US 11,790,279 B2
System and method for class specific deep learning
Arathi Sreekumari, Bangalore (IN); Radhika Madhavan, Latham, NY (US); Suresh Emmanuel Joel, Bangalore (IN); and Hariharan Ravishankar, Bangalore (IN)
Assigned to General Electric Company, Schenectady, NY (US)
Filed by General Electric Company, Schenectady, NY (US)
Filed on Jul. 14, 2022, as Appl. No. 17/864,694.
Application 17/864,694 is a continuation of application No. 16/282,592, filed on Feb. 22, 2019, granted, now 11,410,086.
Prior Publication US 2022/0366321 A1, Nov. 17, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 20/20 (2019.01); G06N 20/10 (2019.01); G06N 3/044 (2023.01); G06N 3/047 (2023.01)
CPC G06N 20/20 (2019.01) [G06N 20/10 (2019.01); G06N 3/044 (2023.01); G06N 3/047 (2023.01)] 20 Claims
OG exemplary drawing
 
11. A system, comprising:
a data acquisition unit that acquires an input dataset corresponding to a physical process of a subject, wherein the input dataset is generated by at least one of a machine;
a database unit communicatively coupled to the data acquisition unit that stores a plurality of reference models corresponding to a plurality of classes of the physical process, wherein each of the plurality of reference models comprises a corresponding plurality of latent space variables of corresponding versions of a machine learning model;
a model generation unit communicatively coupled to the database unit that generates a new data model based on the input dataset, wherein the new data model comprises a plurality of latent space variables of a new version of the machine learning model;
a controller unit communicatively coupled to the model generation unit that:
compares the new data model with each of the plurality of reference models to generate a plurality of distance metric values;
selects a reference model among the plurality of reference models based on the plurality of distance metric values; and
classifies the physical process of the subject with one of the plurality of classes based on the one of the plurality of classes corresponding to the reference model.