Granted Patent B2

US 12,586,002 · App. 17/491,494 · Granted Mar 24, 2026

Inventors: Dzung Tien Phan (Pleasantville, NY); Lam Minh Nguyen (Ossining, NY); Jayant R. Kalagnanam (Briarcliff Manor, NY); Chandrasekhara K. Reddy (Kinnelon, NJ); Srideepika Jayaraman (White Plains, NY)

Assignee: International Business Machines Corporation

G06N20/10

Abstract

Claims (80)

1 . A computer-implemented method for improving an efficiency of a computing device in generating a classifier engine for machine learning, the computer-implemented method comprising:

receiving a plurality of data points;

applying a semi-supervised k-means process to a first set of data points of the plurality of data points and a second set of data points of the plurality of data points, wherein

the first set of data points is labeled with a first class of a plurality of classes and

the second set of data points is labeled with a second class of the plurality of classes, different from the first class;

clustering the first set of data points and the second set of data points into a first plurality of clusters and a second plurality of clusters, respectively, using the semi-supervised k-means process;

constructing multi-polytopes for the first plurality of clusters and the second plurality of clusters;

training, based on the constructing of the multi-polytopes, a plurality of support vector machine (SVM) classifiers, wherein each SVM classifier of the plurality of SVM classifiers is trained on a respective pair of clusters within the first plurality of clusters and the second plurality of clusters;

determining, using the trained plurality of SVM classifiers, separation hyperplanes for the first plurality of clusters and the second plurality of clusters; and

determining labels for each cluster of the first plurality of clusters and each cluster of the second plurality of clusters, based on the separation hyperplanes.

2 . The computer-implemented method of claim 1 , further comprising clustering positive classes of data points of the plurality of data points separately from negative classes of data points of the plurality of data points.

3 . The computer-implemented method of claim 1 , further comprising:

identifying information from one of the first plurality of clusters or the second plurality of clusters; and

using the identified information as a constraint in determining remaining of the one of the first plurality of clusters or the second plurality of clusters.

4 . The computer-implemented method of claim 1 , further comprising:

determining misclassification measurements in the first plurality of clusters and the second plurality of clusters; and

training the semi-supervised k-means process for improved clustering based on the misclassification measurements.

5 . The computer-implemented method of claim 1 , wherein the separation hyperplanes are non-linear boundaries between the respective pair of clusters.

6 . The computer-implemented method of claim 1 , wherein the semi-supervised k-means process is regularized.

7 . A computer program product for improving a computing efficiency of a computing device in generating a classifier engine for machine learning, the computer program product comprising:

one or more non-transitory computer readable storage media, and program instructions collectively stored on the one or more non-transitory computer readable storage media, the program instructions comprising:

receiving a plurality of data points;

applying a semi-supervised k-means process to a first set of data points of the plurality of data points and a second set of data points of the plurality of data points, wherein

the first set of data points is labeled with a first class of a plurality of classes, and

the second set of data points is labeled with a second class of the plurality of classes, different from the first class;

clustering the first set of data points and the second set of data points into a first plurality of clusters and a second plurality of clusters, respectively, using the semi-supervised k-means process;

constructing multi-polytopes for the first plurality of clusters and the second plurality of clusters;

training, based on the constructing of the multi-polytopes, a plurality of support vector machine (SVM) classifiers, wherein each SVM classifier of the plurality of SVM classifiers is trained on a respective pair of clusters within the first plurality of clusters and the second plurality of clusters;

determining, using the trained plurality of SVM classifiers, separation hyperplanes for the first plurality of clusters and the second plurality of clusters; and

determining labels for each cluster of the first plurality of clusters and each cluster of the second plurality of clusters, based on the separation hyperplanes.

8 . The computer program product of claim 7 , wherein the program instructions further comprise clustering positive classes of data points of the plurality of data points separately from negative classes of data points of the plurality of data points.

9 . The computer program product of claim 7 , wherein the program instructions further comprise:

identifying information from one of the first plurality of clusters or the second plurality of clusters; and

using the identified information as a constraint in determining remaining of the one of the first plurality of clusters or the second plurality of clusters.

10 . The computer program product of claim 7 , wherein the program instructions further comprise:

determining misclassification measurements in the first plurality of clusters and the second plurality of clusters; and

training the semi-supervised k-means process for improved clustering based on the misclassification measurements.

