Back to results
US 12,613,941 B1Granted
System, method, and computer program for multi-stage multi-class classification
AssigneeAmdocs Development Limited
InventorsSayali Zirpe, Alexander Murin
Life of this patent
Estimating the term…
Abstract
As described herein, a system, method, and computer program are provided for a multi-stage multi-class classification. A dataset having an imbalanced distribution of data across a plurality of classes is identified. The plurality of classes are grouped into a plurality of clusters, based on a defined criteria. A plurality of machine learning models are trained, each machine learning model of the plurality of machine learning models trained using a subset of the data in the dataset corresponding to a particular cluster of the plurality of clusters. The plurality of machine learning models are used, in stages, to predict a classification for a given input.
Claims19
A non-transitory computer-readable media storing computer instructions which when executed by one or more processors of a device cause the device to:
- identify a dataset having an imbalanced distribution of data across a plurality of classes;
- group the plurality of classes into a plurality of clusters, based on a defined criteria, such that each cluster in the plurality of clusters includes two or more classes in the plurality of classes that are balanced with respect to one another as determined based on the defined criteria;
- train a plurality of machine learning models to classify data, each machine learning model of the plurality of machine learning models trained using a subset of the data in the dataset corresponding to a particular cluster of the plurality of clusters such that the machine learning model learns to classify data according to the two or more classes included in the particular cluster that are balanced with respect to one another as determined based on the defined criteria,
- wherein respectively training the plurality of machine learning models per cluster of the plurality of clusters eliminates bias that would otherwise be introduced in a machine learning model trained on the dataset with the imbalanced distribution of data across the plurality of classes;
- and
- process a given input through the plurality of trained machine learning models, in stages, until a classification for the given input is predicted.
3 more elements
A method, comprising:
- at a computer system:
- identifying a dataset having an imbalanced distribution of data across a plurality of classes;
- grouping the plurality of classes into a plurality of clusters, based on a defined criteria, such that each cluster in the plurality of clusters includes two or more classes in the plurality of classes that are balanced with respect to one another as determined based on the defined criteria;
- training a plurality of machine learning models to classify data, each machine learning model of the plurality of machine learning models trained using a subset of the data in the dataset corresponding to a particular cluster of the plurality of clusters such that the machine learning model learns to classify data according to the two or more classes included in the particular cluster that are balanced with respect to one another as determined based on the defined criteria,
- wherein respectively training the plurality of machine learning models per cluster of the plurality of clusters eliminates bias that would otherwise be introduced in a machine learning model trained on the dataset with the imbalanced distribution of data across the plurality of classes;
- and
- processing a given input through the plurality of trained machine learning models, in stages, until a classification for the given input is predicted.
4 more elements
A system, comprising:
- a non-transitory memory storing instructions;
- and
- one or more processors in communication with the non-transitory memory that execute the instructions to:
- identify a dataset having an imbalanced distribution of data across a plurality of classes;
- group the plurality of classes into a plurality of clusters, based on a defined criteria, such that each cluster in the plurality of clusters includes two or more classes in the plurality of classes that are balanced with respect to one another as determined based on the defined criteria;
- train a plurality of machine learning models to classify data, each machine learning model of the plurality of machine learning models trained using a subset of the data in the dataset corresponding to a particular cluster of the plurality of clusters such that the machine learning model learns to classify data according to the two or more classes included in the particular cluster that are balanced with respect to one another as determined based on the defined criteria,
- wherein respectively training the plurality of machine learning models per cluster of the plurality of clusters eliminates bias that would otherwise be introduced in a machine learning model trained on the dataset with the imbalanced distribution of data across the plurality of classes;
- and
- process a given input through the plurality of trained machine learning models, in stages, until a classification for the given input is predicted.
6 more elements
Dependencies are inferred from the claim text.