Data set filtering for machine learning

Patent metadata

Field	Value
Publication number	US-11526849-B2
Application number	US-201916376675-A
Country	US
Kind code	B2
Filing date	Apr 5, 2019
Priority date	Apr 5, 2019
Publication date	Dec 13, 2022
Grant date	Dec 13, 2022

Abstract

A device may determine an association between a second set of parameters and a third set of parameters using a pseudoinversion network and a multiple regression procedure. The device may determine semantic embeddings based on a set of semantic descriptions of the second set of parameters. The device may determine a semantic similarity between parameters of the second set of parameters based on the semantic embeddings. The device may determine a consistency error based on the semantic similarity. The device may generate, using a regression-based learning model technique, a matrix representing an association between the second set of parameters and the third set of parameters based on the association and the consistency error. The device may perform an action based on the matrix.

First claim

What is claimed is:

1. A device, comprising:
one or more memories; and
one or more processors, communicatively coupled to the one or more memories, to:
receive a first matrix and a second matrix,
wherein the first matrix represents an association between a first set of parameters and a second set of parameters and the second matrix represents an association between the first set of parameters and a third set of parameters;

estimate an association between the second set of parameters and the third set of parameters using a pseudoinversion network and a multiple regression procedure;
receive a set of semantic descriptions of the second set of parameters;
generate a semantic embeddings matrix based on the set of semantic descriptions of the second set of parameters;
determine a semantic similarity between parameters of the second set of parameters based on the semantic embeddings matrix;
determine a consistency error based on the semantic similarity;
generate, using a regression-based learning model technique, a third matrix representing an association between the second set of parameters and the third set of parameters based on the association and the consistency error; and
perform an action based on the third matrix.

2. The device of claim 1 , wherein a semantic similarity between parameters of the third set of parameters corresponds to the semantic similarity between the parameters of the second set of parameters.

3. The device of claim 1 , wherein the one or more processors are further to:
divide the parameters of the second set of parameters into a training set and a prediction set based on the consistency error;
generate a regression-based learning model using the regression-based learning model technique and based on the training set and the prediction set; and
wherein the one or more processors, when generating the third matrix, are to:
generate the third matrix using the regression-based learning model.

4. The device of claim 3 , wherein the one or more processors are further to:
determine the association between parameters of the third set of parameters and the parameters of the second set of parameters, which are included in the prediction set, using the regression-based learning model.

5. The device of claim 1 , wherein the first set of parameters is a set of workforce roles, the second set of parameters is a set of workforce tasks, and the third set of parameters is a set of workforce attributes.

6. The device of claim 1 , wherein the one or more processors, when estimating the association between the second set of parameters and the third set of parameters, are configured to:
set the first matrix as an input layer;
set the second matrix as an output layer; and
set the third matrix as a set of internal parameters and an internal structure for estimation using the multiple regression procedure.

7. The device of claim 6 , wherein the one or more processors are further to:
perform pseudoinversion network prediction over a set of identity matrices representing parameters of the first matrix corresponding to single parameters of the second matrix.

8. The device of claim 1 , wherein the one or more processors are further configured to:
perform natural language processing on the set of semantic descriptions to generate the semantic embeddings matrix.

9. The device of claim 1 , wherein the action is a response action associated with a role assessment relating to tasks and attributes, and
wherein the response action is at least one of:
an automated training enrollment action,
an automated training scheduling action,
an automated role reassignment action,
an automated job application action,
an automated job posting action, or
an automated job searching and reporting action.

10. A method, comprising:
receiving, by a device, a first matrix and a second matrix,
wherein the first matrix represents an association between a first set of parameters and a second set of parameters and the second matrix represents an association between the first set of parameters and a third set of parameters;

determining, by the device, an association between the second set of parameters and the third set of parameters using a pseudoinversion network and a multiple regression procedure;
generating, by the device, a semantic embeddings matrix based on a set of semantic descriptions of the second set of parameters;
determining, by the device, a semantic similarity between parameters of the second set of parameters based on the semantic embeddings matrix;
determining, by the device, a consistency error based on the semantic similarity;
generating, by the device and using a regression-based learning model technique, a third matrix representing an association between the second set of parameters and the third set of parameters based on the association and the consistency error; and
performing, by the device, an action based on the third matrix.

11. The method of claim 10 , wherein the first set of parameters is a set of workforce roles, the second set of parameters is a set of workforce tasks, and the third set of parameters is a set of workforce attributes.

12. The method of claim 10 , wherein determining the association between the second set of parameters and the third set of parameters comprises:
setting the first matrix as an input layer;
setting the second matrix as an output layer; and
setting the third matrix as a set of internal parameters and an internal structure for estimation using the multiple regression procedure.

13. The method of claim 10 , further comprising:
performing pseudoinversion network prediction over a set of identity matrices representing parameters of the first matrix corresponding to single parameters of the second matrix.

14. The method of claim 10 , further comprising
performing natural language processing on the set of semantic descriptions to generate the semantic embeddings matrix.

15. The method of claim 10 , wherein the action is a response action associated with a role assessment relating to tasks and attributes, and
wherein the response action is at least one of:
an automated training enrollment action,
an automated training scheduling action,
an automated role reassignment action,
an automated job application action,
an automated job posting action, or
an automated job searching and reporting action.

16. A non-transitory computer-readable medium storing instructions, the instructions comprising:
one or more instructions that, when executed by one or more processors of a device, cause the one or more processors to:
determine an association between a second set of parameters and a third set of parameters using a pseudoinversion network and a multiple regression procedure,
wherein an association between a first set of parameters and the second set of parameters is defined for the device and an association between the first set of parameters and the third set of parameters is defined for the device;

determine semantic embeddings based on a set of semantic descriptions of the second set of parameters;
determine a semantic similarity between parameters of the second set of parameters based on the semantic embeddings;
determine a consistency error based on the semantic similarity;
generate, using a regression-based learning model technique, a matrix representing an association between the second set of parameters and the third set of parameters based on the association and the consistency error; and
perform an action based on the matrix.

17. The non-transitory computer-read

Assignees

• Accenture Global Solutions Ltd

Inventors

• Lecue Freddy
• Zayats Mykhaylo
• Golla Benedikt Maximilian Johannes

Classifications

• G06Q10/1053 Primary

  Employment or hiring · CPC title

• G06Q10/06311 Primary

  Scheduling, planning or task assignment for a person or group · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N5/022

  Knowledge engineering; Knowledge acquisition · CPC title