US 12,248,880 B2
Using batches of training items for training a network
Eric A. Sather, Palo Alto, CA (US); Steven L. Teig, Menlo Park, CA (US); and Andrew C. Mihal, San Jose, CA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Aug. 27, 2023, as Appl. No. 18/238,507.
Application 18/238,507 is a continuation of application No. 17/514,701, filed on Oct. 29, 2021, granted, now 11,741,369.
Application 17/514,701 is a continuation of application No. 16/852,329, filed on Apr. 17, 2020, granted, now 11,163,986, issued on Nov. 2, 2021.
Application 16/852,329 is a continuation of application No. 15/901,456, filed on Feb. 21, 2018, granted, now 10,671,888, issued on Jun. 2, 2020.
Claims priority of provisional application 62/599,013, filed on Dec. 14, 2017.
Prior Publication US 2023/0409918 A1, Dec. 21, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/084 (2023.01); G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06N 3/04 (2023.01); G06N 3/08 (2023.01); G06T 7/00 (2017.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 40/16 (2022.01)
CPC G06N 3/084 (2013.01) [G06F 18/214 (2023.01); G06F 18/217 (2023.01); G06N 3/04 (2013.01); G06N 3/08 (2013.01); G06T 7/97 (2017.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 40/167 (2022.01); G06V 40/172 (2022.01); G06T 2207/20081 (2013.01); G06T 2207/30201 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for training a machine-trained (MT) network that classifies inputs into a plurality of categories, the method comprising:
propagating a plurality of input training items through the MT network to generate a respective output value for each respective input training item, the plurality of input training items comprising input training items for each of the categories;
identifying a plurality of triplets from the plurality of input training items, wherein each respective triplet comprises two input training items in a respective first category and one input training item in a respective second category, wherein a plurality of the input training items belong to at least two different triplets;
calculating a value of a loss function as a summation of respective individual loss functions for each of the respective identified triplets, the respective individual loss function for each respective triplet based on the output values generated for the input training items of the triplet; and
training the MT network using the calculated loss function value.