Feedback loop learning between artificial intelligence systems

Patent metadata

Field	Value
Publication number	US-11720826-B2
Application number	US-201916521185-A
Country	US
Kind code	B2
Filing date	Jul 24, 2019
Priority date	Jul 24, 2019
Publication date	Aug 8, 2023
Grant date	Aug 8, 2023

Abstract

Techniques that facilitate feedback loop learning between artificial intelligence systems are provided. In one example, a system includes a monitoring component and a machine learning component. The monitoring component identifies a data pattern associated with data for an artificial intelligence system. The machine learning component compares the data pattern to historical data patterns for the artificial intelligence system to facilitate modification of at least a component of the artificial intelligence system and/or one or more dependent systems of the artificial intelligence system.

First claim

What is claimed is:

1. A system, comprising:
a memory that stores computer executable components; and
a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a data analytics component that, based on an analysis of a microservices mesh, constructs a dependency mapping graph for artificial intelligence components of the microservices mesh, wherein the nodes of dependency mapping graph represent the artificial intelligence components, and edges of the dependency mapping graph represent respective dependencies between the artificial intelligence components;
a monitoring component that identifies, in real time, a data pattern associated with data generated during execution of the artificial intelligence components of the microservices mesh in a runtime environment of an artificial intelligence system;
a machine learning component that employs machine learning to:
determine, in real time, a deviation of the data pattern from one or more historical data patterns associated with one or more prior executions of the artificial intelligence components in the runtime environment of the artificial intelligence system, wherein the deviation indicates a need for a modification of at least one corresponding artificial intelligence component, in a development environment of the artificial intelligence system, that corresponds to at least one artificial intelligence component of the artificial intelligence components, and wherein the deviation comprises a new class of data representative of a new use case for the artificial intelligence components, and
determine, based on the dependency mapping graph, a need for another modification of at least one other artificial intelligence component that depends on the at least one corresponding artificial intelligence component based on the modification of at least one corresponding artificial intelligence component; and

a development component that implements the modification of the at least one corresponding artificial intelligence component and the other modification of the at least one other artificial intelligence component in the development environment of the artificial intelligence system.

2. The system of claim 1 , wherein the monitoring component identifies the data pattern associated with the data based on the dependency mapping graph for the artificial intelligence components.

3. The system of claim 1 , wherein the monitoring component monitors one or more application programming interface communications in the data associated with the artificial intelligence components.

4. The system of claim 1 , wherein the monitoring component monitors one or more events in the data associated with the artificial intelligence components.

5. The system of claim 1 , wherein the monitoring component monitors accuracy of output data in the data generated by the at least one artificial intelligence component of the artificial intelligence components.

6. The system of claim 1 , wherein the machine learning component infers a new classification for the at least one corresponding artificial intelligence component.

7. The system of claim 1 , wherein the development component that trains the at least one corresponding artificial intelligence component based on the data pattern.

8. The system of claim 1 , wherein the development component updates respective artificial intelligence models of the at least one corresponding artificial intelligence component based on the data pattern.

9. The system of claim 1 , wherein the machine learning component compares the data pattern to the one or more historical data patterns to improve performance of the artificial intelligence system.

10. A computer-implemented method, comprising:
based on an analysis of a microservices mesh, constructing, by a system operatively coupled to a processor, a dependency mapping graph for artificial intelligence components of the microservices mesh, wherein the nodes of dependency mapping graph represent the artificial intelligence components, and edges of the dependency mapping graph represent respective dependencies between the artificial intelligence components;
identifying, by the system, in real time, a data pattern associated with data generated during execution of artificial intelligence components of the microservices mesh in a runtime environment of a the artificial intelligence system; and
determining, by the system, using machine learning, in real time, a deviation of the data pattern from one or more historical data patterns associated with one or more prior executions of the artificial intelligence components in the runtime environment of the artificial intelligence system, wherein the deviation indicates a need for a modification of at least one corresponding artificial intelligence component, in a development environment of the artificial intelligence system, that corresponds to at least one artificial intelligence component of the artificial intelligence components, and wherein the deviation comprises a new class of data representative of a new use case for the artificial intelligence components;
determining, by the system, based on the dependency mapping graph, a need for another modification of at least one other artificial intelligence component that depends on the at least one corresponding artificial intelligence component based on the modification of at least one corresponding artificial intelligence component; and
implementing, by the system, the modification of the at least one corresponding artificial intelligence component and the other modification of the at least one other artificial intelligence component in the development environment of the artificial intelligence system.

11. The computer-implemented method of claim 10 , further comprising:
inferring, by the system, a new classification for the at least one corresponding artificial intelligence component based on the comparing.

12. The computer-implemented method of claim 10 , wherein the modification comprises:
modifying, by the system, one or more weights of the at least one corresponding artificial intelligence component.

13. The computer-implemented method of claim 10 , wherein the modification comprises:
modifying, by the system, one or more respective models associated with the at least one corresponding artificial intelligence component.

14. The computer-implemented method of claim 10 , wherein the modification comprises:
modifying, by the system, respective training data associated with the at least one corresponding artificial intelligence component.

15. The computer-implemented method of claim 10 , wherein the identifying the data pattern is based on the dependency mapping graph for the artificial intelligence components.

16. A computer program product facilitating feedback loop learning for an artificial intelligence system, 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:
based on an analysis of a microservices mesh, constructing, by the processor, a dependency mapping graph for artificial intelligence components of the microservices mesh, wherein the nodes of dependency mapping graph represent the artificial intelligence components, and edges of the dependency mapping graph represent respective dependencies between the artificial intelligence components;
identify, by the processor, in real time, a data pattern associated with data generated during execution of artificial intelligence

Assignees

• IBM

Inventors

• Hwang Jinho
• Shwartz Larisa
• Völzer Hagen
• Nidd Michael Elton
• Castrillon Rodrigo Otavio

Classifications

• G06N20/20 Primary

  Ensemble learning · CPC title

• G06N3/082 Primary

  modifying the architecture, e.g. adding, deleting or silencing nodes or connections · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/091

  Active learning · CPC title

• G06F9/541

  via adapters, e.g. between incompatible applications · CPC title