Distributed ledger based machine-learning model management
Disclosed are various approaches for managing the status of machine-learning models using distributed ledgers. A registration request for a machine-learning model can be received. The registration request can include a model name for the machine-learning model, a version identifier for the machine-learning model, a network address from which the machine-learning model can be retrieved, a source code hash for a source code version of the machine learning model, and a runtime hash for a binary executable version of the machine-learning model. A registration identifier can then be created based at least in part on the source code hash and the runtime hash. Subsequently, an entry in the distributed ledger can be created for the machine-learning model. The entry can include the registration identifier, the model name, the model version, the network address, the source code hash, and the runtime hash.
1. A system, comprising:
a computing device comprising a processor and a memory; and
machine-readable instructions stored in a distributed ledger that, when executed by the processor, cause the computing device to at least:
retrieve a runtime machine-learning model from an application record in a repository data store, the application record corresponding to an application that uses a machine-learning model;
calculate a runtime hash for the runtime machine-learning model;
concatenate the runtime hash with a source code hash to form a registration identifier;
provide the runtime hash and the registration identifier to a distributed agent as search criteria for the distributed agent to search the distributed ledger for a machine-learning model record;
evaluate a compliance status of the application record with a compliance rule based at least in part on a result of the search; and
send a request to the distributed agent to update the machine-learning model record in the distributed ledger based at least in part on whether the runtime machine-learning model at least meets a performance metric associated with the compliance status of the application record.
2. The system of claim 1 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
obtain a machine-learning model record stored in the distributed ledger in response to the search; and
confirm that the application record satisfies the compliance rule in response to obtaining the machine-learning model record.
3. The system of claim 1 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
determine that the distributed ledger fails to contain a machine-learning model record with a matching runtime hash; and
determine that the application record fails to satisfy the compliance rule in response to obtaining the machine-learning model record.
4. The system of claim 1 , wherein the compliance rule specifies a minimum level of accuracy for the machine-learning model.
5. The system of claim 1 , wherein the compliance rule specifies a maximum level of bias or variance for the machine-learning model.
6. The system of claim 1 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
receive a registration request for a machine-learning model, the registration request comprising a model name for the machine-learning model, a version identifier for the machine-learning model, a network address from which the machine-learning model can be retrieved, the source code hash for a source code version of the machine learning model, and a runtime hash for a binary executable version of the machine-learning model;
create the registration identifier based at least in part on the source code hash and the runtime hash; and
create an immutable entry in the distributed ledger for the machine-learning model, the immutable entry comprising the registration identifier, the model name, the version identifier, the network address, the source code hash, and the runtime hash.
7. The system of claim 6 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
receive a rating for the machine-learning model and the registration identifier or runtime hash of the machine-learning model;
select the immutable entry in the distributed ledger for the machine-learning model based on the registration identifier or runtime hash; and
update a score for the selected immutable entry in the distributed ledger for the machine-learning model based at least in part on the rating.
8. A computer-implemented method, comprising:
retrieving a runtime machine-learning model from an application record in a repository data store, the application record corresponding to an application that uses a machine-learning model;
calculating a runtime hash for the runtime machine-learning model;
concatenating the runtime hash with a source code hash to form a registration identifier;
providing the runtime hash and the registration identifier to a distributed agent as search criteria for the distributed agent to search a distributed ledger for a machine-learning model record;
evaluating a compliance status of the application record with a compliance rule based at least in part on a result of the search; and
sending a request to the distributed agent to update the machine-learning model record in the distributed ledger based at least in part on whether the runtime machine-learning model at least meets a performance metric associated with the compliance status of the application record.
9. The computer-implemented method of claim 8 , further comprising:
obtaining a machine-learning model record stored in the distributed ledger in response to the search; and
confirming that the application record satisfies the compliance rule in response to obtaining the machine-learning model record.
10. The computer-implemented method of claim 9 , further comprising:
determining that the distributed ledger fails to contain a machine-learning model record with a matching runtime hash; and
determining that the application record fails to satisfy the compliance rule in response to obtaining the machine-learning model record.
11. The computer-implemented method of claim 8 , wherein the compliance rule specifies a minimum level of accuracy for the machine-learning model.
12. The computer-implemented method of claim 8 , wherein the compliance rule specifies a maximum level of bias or variance for the machine-learning model.
13. The computer-implemented method of claim 8 , further comprising:
receiving a registration request for a machine-learning model, the registration request comprising a model name for the machine-learning model, a version identifier for the machine-learning model, a network address from which the machine-learning model can be retrieved, the source code hash for a source code version of the machine learning model, and a runtime hash for a binary executable version of the machine-learning model;
creating the registration identifier based at least in part on the source code hash and the runtime hash; and
creating an immutable entry in the distributed ledger for the machine-learning model, the immutable entry comprising the registration identifier, the model name, the version identifier, the network address, the source code hash, and the runtime hash.
14. The computer-implemented method of claim 13 , further comprising:
receiving a rating for the machine-learning model and the registration identifier or runtime hash of the machine-learning model;
selecting the immutable entry in the distributed ledger for the machine-learning model based on the registration identifier or runtime hash; and
updating a score for the selected immutable entry in the distributed ledger for the machine-learning model based at least in part on the rating.
15. A non-transitory, computer-readable medium, comprising machine-readable instructions that, when executed by a processor of a computing device, cause the computing device to at least:
retrieve a runtime machine-learning model from an application record in a repository data store, the application record corresponding to an application that uses a machine-learning model;
calculate a runtime hash for the runtime machine-learning model;
concatenate the runtime hash with a source code hash to form a registration identifier;
provide the runtime hash and the registration identifier to a distributed agent as search criteria for the distributed agent to search a distributed ledger for a machine-learning model record;
evaluate a compliance status of the application record with a compliance rule based at least in part on a result of the search; and
send a request to the distributed agent to update the machine-learning model record in the distributed ledger based at least in part on whether the runtime machine-learning model at least meets a performance metric associated with the compliance status of the application record.
16. The non-transitory, computer-readable medium of claim 15 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
receive a registration request for a machine-learning model, the registration request comprising a model name for the machine-learning model, a version identifier for the machine-learning model, a network address from which the machine-learning model can be retrieved, the source code hash for a source code version of the machine learning model, and a runtime hash for a binary executable version of the machine-learning model;
create the registration identifier based at least in part on the source code hash and the runtime hash; and
create an immutable entry in the distributed ledger for the machine-learning model, the immutable entry comprising the registration identifier, the model name, the version identifier, the network address, the source code hash, and the runtime hash.
17. The non-transitory, computer-readable medium of claim 15 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
obtain a machine-learning model record stored in the distributed ledger in response to the search; and
confirm that the application record satisfies the compliance rule in response to obtaining the machine-learning model record.
18. The non-transitory, computer-readable medium of claim 15 , wherein the machine-readable instructions, when executed by the processor, further cause the computing device to at least:
determine that the distributed ledger fails to contain a machine-learning model record with a matching runtime hash; and
determine that the application record fails to satisfy the compliance rule in response to obtaining the machine-learning model record.
19. The non-transitory, computer-readable medium of claim 15 , wherein the compliance rule specifies a minimum level of accuracy for the machine-learning model.
20. The non-transitory, computer-readable medium of claim 15 , wherein the compliance rule specifies a maximum level of bias or variance for the machine-learning model.