Generative Content Attribution Without Selected Training Examples

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The challenge of automating high-quality media content generation and modification while ensuring accurate attribution and protecting intellectual property rights is hindered by the need for significant manual effort and the difficulty in enforcing proper attribution, leading to issues like plagiarism and copyright infringement.

Innovation Solution

A generative model is trained using a plurality of training examples from different sources, allowing for the generation of new media content without relying on specific training examples, and attributing the content to its source through analysis of textual and media content properties.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a generative model is trained using a plurality of training examples from different sources, then automated media content generation efficiency is improved, but the difficulty in enforcing proper attribution and protecting intellectual property rights worsens

Engineering Contradiction:

Improveautomated media content generation efficiencyVSAvoidattribution accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary actions by training the generative model with multiple sources of training examples and proactively generating attribution information along with the synthesized media content. This allows attribution to be established before any potential copyright issues arise, rather than attempting to enforce attribution after the fact.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback mechanisms by analyzing the synthesized media content to determine which training examples contributed to its creation, and then using this information to generate accurate attribution. This closed-loop approach ensures that attribution reflects the actual influences on the generated content.

Inventive Principle:

Principle #23Feedback

2Reliability

If manual effort is used to ensure high quality results in media content generation, then content integrity is improved, but the time consumption and operational complexity worsen

Engineering Contradiction:

Improvecontent integrityVSAvoidtime consumption

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system enables self-service by automatically analyzing the synthesized media content to determine its relationship to training examples and generating appropriate attribution information without requiring manual intervention. This maintains content integrity through systematic analysis while eliminating time-consuming manual processes.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system replaces manual mechanical processes with automated computational analysis. Instead of manually tracking and attributing training examples, the system uses machine learning models to automatically analyze and determine relationships between synthesized content and training data, significantly reducing time consumption while maintaining or improving accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Manufacturing precision

If selected training examples are used for generating content, then the quality of generated content is improved, but the risk of plagiarism and copyright infringement worsens

Engineering Contradiction:

Improvegenerated content qualityVSAvoidcopyright infringement risk

Core Design Contradiction:

Manufacturing precisionVSObject-affected harmful factors

Solution Approach 1:

The system changes the parameter of training data diversity by using a plurality of training examples from different sources rather than relying on selected training examples. This diversification maintains content quality through multiple influences while reducing copyright risk by avoiding over-reliance on any single source.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system introduces an intermediary mechanism in the form of automated attribution analysis that mediates between the use of training examples and copyright protection. By actively analyzing and attributing influences from multiple training sources, the system maintains quality while providing a defensive layer against copyright infringement accusations.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20250245440A1Generating content using a generative model without relying on selected training examples

Publication Date: 2025.07.31 BRIA ARTIFICIAL INTELLIGENCE LTD

AI summary

Systems, methods and non-transitory computer readable media for generating content using a generative model without relying on selected training examples are provided. An input indicative of a desire to generate a new content using a generative model may be received. The generative model may be a result of training a machine learning model using a plurality of training examples. Each training example of the plurality of training examples may be associated with a respective content. Further, an indication of a particular subgroup of at least one but not all of the plurality of training examples may be obtained. Based on the indication, the input and the generative model may be used to generate the new content, abstaining from basing the generation of the new content on any training example included in the particular subgroup. The new content may be provided.