Back to results
US 12,748,653 B2Granted
Correction data determination apparatus, correction data determination method, and storage medium
AssigneeNEC Corporation
InventorsYuyang Dong, Masafumi Enomoto
Life of this patent
Estimating the term…
Abstract
To enable appropriate error correction that suits an analysis task, an information processing apparatus (1) includes: an acquisition unit (11) that acquires target data; a calculation unit (12) that calculates, for respective ones of a plurality of errors included in the target data or for respective ones of attributes of the plurality of errors, corresponding degrees of influence that the respective ones of the plurality of errors exert on an evaluation index of a machine learning model; and a determination unit (13) that determines data to be corrected in the target data on the basis of the degrees of influence calculated by the calculation unit (12).
Claims10
A correction data determination apparatus comprising:
- at least one processor, the at least one processor carrying out:
- an acquisition process for acquiring target data;
- a calculation process for calculating, for respective ones of a plurality of errors included in the target data or for respective ones of attributes of the plurality of errors, corresponding degrees of influence that the respective ones of the plurality of errors or the respective ones of the attribute exert on an evaluation index of a machine learning model;
- wherein the calculation process includes:
- generating, for each of the respective ones of the plurality of errors or the respective ones of the attributes, evaluation data obtained by including a pseudo error corresponding to the error or the attribute in the target data;
- and
- calculating the degree of influence corresponding to the error or the attribute on the basis of a comparison between performance of a machine learning model generated with use of the target data and performance of a machine learning model generated with use of the evaluation data;
- a determination process for determining data to be corrected in the target data on the basis of the degrees of influence calculated in the calculation process, wherein the determination process includes determining, from among pieces of data included in the target data, the data to be corrected in accordance with one or more errors or one or more attributes of one or more errors included in the pieces of data and the degrees of influence corresponding to the one or more errors or the one or more attributes;
- and
- a data cleansing process for correcting the data that has been determined to be corrected in the target data.
7 more elements
A correction data determination method comprising:
- at least one processor acquiring target data;
- the at least one processor calculating, for respective ones of a plurality of errors included in the target data or for respective ones of attributes of the plurality of errors, corresponding degrees of influence that the respective ones of the plurality of errors or the respective ones of the attribute exert on an evaluation index of a machine learning model, wherein calculating the corresponding degrees of influence includes:
- generating, for each of the respective ones of the plurality of errors or the respective ones of the attributes, evaluation data obtained by including a pseudo error corresponding to the error or the attribute in the target data;
- and
- calculating the degree of influence corresponding to the error or the attribute on the basis of a comparison between performance of a machine learning model generated with use of the target data and performance of a machine learning model generated with use of the evaluation data;
- the at least one processor determining data to be corrected in the target data on the basis of the calculated degrees of influence, wherein determining the data to be corrected in the target data includes determining, from among pieces of data included in the target data, the data to be corrected in accordance with one or more errors or one or more attributes of one or more errors included in the pieces of data and the degrees of influence corresponding to the one or more errors or the one or more attributes;
- and
- the at least one processor correcting the data that has been determined to be corrected in the target data.
5 more elements
A computer-readable non-transitory storage medium storing a program for causing a computer to function as a correction data determination apparatus, the program causing the computer to carry out:
- an acquisition process for acquiring target data;
- a calculation process for calculating, for respective ones of a plurality of errors included in the target data or for respective ones of attributes of the plurality of errors, corresponding degrees of influence that the respective ones of the plurality of errors or the respective ones of the attribute exert on an evaluation index of a machine learning model, wherein the calculation process includes:
- generating, for each of the respective ones of the plurality of errors or the respective ones of the attributes, evaluation data obtained by including a pseudo error corresponding to the error or the attribute in the target data;
- and
- calculating the degree of influence corresponding to the error or the attribute on the basis of a comparison between performance of a machine learning model generated with use of the target data and performance of a machine learning model generated with use of the evaluation data;
- a determination process for determining data to be corrected in the target data on the basis of the degrees of influence calculated in the calculation process, wherein the determination process includes determining, from among pieces of data included in the target data, the data to be corrected in accordance with one or more errors or one or more attributes of one or more errors included in the pieces of data and the degrees of influence corresponding to the one or more errors or the one or more attributes;
- and
- a data cleansing process for correcting the data that has been determined to be corrected in the target data.
5 more elements
Dependencies are inferred from the claim text.