11 . The computer program product of claim 7 , wherein the separation hyperplanes are non-linear boundaries between the respective pair of clusters.

12 . The computer program product of claim 7 , wherein the semi-supervised k-means process is regularized.

13 . A computer server for improving computational efficiency in generating a classifier engine for machine learning, the computer server comprising:

a network connection;

one or more computer readable storage media;

a processor coupled to the network connection and coupled to the one or more computer readable storage media; and

a computer program product comprising program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising:

receiving a plurality of data points;

applying a semi-supervised k-means process to a first set of data points of the plurality of data points and a second set of data points of the plurality of data points, wherein

the first set of data points is labeled with a first class of a plurality of classes, and

the second set of data points is labeled with a second class of the plurality of classes, different from the first class;

clustering the first set of data points and the second set of data points into a first plurality of clusters and a second plurality of clusters, respectively, using the semi-supervised k-means process;

constructing multi-polytopes for the first plurality of clusters and the second plurality of clusters;

training, based on the constructing of the multi-polytopes, a plurality of support vector machine (SVM) classifiers, wherein each SVM classifier of the plurality of SVM classifiers is trained on a respective pair of clusters within the first plurality of clusters and the second plurality of clusters;

determining, using the trained plurality of SVM classifiers, separation hyperplanes for the first plurality of clusters and the second plurality of clusters; and

determining labels for each cluster of the first plurality of clusters and each cluster of the second plurality of clusters, based on the separation hyperplanes.

14 . The computer server of claim 13 , wherein the program instructions further comprise clustering positive classes of data points of the plurality of data points separately from negative classes of data points of the plurality of data points.

15 . The computer server of claim 13 , wherein the program instructions further comprise;

identifying information from one of the first plurality of clusters or the second plurality of clusters; and

using the identified information as a constraint in determining remaining of the one of the first plurality of clusters or the second plurality of clusters.

16 . The computer server of claim 13 , wherein the program instructions further comprise:

determining misclassification measurements in the first plurality of clusters and the second plurality of clusters clustered; and

training the semi-supervised k-means process for improved clustering based on the misclassification measurements.

17 . The computer server of claim 13 , wherein the semi-supervised k-means process is regularized.

18 . A software as a service (SaaS) computer-implemented method for improving a computational efficiency in generating a classifier engine for machine learning, the SaaS computer-implemented comprising:

receiving a plurality of data points and a request for classification services from a remote user through a network connection;

applying, by a local computer server, a semi-supervised k-means process to a first set of data points of the plurality of data points and a second set of data points of the plurality of data points, wherein

the first set of data points is labeled with a first class of a plurality of classes, and

the second set of data points is labeled with a second class of the plurality of classes, different from the first class;

clustering the first set of data points and the second set of data points into a first plurality of clusters and a second plurality of clusters, respectively, using the semi-supervised k-means process;

constructing multi-polytopes for the first plurality of clusters and the second plurality of clusters;

training, based on the constructing of the multi-polytopes, a plurality of support vector machine (SVM) classifiers, wherein each SVM classifier of the plurality of SVM classifiers is trained on a respective pair of clusters within the first plurality of clusters and the second plurality of clusters;

determining, using the trained plurality of SVM classifiers, separation hyperplanes for the first plurality of clusters and the second plurality of clusters;

determining labels for each cluster of the first plurality of clusters and each cluster of the second plurality of clusters, based on the separation hyperplanes; and

providing the determined labels for each cluster of the first plurality of clusters and each cluster of the second plurality of clusters, to the remote user.

19 . The SaaS computer-implemented method of claim 18 , further comprising clustering positive classes of data points of the plurality of data points separately from negative classes of data points of the plurality of data points.

20 . The SaaS computer-implemented method of claim 18 , further comprising:

identifying information from one of the first plurality of clusters or the second plurality of clusters; and

using the identified information as a constraint in determining remaining of the one of the first plurality of clusters or the second plurality of clusters.

21 . The SaaS computer-implemented method of claim 18 , further comprising:

determining misclassification measurements in the first plurality of clusters and the second plurality of clusters clustered; and

training the semi-supervised k-means process for improved clustering based on the misclassification measurements.

22 . The SaaS computer-implemented method of claim 18 , wherein the separation hyperplanes are non-linear boundaries between the respective pair of clusters.

Assignments (1)

Continuity (1)

References Cited (29)