Generalized metric for machine learning model evaluation for unsupervised classification

1. A method comprising:

for each unsupervised machine learning model of a plurality of unsupervised machine learning models:

generating, by a computing device comprising one or more processors, a first set of synthetic inputs for the unsupervised machine learning model of the plurality of unsupervised machine learning models;

providing, by the computing device, the first set of synthetic inputs to the unsupervised machine learning model trained to output a prediction for each input of the first set of synthetic inputs, wherein the prediction indicates whether the input is of a first class of a plurality of classes;

identifying, by the computing device, based on an output of the unsupervised machine learning model, a second set of synthetic inputs predicted to be of the first class;

determining by the computing device, based on a set of expected normal inputs for the unsupervised machine learning model and the second set of synthetic inputs, an accuracy score for the unsupervised machine learning model; and

providing, by the computing device, the accuracy score for display via a display device to a requestor;

selecting, by the computing device, based on the corresponding accuracy scores of the plurality of unsupervised machine learning models, a particular unsupervised machine learning model of the plurality of unsupervised machine learning models for deployment into a production environment; and

deploying, by the computing device, based on the selecting, the particular unsupervised machine learning model into the production environment, wherein a software application in the production environment uses the particular unsupervised machine learning model to generate one or more predictions.

2. The method of claim 1 , wherein determining the accuracy score comprises:

determining, based on the set of expected normal inputs and the second set of synthetic inputs, an area of overlap between the set of expected normal inputs and the second set of synthetic inputs.

3. The method of claim 2 , wherein determining the accuracy score comprises:

determining, based on the set of expected normal inputs and the second set of synthetic inputs, an area of union between the set of expected normal inputs and the second set of synthetic inputs.

4. The method of claim 3 , wherein determining the accuracy score comprises:

determining a number of unique common inputs between the set of expected normal inputs and the second set of synthetic inputs; and

determining a total number of unique common inputs between the set of expected normal inputs and the second set of synthetic inputs.

5. The method of claim 1 , wherein the second set of synthetic inputs are a subset of the first set of synthetic inputs.

6. The method of claim 1 , wherein the accuracy score of the particular unsupervised machine learning model is greater than the corresponding accuracy scores of other unsupervised machine learning models of the plurality of unsupervised machine learning models.

7. The method of claim 1 , wherein the set of expected normal inputs are received from the requestor.

8. The method of claim 1 , wherein the plurality of unsupervised machine learning models are received from the requestor.

9. A processing system, comprising:

a memory comprising computer-executable instructions;

a processor configured to execute the computer-executable instructions and cause the processing system to:

for each unsupervised machine learning model of a plurality of unsupervised machine learning models:

generate a first set of synthetic inputs for the unsupervised machine learning model of the plurality of unsupervised machine learning models;

provide the first set of synthetic inputs to the unsupervised machine learning model trained to output a prediction for each input of the first set of synthetic inputs, wherein the prediction indicates whether the input is of a first class of a plurality of classes;

identify, based on an output of the unsupervised machine learning model, a second set of synthetic inputs predicted to be of the first class;

determine, based on a set of expected normal inputs for the unsupervised machine learning model and the second set of synthetic inputs, an accuracy score for the unsupervised machine learning model; and

provide the accuracy score for display on a display device to a requestor;

select, based on the corresponding accuracy scores of the plurality of unsupervised machine learning models, a particular unsupervised machine learning model of the plurality of unsupervised machine learning models for deployment into a production environment; and

deploy, based on the selecting, the particular unsupervised machine learning model into the production environment, wherein a software application in the production environment uses the particular unsupervised machine learning model to generate one or more predictions.

10. The processing system of claim 9 , wherein to determine the accuracy score comprises:

determine, based on the set of expected normal inputs and the second set of synthetic inputs, an area of overlap between the set of expected normal inputs and the second set of synthetic inputs.

11. The processing system of claim 10 , wherein to determine the accuracy score comprises:

determine, based on the set of expected normal inputs and the second set of synthetic inputs, an area of union between the set of expected normal inputs and the second set of synthetic inputs.

12. The processing system of claim 11 , wherein to determine the accuracy score comprises:

determine a number of unique common inputs between the set of expected normal inputs and the second set of synthetic inputs; and

determine a total number of unique common inputs between the set of expected normal inputs and the second set of synthetic inputs.

13. The processing system of claim 10 , wherein the second set of synthetic inputs are a subset of the first set of synthetic inputs.

14. The processing system of claim 9 , wherein the accuracy score of the particular unsupervised machine learning model is greater than the corresponding accuracy scores of other unsupervised machine learning models of the plurality of unsupervised machine learning models.