regarding at least what to share for data set sharing between the apparatus AP1 - APX and the device DEI - DEY, an efficient separated or sequential training of the two-sided model can be generated.