System and method for automatic labeling of clusters created by machine learning methods
AssigneePaypal, Inc.
InventorAviv Ben-Arie
Life of this patent
Estimating the term…
Abstract
Aspects of the present disclosure involve systems, methods, devices, and the like for auto-labeling clusters generated by machine learning models. In one embodiment, a system is introduced that can perform a series of operations for determining comprehensive labels for clusters output from machine learning methods used to classify data sets. The auto-labeling system may include generating labels determined using a computation of a frequency count, ratio, and coverage. These computations may use feature-based dictionaries which aid in the determination, storage, and analysis of the relevant features useful in labeling the clusters.
Claims20
A system, comprising:
- a non-transitory memory;
- and
- one or more hardware processors coupled with the non-transitory memory and configured to execute instructions from the non-transitory memory to cause the system to perform operations comprising:
- retrieving a set of clusters output by a machine learning model, each of the set of clusters being grouped based on one or more similarities;
- retrieving a set of features used by the machine learning model to cluster an original dataset into the set of clusters;
- transforming each of the set of features from a numerical representation to a state readable by an auto-labeling algorithm;
- determining a frequency count for each of the set of features in each of the set of clusters;
- creating dictionaries corresponding to each of the set of features for each of the set of clusters;
- determining a ratio value and coverage for each of the set of features in each of the set of clusters based on the frequency count and the dictionaries;
- and
- generating, for each of the set of clusters using the auto-labeling algorithm and based on the ratio value and coverage for each of the set of features, labels corresponding to the set of features based on relevancy to each of the set of clusters.
A method, comprising:
- retrieving a set of clusters output by a machine learning model, each of the set of clusters being grouped based on one or more similarities;
- retrieving a first set of features used by the machine learning model to cluster an original dataset into the set of clusters;
- adding a second set of features not used by the machine learning model to the first set of features to create a third set of feature;
- transforming each of the third set of features from a numerical representation to a state readable by an auto-labeling algorithm;
- determining a frequency count for each of the third set of features in each of the set of clusters;
- creating dictionaries corresponding to each of the third set of features for each of the set of clusters;
- determining a ratio value and coverage for each of the third set of features in each of the set of clusters based on the frequency count and the dictionaries;
- and
- generating, for each of the set of clusters using the auto-labeling algorithm and based on the ratio value and coverage for each of the third set of features, labels corresponding to the third set of features based on relevancy to each of the set of clusters.
A non-transitory machine-readable medium having instructions stored thereon, the instructions executable to cause performance of operations comprising:
- retrieving a set of clusters output by a machine learning model, each of the set of clusters being grouped based on one or more similarities;
- retrieving a set of features used by the machine learning model to cluster an original dataset into the set of clusters;
- transforming each of the set of features from a numerical representation to a state readable by an auto-labeling algorithm;
- determining a frequency count for each of the set of features in each of the set of clusters;
- creating dictionaries corresponding to each of the set of features for each of the set of clusters;
- determining a ratio value and coverage for each of the set of features in each of the set of clusters based on the frequency count and the dictionaries;
- and
- generating, for each of the set of clusters using the auto-labeling algorithm and based on the ratio value and coverage for each of the set of features, labels corresponding to the set of features based on relevancy to each of the set of clusters.
Dependencies are inferred from the claim text.