Blockchain Validation Records for Auditable AI Training Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing training datasets for AI models are vulnerable to alteration or contamination, which can lead to inaccurate predictions and potential harm, and there is a lack of transparent and auditable methods to ensure data integrity and lineage.

Innovation Solution

A decentralized peer-to-peer (P2P) computer network using blockchain technology to validate and record data based on pre-defined criteria, including digital image processing and adversarial vulnerability testing, generating immutable data records that include metadata and hash values, and achieving consensus among nodes for data integrity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional centralized data management systems are used for training datasets, then data processing and model training can be performed efficiently, but the system becomes vulnerable to data alteration, contamination, and lacks transparent auditing capabilities

Engineering Contradiction:

Improvedata integrityVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system segments data management into distributed ledger entries, where each data item is recorded as a separate block with its own hash, allowing individual verification without affecting the entire system. This segmentation enables reliable auditing while maintaining distributed processing capabilities.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The blockchain acts as an intermediary layer between data sources and AI models, providing a transparent and immutable record of data lineage. This intermediary enables auditing and integrity verification without requiring complex centralized management systems.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If comprehensive validation criteria including digital image processing and adversarial vulnerability testing are applied to all training data, then data safety and authenticity improve, but processing time and computational resources increase

Engineering Contradiction:

Improvedata safetyVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs validation actions in advance by recording validation results in the blockchain before data is used for training. This preliminary action ensures data safety is established beforehand, allowing subsequent processing to proceed without repeated validation overhead.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates a copy of validation results and stores them in the blockchain ledger, separating the validation process from the actual training process. This copying allows the same validation criteria to be applied efficiently without repeating computational intensive operations during model training.

Inventive Principle:

Principle #26Copying

3Loss of information

If immutable blockchain records are maintained for all training data, then auditability and transparency improve, but storage requirements and system complexity increase

Engineering Contradiction:

ImproveauditabilityVSAvoiddata storage

Core Design Contradiction:

Loss of informationVSQuantity of substance

Solution Approach 1:

The system extracts only the essential validation information and metadata needed for auditing into the blockchain, rather than storing complete data copies. This extraction approach maintains auditability while minimizing storage requirements by recording only hash values, timestamps, and validation results.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system uses lightweight hash values and condensed metadata in the blockchain rather than storing full data records. This approach creates inexpensive, disposable-like records that provide sufficient audit information without requiring substantial storage capacity.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

4Reliability

If multiple validation nodes perform independent validation of the same data, then consensus and reliability improve, but computational workload and processing time increase

Engineering Contradiction:

Improveconsensus accuracyVSAvoidcomputational workload

Core Design Contradiction:

ReliabilityVSPower

Solution Approach 1:

The system merges validation results from multiple nodes into a single blockchain record through consensus mechanisms. This combining approach maintains reliability by aggregating multiple validation perspectives while reducing overall computational workload by avoiding redundant independent validation of the same data.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS20250358310A1System and method of managing and auditing training data based on distributed ledger technology

Publication Date: 2025.11.20 PLAINSIGHT TECH INC

AI summary

A computing node in a P2P computer network obtaining a media item to be verified, determining a hash value for the media item based on a digital cryptographic hash function, retrieving a plurality of data records associated with the media item from a blockchain based at least in part on the hash value, wherein a data record is generated by a validation node in the decentralized P2P computer network, evaluating the plurality of data records to determine whether the media item has satisfied pre-defined validation criteria, and providing information describing the media item based at least in part on the plurality of data records, wherein the information provides an indication as to whether the media item satisfied the pre-defined validation criteria.