Distributed-Ledger GAN Training for Private Multi-Party Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional approaches struggle to effectively train machine learning models for contract clause detection due to the challenge of accessing disparate data from multiple parties, leading to overfitting and limited detection capabilities.

Innovation Solution

A system utilizing a Generative Adversarial Network (GAN) combined with a distributed ledger, such as a blockchain, allows parties to contribute data for model training while maintaining data privacy and transparency, enabling model refinement through competitive data generation and reward mechanisms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If data from multiple parties is accessed for model training, then model detection capability is improved, but data privacy and security risks increase

Engineering Contradiction:

Improvemodel detection capabilityVSAvoiddata privacy risks

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

A blockchain-based intermediary system is introduced to mediate between data providers and the machine learning model. The blockchain stores hashed versions of training data and tracks data usage without exposing actual data content, enabling multi-party data contribution while maintaining privacy through cryptographic intermediaries.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

Instead of sharing actual training data, the system uses cryptographic hashes (copies of data fingerprints) stored on blockchain. These hashes verify data integrity and origin without revealing sensitive information, allowing the model to be trained on diverse data while preserving data privacy through information copying rather than direct access.

Inventive Principle:

Principle #26Copying

2Adaptability or versatility

If data from multiple parties is accessed for model training, then model detection capability is improved, but model overfitting increases

Engineering Contradiction:

Improvemodel detection capabilityVSAvoidmodel overfitting

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The training data is segmented into contributions from multiple independent parties, each adding diverse data samples. The blockchain tracks and verifies contributions from different sources, ensuring the model receives varied training examples that prevent overfitting to any single data source while improving overall detection capability.

Inventive Principle:

Principle #1Segmentation

3Reliability

If data usage is tracked transparently using distributed ledger, then data privacy is improved, but system complexity increases

Engineering Contradiction:

Improvedata privacyVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The blockchain system performs multiple functions simultaneously: it stores data integrity verification hashes, tracks data provenance, manages participant credentials, and enables transparent auditing. This multi-functional approach consolidates what would otherwise require separate systems into a single universal platform, reducing overall system complexity while maintaining privacy.

Inventive Principle:

Principle #6Universality (Multi-functionality)

4Ease of manufacture

If conventional approaches are used for contract analysis, then implementation simplicity is maintained, but clause detection accuracy deteriorates

Engineering Contradiction:

Improveimplementation simplicityVSAvoidclause detection accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

Conventional text-based contract analysis is replaced with a machine learning model trained on diverse contractual data. The model uses semantic understanding and pattern recognition capabilities to identify clauses with varying wording, achieving superior detection accuracy compared to simple keyword or rule-based mechanical approaches.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20250299095A1Generative adversarial network model training using distributed ledger

Publication Date: 2025.09.25 DOCUSIGN INT EMEA LTD

AI summary

Embodiments are directed to the tracking of data in a generative adversarial network (GAN) model using a distributed ledger system, such as a blockchain. A learning platform implementing a classification model receives, from a third party, a set of data examples generated by a generator model. The set of data examples are processed by the classification model, which outputs a prediction for each data example indicating whether each data example is true or false. The distributed ledger keeps a record of data examples submitted to the learning platform, as well as of predictions determined by the classification model on the learning platform. The learning platform analyzes the records of the distributed ledger, and pairs the records corresponding to the submitted data examples and the generated predictions determined by the classification model, and determines if the predictions were correct. The classification model may then be updated based upon the prediction results.