US 10,769,551 B2Granted

Training data set determination

AssigneeInternational Business Machines Corporation

InventorsKanako Fujitani, Mariko Ishige, Yusuke Nishitani, Yutaka Oishi, Tatsuya Sobue, Masayuki Yamana

Life of this patent

Estimating the term…

Loading patent…

Abstract

Provided herein is a computer readable medium having instructions embodied therewith, the instructions executable by a processor or programmable circuitry of a computer to cause the processor or programmable circuitry to perform operations comprising: selecting a first set of reference training data from a training data set based on a degree of difference between each set of reference training data in the training data set and target training data; determining whether a degree of difference between the target training data and the first set of reference training data is below a first threshold; and determining whether to include the target training data in the training data set using at least a degree of difference between the target training data and at least one second set of reference training data, wherein the at least one second set of reference training data is included in the training data set and does not include the first set of reference training data.

Claims16

Dependent claims

A computer program product including one or more computer readable storage media collectively having instructions embodied therewith, the instructions executable by a processor to cause the processor to perform operations comprising:

selecting a first set of reference training data from a training data set based on a degree of difference between each set of reference training data in the training data set and target training data, wherein the first set of reference training data yields a smallest degree of difference from the target training data among all sets of reference training data in the training data set;
determining that a degree of difference between the target training data and the first set of reference training data is below a first threshold, wherein the degree of difference is based on a distance;
determining to include the target training data in the training data set in response to determining that an average degree of difference between the target training data and a first plurality of second sets of reference training data is larger than an average degree of difference between the first set of reference training data and a second plurality of second sets of reference training data, wherein the first and second plurality of second sets of reference training data are included in the training data set and do not include data from the first set of reference training data;
and
performing machine learning using the training data set with the included target training data.

2 more elements

An apparatus comprising:

a processor;
and
one or more computer readable storage media collectively including instructions that, when executed by the processor, cause the processor to perform operations comprising:
selecting a first set of reference training data from a training data set based on a degree of difference between each set of reference training data in the training data set and target training data, wherein the first set of reference training data yields a smallest degree of difference from the target training data among all sets of reference training data in the training data set;
determining that a degree of difference between the target training data and the first set of reference training data is below a first threshold, wherein the degree of difference is based on a distance;
determining to include the target training data in the training data set in response to determining that an average degree of difference between the target training data and a first plurality of second sets of reference training data is larger than an average degree of difference between the first set of reference training data and a second plurality of second sets of reference training data, wherein the first and second plurality of second sets of reference training data are included in the training data set and do not include data from the first set of reference training data;
and
performing machine learning using the training data set with the included target training data.

5 more elements

Dependencies are inferred from the claim text.