Self-learning selection of information-analysis runtimes

Patent metadata

Field	Value
Publication number	US-11288601-B2
Application number	US-201916360118-A
Country	US
Kind code	B2
Filing date	Mar 21, 2019
Priority date	Mar 21, 2019
Publication date	Mar 29, 2022
Grant date	Mar 29, 2022

Abstract

A self-learning computer-based system has access to multiple runtime modules that are each capable of performing a particular algorithm. Each runtime module implements the algorithm with different code or runs in a different runtime environment. The system responds to a request to run the algorithm by selecting the runtime module or runtime environment that the system predicts will provide the most desirable results based on parameters like accuracy, performance, cost, resource-efficiency, or policy compliance. The system learns how to make such predictions through training sessions conducted by a machine-learning component. This training teaches the system that previous module selections produced certain types of results in the presence of certain conditions. After determining whether similar conditions currently exist, the system uses rules inferred from the training sessions to select the runtime module most likely to produce desired results.

First claim

What is claimed is:

1. A self-learning information-analysis system comprising a processor, a memory coupled to the processor, and a computer-readable hardware storage device coupled to the processor, the storage device containing program code configured to be run by the processor via the memory to implement a method for self-learning selection of information-analysis runtimes, the method comprising:
the system receiving a request to perform an algorithm, where the system has access to a plurality of runtime modules that are each designed to perform the algorithm;
the system identifying a set of influencing factors capable of influencing the system's selection of an optimal runtime module, from the plurality of runtime modules, to perform the requested algorithm;
the system predicting outcomes that would be produced by running, in response to the request, each module of the plurality of runtime modules,
where the system has been trained to predict the outcomes by a machine-learning component that infers future outcomes from a record of previous outcomes produced by running any of the plurality of runtime modules in a presence of at least one factor of the set of influencing factors;

the system selecting an optimal runtime module of the plurality of runtime modules as a function of the predicting; and
the system initiating steps that direct the optimal runtime module to perform the algorithm.

2. The system of claim 1 ,
where at least two of the plurality of runtime modules are configured to run in different runtime environments, and
where the optimal runtime module is selected as a function of a runtime environment in which each module of the plurality of runtime modules is configured to run.

3. The system of claim 1 ,
where each factor of the set of influencing factors identifies a value selected from the group consisting of:
value of extrinsic factors that identify current conditions that are external to the system and that are external to the platform on which the system is implemented,
values of operating conditions that identify current characteristics of one or more components of the platform on which the system is implemented, and
dataset characteristics that characterize a dataset upon which the algorithm is to be performed.

4. The system of claim 1 , further comprising:
the system receiving feedback about the system's selection of the optimal runtime module,
where the feedback is used, by the machine-learning component, to further train the system to more accurately predict future outcomes produced by running the optimal runtime module.

5. The system of claim 1 , further comprising:
the system requesting and receiving authorization for the system to initiate steps that direct the optimal runtime module to perform the algorithm,
where the authorization is used, by the machine-learning component, to further train the system to more accurately predict future outcomes produced by running the optimal runtime module.

6. The system of claim 1 ,
where the selecting is performed as a further function of selection criteria selected from the group consisting of: resource efficiency, cost, accuracy, performance, security, privacy, data-pipeline priority, policy compliance, and conformance to conventions and standards.

7. The system of claim 6 ,
where at least one criterion of the selection criteria is assigned a weighting that alters a degree of influence of the one criterion upon the selection, relative to degrees of influence upon the selection of other criteria of the selection criteria.

8. A method for self-learning selection of information-analysis runtimes, the method comprising:
a self-learning information-analysis system receiving a request to perform an algorithm, where the system has access to a plurality of runtime modules that are each designed to perform the algorithm;
the system identifying a set of influencing factors capable of influencing the system's selection of an optimal runtime module, from the plurality of runtime modules, to perform the requested algorithm;
the system predicting outcomes that would be produced by running, in response to the request, each module of the plurality of runtime modules,
where the system has been trained to predict the outcomes by a machine-learning component that infers future outcomes from a record of previous outcomes produced by running any of the plurality of runtime modules in a presence of at least one factor of the set of influencing factors;

the system selecting an optimal runtime module of the plurality of runtime modules as a function of the predicting; and
the system initiating steps that direct the optimal runtime module to perform the algorithm.

9. The method of claim 8 ,
where at least two of the plurality of runtime modules are configured to run in different runtime environments, and
where the optimal runtime module is selected as a function of a runtime environment in which each module of the plurality of runtime modules is configured to run.

10. The method of claim 8 ,
where each factor of the set of influencing factors identifies a value selected from the group consisting of:
value of extrinsic factors that identify current conditions that are external to the system and that are external to the platform on which the system is implemented,
values of operating conditions that identify current characteristics of one or more components of the platform on which the system is implemented, and
dataset characteristics that characterize a dataset upon which the algorithm is to be performed.

11. The method of claim 8 , further comprising:
the system receiving feedback about the system's selection of the optimal runtime module,
where the feedback is used, by the machine-learning component, to further train the system to more accurately predict future outcomes produced by running the optimal runtime module.

12. The method of claim 8 , further comprising:
the system requesting and receiving authorization for the system to initiate steps that direct the optimal runtime module to perform the algorithm,
where the authorization is used, by the machine-learning component, to further train the system to more accurately predict future outcomes produced by running the optimal runtime module.

13. The method of claim 8 ,
where the selecting is performed as a further function of selection criteria selected from the group consisting of: resource efficiency, cost, accuracy, performance, security, privacy, data-pipeline priority, policy compliance, and conformance to conventions and standards, and
where at least one criterion of the selection criteria is assigned a weighting that alters a degree of influence of the one criterion upon the selection, relative to degrees of influence upon the selection of other criteria of the selection criteria.

14. The method of claim 8 , further comprising providing at least one support service for at least one of creating, integrating, hosting, maintaining, and deploying computer-readable program code in the computer system, wherein the computer-readable program code in combination with the computer system is configured to implement the receiving, the identifying, the predicting, the selecting, and the initiating.

15. A computer program product, comprising a computer-readable hardware storage device having a computer-readable program code stored therein, the program code configured to be executed by An information-analysis system comprising a processor, a memory coupled to the processor, and a computer-readable hardware storage device coupled to the processor, the storage device contai

Assignees

• IBM

Inventors

• Gupta Ritesh Kumar
• Kabra Namit
• Jacobson Eric Allen
• Brokaw Scott Louis
• Ramos Jo Arao

Classifications

• G06N5/025 Primary

  Extracting rules from data · CPC title

• G06N20/20 Primary

  Ensemble learning · CPC title

• G06F18/2155

  characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06N5/04

  Inference or reasoning models · CPC title