US 11,868,443 B1
System for training neural network using ordered classes
Rajeev Ranjan, Seattle, WA (US); Prithviraj Banerjee, Redmond, WA (US); Manoj Aggarwal, Seattle, WA (US); Gerard Guy Medioni, Los Angeles, CA (US); and Dilip Kumar, Seattle, WA (US)
Assigned to AMAZON TECHNOLOGIES, INC., Seattle, WA (US)
Filed by AMAZON TECHNOLOGIES, INC., Seattle, WA (US)
Filed on May 12, 2021, as Appl. No. 17/302,770.
Int. Cl. G06F 18/2431 (2023.01); G06N 3/08 (2023.01); G06F 18/214 (2023.01); G06F 18/2113 (2023.01); G06N 3/045 (2023.01)
CPC G06F 18/2431 (2023.01) [G06F 18/214 (2023.01); G06F 18/2113 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system comprising:
a memory, storing first computer-executable instructions; and
a hardware processor to execute the first computer-executable instructions to:
determine a set of ordered classes, wherein the set of ordered classes is indicative of a continuum of classes;
determine training data comprising a plurality of samples, wherein each sample of the plurality of samples comprises input data and a corresponding label value indicative of one of the ordered classes of the set of ordered classes; and
train, using the training data and an ordered classification loss function, a neural network to determine a first value associated with a first ordered class of the set of ordered classes.