IP Library › Granted Patent US 12,073,298
Granted Patent B2
US 12,073,298 · App. 17/811,555 · Granted Aug 27, 2024

Machine learning service

Inventors: Leo Parker Dirac (Seattle, WA); Nicolle M. Correa (Seattle, WA); Aleksandr Mikhaylovich Ingerman (Seattle, WA); Sriram Krishnan (Sammamish, WA); Jin Li (Bellevue, WA); Sudhakar Rao Puvvadi (Bellevue, WA); Saman Zarandioon (Seattle, WA)
Assignee: Amazon Technologies, Inc.
G06N20/00
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,073,298
App. No.
17/811,555
Filed
Jul 8, 2022
Granted
Aug 27, 2024
Kind
B2
Art Unit
2127
USPC
706/12
Abstract

A machine learning service implements programmatic interfaces for a variety of operations on several entity types, such as data sources, statistics, feature processing recipes, models, and aliases. A first request to perform an operation on an instance of a particular entity type is received, and a first job corresponding to the requested operation is inserted in a job queue. Prior to the completion of the first job, a second request to perform another operation is received, where the second operation depends on a result of the operation represented by the first job. A second job, indicating a dependency on the first job, is stored in the job queue. The second job is initiated when the first job completes.

Claims (49)

1. A computer implemented method, comprising:

receiving, via a user interface of a machine learning service (MLS) of an infrastructure provider network, a data source creation request from a client specifying an address for a dataset;

responsive to the data source creation request, generating and storing a data source artifact using the address;

receiving, via the user interface, a hyper-parameter auto-tuning request from the client to optimize a set of hyper-parameters associated with a machine learning model managed by the MLS, wherein the hyper-parameter auto-tuning request specifies one or more optimization criteria for the set of hyper-parameters;

responsive to the hyper-parameter auto-tuning request, executing an automated parameter exploration process to automatically explore a parameter value space of the set of hyper-parameters and determine values of the set of hyper-parameters that satisfy the specified one or more optimization criteria;

receiving, via the user interface, a model execution request from the client to train the machine learning model, the model execution request specifying the data source artifact as input data and a type of output of the machine learning model; and

executing one or more jobs to train the machine learning model responsive to the model execution request, including:

instantiate the machine learning model on a set of resources that includes one or more virtual computing resources hosted on one or more physical hosts of the infrastructure provider network, wherein virtual computing resources are isolated to the client to execute only workloads of the client;

executing the machine learning model on the one or more virtual computing resources using training data retrieved using the data source artifact and according to the values of the set of hyper-parameters; and

modifying the machine learning model based on output generated by the machine learning model.

2. The computer implemented method of claim 1 , wherein the MLS supports machine learning models that process video data or image data.

3. The computer implemented method of claim 1 , wherein the MLS supports machine learning models that performs classification or regression tasks.

4. The computer implemented method of claim 1 , further comprising the MLS storing the data source artifact at a storage service of the infrastructure provider network.

5. The computer implemented method of claim 1 , further comprising the MLS determining performance metrics of the machine learning model based on the output generated by the machine learning model.

6. The computer implemented method of claim 1 , further comprising the MLS sending a notification to the client when the model execution request is completed.

7. The computer implemented method of claim 1 , further comprising the MLS assigning an endpoint to the machine learning model configured to receive additional input data for the machine learning model.

8. The computer implemented method of claim 1 , wherein the model execution request specifies a resource budget limit for executing the machine learning model.

9. The computer implemented method of claim 1 , wherein the model execution request specifies a maximum number of servers to use to execute the machine learning model.

10. The computer implemented method of claim 1 , wherein the model execution request specifies one or more types of resources to use to execute the machine learning model.

11. The computer implemented method of claim 1 , wherein the user interface comprises a web-based graphical user interface.

12. A system, comprising:

one or more computing systems that implement a machine learning service (MLS) of an infrastructure provider network, configure to:

receive, via a user interface of the MLS, a data source creation request from a client specifying an address for a dataset;

responsive to the data source creation request, generate and store a data source artifact using the address;

receive, via the user interface, a hyper-parameter auto-tuning request from the client to optimize a set of hyper-parameters associated with a machine learning model managed by the MLS, wherein the hyper-parameter auto-tuning request specifies one or more optimization criteria for the set of hyper-parameters;

responsive to the hyper-parameter auto-tuning request, executing an automated parameter exploration process to automatically explore a parameter value space of the set of hyper-parameters and determine values of the set of hyper-parameters that satisfy the specified one or more optimization criteria;

receive, via the user interface, a model execution request from the client to train the machine learning model, the model execution request specifying the data source artifact as input data and a type of output of the machine learning model; and

execute one or more jobs to train the machine learning model responsive to the model execution request, including to:

instantiate the machine learning model on a set of resources that includes one or more virtual computing resources hosted on one or more physical hosts of the infrastructure provider network, wherein virtual computing resources are isolated to the client to execute only workloads of the client;

execute the machine learning model on the one or more virtual computing resources using training data retrieved using the data source artifact and according to the values of the set of hyper-parameters; and

modify the machine learning model based on output generated by the machine learning model.

13. The system of claim 12 , wherein the MLS supports machine learning models that process video data or image data.

14. The system of claim 12 , wherein the MLS supports machine learning models that performs classification or regression tasks.

15. The system of claim 12 , wherein the MLS is configured to store the data source artifact at a storage service of the infrastructure provider network.

16. The system of claim 12 , wherein the MLS is configured to determine performance metrics of the machine learning model based on the output generated by the machine learning model.

17. The system of claim 12 , wherein the MLS is configured to send a notification to the client when the model execution request is completed.

18. The computer implemented method of claim 1 , wherein the user interface comprises a web-based graphical user interface.

19. One or more non-transitory computer-readable media having stored instructions that when executed on or across one or more processors implement at least a portion of a machine learning service (MLS) of an infrastructure provider network and cause the MLS to:

receive, via a user interface of the MLS, a data source creation request from a client specifying an address for a dataset;

responsive to the data source creation request, generate and store a data source artifact using the address;

receive, via the user interface, a hyper-parameter auto-tuning request from the client to optimize a set of hyper-parameters associated with a machine learning model managed by the MLS, wherein the hyper-parameter auto-tuning request specifies one or more optimization criteria for the set of hyper-parameters;

responsive to the hyper-parameter auto-tuning request, executing an automated parameter exploration process to automatically explore a parameter value space of the set of hyper-parameters and determine values of the set of hyper-parameters that satisfy the specified one or more optimization criteria;

receive, via the user interface, a model execution request from the client to train the machine learning model, the model execution request specifying the data source artifact as input data and a type of output of the machine learning model; and

execute one or more jobs to train the machine learning model responsive to the model execution request, including to:

instantiate the machine learning model on a set of resources that includes one or more virtual computing resources hosted on one or more physical hosts of the infrastructure provider network, wherein virtual computing resources are isolated to the client to execute only workloads of the client;

execute the machine learning model on the one or more virtual computing resources using training data retrieved using the data source artifact and according to the values of the set of hyper-parameters; and

modify the machine learning model based on output generated by the machine learning model.

20. The one or more non-transitory computer-readable media of claim 19 , wherein the stored instructions when executed on or across the one or more processors cause the MLS to:

output the value of the hyper-parameter as a recommended value via the user interface.

Continuity (3)
Continuation 16159441 · Oct 12, 2018
Continuation 14319902 · Jun 30, 2014
Related Publication 20220391763A1 · Dec 8, 2022