Model-tiering machine learning model

1 . A method, comprising:

receiving, by a processor set, an input machine learning model being developed for deployment in a production environment;

creating, by the processor set, a feature vector based on metadata of the input machine learning model and a data source used by the input machine learning model;

determining, by the processor set, a risk tier of the input machine learning model by classifying the input machine learning model into one of plural predefined tiers using the feature vector with a model-tiering machine learning model;

providing, by the processor set, validation information to an evaluator, wherein the validation information is based on the risk tier and is used to validate the input machine learning model in accordance with the risk tier, and wherein providing the validation information comprises granting time-limited permissions to access data associated with the input machine learning model;

receiving, by the processor set, a validation annotation associated with the input machine learning model, the validation annotation indicating a successful validation or an unsuccessful validation; and

selectively deploying, by the processor set, the input machine learning model in the production environment, the selectively deploying being performed based on the validation annotation.

2 . The method of claim 1 , wherein the creating the feature vector comprises:

determining one or more techniques used by the input machine learning model by analyzing the metadata; and

including, in the feature vector, content based on the one or more techniques.

3 . The method of claim 1 , wherein the creating the feature vector comprises:

determining one or more areas of use of the input machine learning model by analyzing the metadata and the data source; and

including, in the feature vector, content based on the one or more areas of use.

4 . The method of claim 1 , wherein the creating the feature vector comprises:

identifying information in a knowledge base that is related to using the input machine learning model; and

including, in the feature vector, content based on the identified information in the knowledge base.

5 . The method of claim 1 , wherein the creating the feature vector comprises:

identifying information in a regulation that is related to using the input machine learning model; and

including, in the feature vector, content based on the identified information in the regulation.

6 . The method of claim 1 , wherein the creating the feature vector comprises:

identifying a type of sensitive information in the data source and used by the input machine learning model; and

including, in the feature vector, content based on the identified type of sensitive information.

7 . The method of claim 1 , further comprising training the model-tiering machine learning model using labeled training data.

8 . The method of claim 7 , further comprising creating the labeled training data, wherein the labeled training data comprises plural data pairs each comprising a feature vector of a historic machine learning model and a label corresponding to one of the plural predefined tiers.

9 . The method of claim 1 , wherein the model-tiering machine learning model comprises one an artificial neural network model and an extreme gradient boost model.

10 . The method of claim 1 , further comprising one of:

providing an indication of successful validation in response to the successful validation by the evaluator; and

providing an indication of unsuccessful validation in response to the unsuccessful validation by the evaluator.

11 . A computer program product comprising one or more computer readable storage media having program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:

receive an input machine learning model being developed for deployment in a production environment;

create a feature vector based on metadata of the input machine learning model and a data source used by the input machine learning model;

determine a risk tier of the input machine learning model by classifying the input machine learning model into one of plural predefined tiers using the feature vector with a model-tiering machine learning model;

provide validation information to an evaluator, wherein the validation information is based on the risk tier and is used to validate the input machine learning model in accordance with the risk tier, and wherein the validation information comprises time-limited permissions to access data associated with the input machine learning model;

receive a validation annotation associated with the input machine learning model, the validation annotation indicating a successful validation or an unsuccessful validation; and

selectively deploy the input machine learning model in the production environment based on the validation annotation.

12 . The computer program product of claim 11 , wherein the creating the feature vector comprises one or more operations selected from a group consisting of:

determining one or more techniques used by the input machine learning model by analyzing the metadata, and including, in the feature vector, content based on the one or more techniques;

determining one or more areas of use of the input machine learning model by analyzing the metadata and the data source, and including, in the feature vector, content based on the one or more areas of use;

identifying information in a knowledge base that is related to using the input machine learning model, and including, in the feature vector, content based on the identified information in the knowledge base;

identifying information in a regulation that is related to using the input machine learning model, and including, in the feature vector, content based on the identified information in the regulation; and

identifying a type of sensitive information in the data source and used by the input machine learning model, and including, in the feature vector, content based on the identified type of sensitive information.

13 . The computer program product of claim 11 , wherein the program instructions are executable to train the model-tiering machine learning model using labeled training data, wherein the labeled training data comprises plural data pairs each comprising a feature vector of a historic machine learning model and a label corresponding to one of the plural predefined tiers.

14 . The computer program product of claim 11 , wherein the model-tiering machine learning model comprises one an artificial neural network model and an extreme gradient boost model.

15 . The computer program product of claim 11 , wherein the program instructions are executable to:

provide an indication of successful validation in response to the successful validation by the evaluator; and

provide an indication of unsuccessful validation in response to the unsuccessful validation by the evaluator.

16 . A system comprising:

a processor set, one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:

receive an input machine learning model being developed for deployment in a production environment;

create a feature vector based on metadata of the input machine learning model and a data source used by the input machine learning model;

determine a risk tier of the input machine learning model by classifying the input machine learning model into one of plural predefined tiers using the feature vector with a model-tiering machine learning model;

provide validation information to an evaluator, wherein the validation information is based on the risk tier and is used to validate the input machine learning model in accordance with the risk tier, and wherein the validation information comprises time-limited permissions to access data associated with the input machine learning model;

receive a validation annotation associated with the input machine learning model, the validation annotation indicating a successful validation or an unsuccessful validation; and

selectively deploy the input machine learning model in the production environment based on the validation annotation.

17 . The system of claim 16 , wherein the creating the feature vector comprises one or more operations selected from a group consisting of:

determining one or more techniques used by the input machine learning model by analyzing the metadata, and including, in the feature vector, content based on the one or more techniques;

determining one or more areas of use of the input machine learning model by analyzing the metadata and the data source, and including, in the feature vector, content based on the one or more areas of use;

identifying information in a knowledge base that is related to using the input machine learning model, and including, in the feature vector, content based on the identified information in the knowledge base;

identifying information in a regulation that is related to using the input machine learning model, and including, in the feature vector, content based on the identified information in the regulation; and

identifying a type of sensitive information in the data source and used by the input machine learning model, and including, in the feature vector, content based on the identified type of sensitive information.

18 . The system of claim 16 , wherein the program instructions are executable to train the model-tiering machine learning model using labeled training data, wherein the labeled training data comprises plural data pairs each comprising a feature vector of a historic machine learning model and a label corresponding to one of the plural predefined tiers.

19 . The system of claim 16 , wherein the model-tiering machine learning model comprises one an artificial neural network model and an extreme gradient boost model.

20 . The system of claim 16 , wherein the program instructions are executable to:

provide an indication of successful validation in response to the successful validation by the evaluator; and

provide an indication of unsuccessful validation in response to the unsuccessful validation by the evaluator.