Blockchain framework for registering 3D machine-learning models with disclosed training datasets

We propose a framework for managing machine-learning (ML) models for three dimensional image data in a blockchain network. ML models are registered to the blockchain with their associated datasets in a relevant state. The key enabler is an aggregation of the hash values calculated by ML models and their associated datasets. This allows third parties to prove the authenticity of these data for them, even if only part of the data is disclosed to third parties as raw data.