Systems and methods for machine learning model generation

Patent metadata

Field	Value
Publication number	US-12468923-B2
Application number	US-202418409687-A
Country	US
Kind code	B2
Filing date	Jan 10, 2024
Priority date	Jan 10, 2024
Publication date	Nov 11, 2025
Grant date	Nov 11, 2025

Abstract

An apparatus and method for machine learning model generation are described herein. In some embodiments, an apparatus may detect a divergence in system state data and may determine a machine learning model using a metamodel. In some embodiments, apparatus may then train the machine learning model. In some embodiments, apparatus may then determine a visual element data structure.

First claim

What is claimed is:

1 . An apparatus for machine learning model generation, the apparatus comprising:
at least a processor; and a memory communicatively connected to the at least a processor, the memory containing instructions configuring the at least a processor to:
obtain system state data, wherein the system state data comprises a plurality of points;
receive a metamodel training data set, wherein the metamodel training data set comprises historical system state data and historical machine learning model forms which were successfully applied to the historical system state data;
sanitize the metamodel training data set using a dedicated hardware unit comprising circuitry configured to perform signal processing operations, wherein sanitizing the metamodel training data set comprises:
determining by the dedicated hardware unit that at least one training data entry of the metamodel training data set has a signal to noise ratio below a threshold value; and
removing the at least one training data entry from the metamodel training data set to create sanitized metamodel training data set;

iteratively train a metamodel as a function of the sanitized metamodel training data set;
generate a machine learning model form as a function of the trained metamodel, wherein generating the machine learning model form comprises:
determining a visual element data structure as a function of the generated machine learning model form, wherein determining the visual element data structure comprises:
ranking the plurality of points of the system state data;
applying rules based on a comparison between a ranking; and
transmitting the visual element data structure to a display device;

detect a divergence in the system state data as a function of feedback related to an effectiveness of the machine learning model form; and
retrain the metamodel using one or more results indicating the effectiveness of the machine learning model form.

2 . The apparatus of claim 1 , wherein detecting the divergence in the system state data comprises comparing the system state data to a stable data set.

3 . The apparatus of claim 2 , wherein the memory contains instructions configuring the at least a processor to identify the divergence in the system state data if the system state data departs from a distribution of data of the stable data set.

4 . The apparatus of claim 1 , wherein detecting the divergence in the system state data comprises comparing a parameter of the system state data to a predetermined threshold.

5 . The apparatus of claim 1 , wherein the memory contains instructions configuring the at least a processor to add the system state data to a stable data set if no divergence in the system state data is detected.

6 . The apparatus of claim 1 , wherein generating the machine learning model form comprises:
identifying the metamodel training data set; training the metamodel on the metamodel training data set; inputting the system state data into the metamodel; and receiving a machine learning model form datum.

7 . The apparatus of claim 6 , wherein the metamodel comprises a classifier.

8 . The apparatus of claim 1 , wherein the memory contains instructions configuring the at least a processor to train a machine learning model of the machine learning model form using the system state data.

9 . The apparatus of claim 1 , wherein the memory contains instructions configuring the at least a processor to:
determine the visual element data structure as a function of the machine learning model form; and transmit the visual element data structure to a user device.

10 . The apparatus of claim 9 , wherein the visual element data structure configures the user device to display a visual element to a user.

11 . A method for machine learning model generation, comprising:
using at least a processor, obtaining system state data, wherein the system state data comprises a plurality of points; using the at least a processor, receiving a metamodel training data set, wherein the metamodel training data set comprises historical system state data and historical machine learning model forms which were successfully applied to the historical system state data; using the at least a processor, sanitizing the metamodel training data set using a dedicated hardware unit comprising circuitry configured to perform signal processing operations, wherein sanitizing the metamodel training data set comprises:
determining by the dedicated hardware unit that at least one training data entry of the metamodel training data set has a signal to noise ratio below a threshold value; and
removing the at least one training data entry from the metamodel training data set to create sanitized metamodel training data set;
using the at least a processor, iteratively training a metamodel as a function of the sanitized metamodel training data set; using the at least a processor, generating a machine learning model form as a function of the trained metamodel, wherein generating the machine learning model form comprises:
determining a visual element data structure as a function of the generated machine learning model form, wherein determining the visual element data structure comprises:
ranking the plurality of points of the system state data;
applying rules based on a comparison between a ranking; and
transmitting the visual element data structure to a display device;

using the at least a processor, detecting a divergence in the system state data as a function of feedback related to an effectiveness of the machine learning model form; and using the at least a processor, retraining the metamodel using one or more results indicating the effectiveness of the machine learning model form.

12 . The method of claim 11 , wherein detecting the divergence in the system state data comprises comparing the system state data to a stable data set.

13 . The method of claim 12 , wherein the divergence in the system state data is identified if the system state data departs from a distribution of data of the stable data set.

14 . The method of claim 11 , wherein detecting the divergence in the system state data comprises comparing a parameter of the system state data to a predetermined threshold.

15 . The method of claim 11 , further comprising adding the system state data to a stable data set if no divergence in the system state data is detected.

16 . The method of claim 11 , wherein generating the machine learning model form comprises:
using the at least a processor, identifying the metamodel training data set; using the at least a processor, training the metamodel on the metamodel training data set; using the at least a processor, inputting the system state data into the metamodel; and using the at least a processor, receiving a machine learning model form datum.

17 . The method of claim 16 , wherein the metamodel comprises a classifier.

18 . The method of claim 11 , further comprising using the at least a processor, training a machine learning model of the machine learning model form using the system state data.

19 . The method of claim 11 , further comprising:
using the at least a processor, determining the visual element data structure as a function of the machine learning model form; and using the at least a processor, transmitting the visual element data structure to a user device.

20 . The method of claim 19 , wherein the visual element data structure configures the user device to display a visual element to a user.

Assignees

• The Strategic Coach Inc

Inventors

• Smith Barbara Sue
• Sullivan Daniel J

Classifications

• G06N3/08

  Learning methods · CPC title

• G06N3/0464 Primary

  Convolutional networks [CNN, ConvNet] · CPC title