Shared AI Model Prioritization Across Multiple Vendor Apps

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional machine learning methods rely on external servers for model training, leading to inaccurate and low-quality results due to the lack of direct user data processing, which compromises the accuracy and quality of machine learning outcomes.

Innovation Solution

An electronic device trains an artificial intelligence model using user data from multiple applications through confidential computing technology, enabling local data processing and enhancing the quality and accuracy of results by integrating a shared AI model that operates within a confidential computing environment.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If user data is transmitted to an external server for machine learning processing, then the device can perform machine learning tasks, but the accuracy and quality of machine learning results deteriorate due to processing only model parameters rather than actual user data

Engineering Contradiction:

Improveaccuracy of machine learning resultsVSAvoidcomplexity of data processing architecture

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces a confidential computing environment as an intermediary layer between the user data and the machine learning model. This mediator enables the server to process actual user data for training while maintaining security through cryptographic protection, thus improving result accuracy without requiring complex local processing architecture at the device end

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the conventional mechanical data transmission system with a cryptographic system. Instead of directly transmitting sensitive user data to servers, the system uses encrypted computing environments where data remains protected throughout the processing pipeline, substituting physical security mechanisms with mathematical cryptography

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If user data is processed locally on the electronic device, then the accuracy of machine learning results improves, but the security and confidentiality of user data may be compromised

Engineering Contradiction:

Improvequality of machine learning resultsVSAvoidrisk to data confidentiality

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent creates a confidential computing environment that acts as an inert or secure atmosphere for data processing. Within this isolated cryptographic environment, user data can be processed locally on the device with improved accuracy while the environment itself protects against external threats and unauthorized access, eliminating the confidentiality risk

Inventive Principle:

Principle #39Inert atmosphere (Inert environment)

3Reliability

If conventional federated learning is used to protect user privacy, then data security is maintained, but machine learning accuracy deteriorates because only model parameters are processed rather than actual user data

Engineering Contradiction:

Improvedata securityVSAvoidaccuracy of learning results

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent inverts the conventional federated learning approach. Instead of sending only model parameters to the server for aggregation, the system enables local processing of actual user data within a confidential computing environment and then securely shares only the learned insights or aggregated results, reversing the traditional data flow direction while maintaining security

Inventive Principle:

Principle #13The other way round (Inversion)

Data Source

PatentUS20250371423A1Electronic device providing results output through shared artificial intelligence model to multiple applications and method for controlling the same

Publication Date: 2025.12.04 SAMSUNG ELECTRONICS CO LTD

AI summary

An electronic device may comprise: a memory, and at least one processor, comprising processing circuitry. At least one processor, individually and/or collectively, may be configured to cause the electronic device to: transmit, to a shared application, first user input data input through a first application corresponding to a first vendor and second user input data input through a second application corresponding to a second vendor, stored in the memory, train a first artificial intelligence model of the shared application based on the first user input data and the second user input data, estimate results of a third user input data input through the first application or the second application, through the first artificial intelligence model, based on training the artificial intelligence model, and determine a priority for the estimated results and transmit information about the determined priority and the estimated results to the first application or the second application.