Configurable and scalable machine learning model lifecycle operator

Patent metadata

Field	Value
Publication number	US-12190210-B2
Application number	US-202117554166-A
Country	US
Kind code	B2
Filing date	Dec 17, 2021
Priority date	Dec 17, 2021
Publication date	Jan 7, 2025
Grant date	Jan 7, 2025

Abstract

A method of using a computing device to manage a lifecycle of machine learning models includes receiving, by a computing device, multiple pre-defined machine learning lifecycle tasks. The computing device manages executing a management-layer software layer for the multiple pre-defined machine learning lifecycle tasks. The computing device further generates and updates a machine learning pipeline using the management-layer software layer.

First claim

What is claimed is:

1. A method of using a computing device to manage a lifecycle of machine learning models, the method comprising:
receiving, by a computing device, a plurality of pre-defined machine learning lifecycle tasks that represent stages in the lifecycle of the machine learning models, wherein inputs for pre-defined machine learning lifecycle tasks are provided by outputs of preceding pre-defined machine learning lifecycle tasks from the plurality of pre-defined machine learning lifecycle tasks;
managing, by the computing device, execution of a management-layer software layer for the plurality of pre-defined machine learning lifecycle tasks;
generating and updating, by the computing device, a machine learning pipeline using the management-layer software layer;
adding, by the computing device, new lifecycle tasks to the plurality of pre-defined machine learning lifecycle tasks;
monitoring, by the computing device, external dependencies; and
launching, by the computing device, pipelines containing a subset of the machine learning lifecycle tasks, wherein the launched pipelines update machine learning model instances, and wherein the updates generate an end-to-end lifecycle of the machine learning model instances associated with a corresponding desired state.

2. The method of claim 1 , wherein each machine learning lifecycle task from the plurality of pre-defined machine learning lifecycle tasks comprises code that executes a specific stage in the lifecycle of a machine learning model.

3. The method of claim 1 , wherein a machine learning model lifecycle configuration specifies a set of new lifecycle tasks, that generates an end-to-end lifecycle for a particular machine learning model instance.

4. The method of claim 3 , wherein a lifecycle operator process orchestrates execution of the plurality of pre-defined machine learning tasks, pipelines and triggers that manage one or more machine learning model instances.

5. The method of claim 4 , wherein the lifecycle operator process manages a plurality of model versions for each managed machine learning model instance.

6. The method of claim 5 , wherein the machine learning model lifecycle configuration is constructed from a pre-build lifecycle template and composed of pre-build lifecycle tasks.

7. A computer program product for managing a lifecycle of machine learning models, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
receive, by the processor, a plurality of pre-defined machine learning lifecycle tasks that represent stages in the lifecycle of the machine learning models, wherein inputs for pre-defined machine learning lifecycle tasks are provided by outputs of preceding pre-defined machine learning lifecycle tasks from the plurality of pre-defined machine learning lifecycle tasks;
manage, by the processor, execution of a management-layer software layer for the plurality of pre-defined machine learning lifecycle tasks;
generate and update, by the processor, a machine learning pipeline using the management-layer software layer;
adding, by the processor, new lifecycle tasks to the plurality of pre-defined machine learning lifecycle tasks;
monitoring, by the processor, external dependencies; and
launching, by the processor, pipelines containing a subset of the machine learning lifecycle tasks, wherein the launched pipelines update machine learning model instances, and wherein the updates generate an end-to-end lifecycle of the machine learning model instances associated with a corresponding desired state.

8. The computer program product of claim 7 , wherein each machine learning lifecycle task from the plurality of pre-defined machine learning lifecycle tasks comprises code that executes a specific stage in the lifecycle of a machine learning model.

9. The computer program product of claim 7 , wherein a machine learning model lifecycle configuration specifies a set of new lifecycle tasks, that generates an end-to-end lifecycle for a particular machine learning model instance.

10. The computer program product of claim 9 , wherein a lifecycle operator process orchestrates execution of the plurality of pre-defined machine learning tasks, pipelines and triggers that manage one or more machine learning model instances.

11. The computer program product of claim 10 , wherein the lifecycle operator process manages a plurality of model versions for each managed machine learning model instance.

12. The computer program product of claim 11 , wherein the machine learning model lifecycle configuration is constructed from a pre-build lifecycle template and composed of pre-build lifecycle tasks.

13. An apparatus comprising:
a memory configured to store instructions; and
a processor configured to execute the instructions to:
receive a plurality of pre-defined machine learning lifecycle tasks that represent stages in lifecycle of machine learning models, wherein inputs for pre-defined machine learning lifecycle tasks are provided by outputs of preceding pre-defined machine learning lifecycle tasks from the plurality of pre-defined machine learning lifecycle tasks;
manage execution of a management-layer software layer for the plurality of pre-defined machine learning lifecycle tasks;
generate and update a machine learning pipeline using the management-layer software layer;
add new lifecycle tasks to the plurality of pre-defined machine learning lifecycle tasks;
monitor external dependencies; and
launch pipelines containing a subset of the machine learning lifecycle tasks, wherein the launched pipelines update machine learning model instances, and wherein the updates generate an end-to-end lifecycle of the machine learning model instances associated with a corresponding desired state.

14. The apparatus of claim 13 , wherein each machine learning lifecycle task from the plurality of pre-defined machine learning lifecycle tasks comprises code that executes a specific stage in the lifecycle of a machine learning model.

15. The apparatus of claim 13 , wherein a machine learning model lifecycle configuration specifies a set of new lifecycle tasks, that generates an end-to-end lifecycle for a particular machine learning model instance.

16. The apparatus of claim 15 , wherein a lifecycle operator process orchestrates execution of the plurality of pre-defined machine learning tasks, pipelines and triggers that manage one or more machine learning model instances.

17. The apparatus of claim 16 , wherein the lifecycle operator process manages a plurality of model versions for each managed machine learning model instance, and the machine learning model lifecycle configuration is constructed from a pre-build lifecycle template and composed of pre-build lifecycle tasks.

Assignees

• IBM

Inventors

• Herta Benjamin
• Reimer Darrell Christopher
• Duesterwald Evelyn
• Fang Gaodan
• Oum Punleuk
• Saha Debashish
• Verma Archit

Classifications

• G06F9/485 Primary

  Task life-cycle, e.g. stopping, restarting, resuming execution (G06F9/4881 takes precedence) · CPC title

• G06F9/4887

  involving deadlines, e.g. rate based, periodic · CPC title

• G06F9/5038

  considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration (scheduling strategies G06F9/4881 and subgroups) · CPC title

• G06N20/20

  Ensemble learning · CPC title

• G06N20/00 Primary

  Machine learning · CPC title