Immutable Ledger Storage for Machine Learning Model Replication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Replicating computer models, such as trained machine-learning algorithms, is challenging due to the lack of access to specific details like training data, code, and computing environments, which are often not accessible or not in the correct version.
Innovation Solution
The use of an immutable and decentralized ledger system, like a blockchain ledger, combined with a distributed database, to store and provide access to data associated with computer models, ensuring replicability and reproducibility by maintaining a transparent record of the model's assets and configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional academic publications are used to describe machine-learning algorithms, then high-level information on design and configuration is provided, but specific details on setup, configuration, and source data are omitted making replication difficult
Solution Approach 1:
The patent creates digital copies of all model assets (training data, code, configuration files, computing environment descriptions) and stores them in an immutable ledger system. This copying approach ensures that complete and accurate information is preserved without requiring complex management systems, as the ledger permanently stores all necessary details for replication.
Solution Approach 2:
The patent introduces an immutable ledger system as an intermediary between model creators and replicators. This intermediary permanently stores all model assets and metadata, providing a trusted source that eliminates information loss while maintaining a relatively simple system structure through its decentralized and transparent nature.
2Reliability
If assets like training data and code are made accessible, then replication becomes possible, but there is no guarantee that the correct version will be available
Solution Approach 1:
The patent performs preliminary action by immediately storing all model assets in the immutable ledger at the time of model creation. This ensures that the exact versions of training data, code, and configuration are permanently preserved before any replication attempt, eliminating versioning problems and guaranteeing that the correct assets are always available.
Solution Approach 2:
The patent replaces traditional mechanical file storage and version control systems with an immutable distributed ledger system. This substitution provides inherent guarantees of correctness and availability through cryptographic hashing and decentralized storage, eliminating the need for complex version management while ensuring reliable access to the correct asset versions.
3Productivity
If extensive experimentation is performed to replicate a model without exact assets, then some reproduction may be achieved, but there is no guarantee of success and significant time is consumed
Solution Approach 1:
The patent enables direct copying of all model assets including training data, code, configuration files, and computing environment descriptions in the immutable ledger. This allows replication to proceed by simply retrieving and executing the copied assets rather than performing extensive experimentation, dramatically increasing productivity and reducing time loss.
4Reliability
If a decentralized ledger system is used to store model data, then transparent and tamper-proof record is achieved, but system complexity increases
Solution Approach 1:
The patent replaces complex centralized storage and verification systems with a decentralized immutable ledger. This substitution achieves tamper-proof records through cryptographic hashing and distributed consensus mechanisms, providing high reliability while actually simplifying the system by eliminating single points of failure and centralized control.
Data Source
AI summary
The present disclosure relates generally to storing computer models, and more specifically to a platform for achieving replicability of a computer model (e.g., a trained machine-learning algorithm) by storing and providing access to data associated with the computer model using an immutable and decentralized ledger system (e.g., a blockchain ledger) and a distributed database. An exemplary computer-enabled method for storing a computer model, the method comprises: receiving data associated with the computer model; generating one or more asset files based on the data associated with the computer model; generating one or more hash values corresponding to the one or more asset files; generating one or more of location trackers corresponding to the one or more asset files; generating a ledger entry comprising the one or more hash values and the one or more location trackers; and adding the ledger entry to a blockchain ledger.