	US 12,340,569 B2
	Method and apparatus for training neural network models to increase performance of the neural network models
Jaewook Yoo, Suwon-si (KR); Dokwan Oh, Hwaseong-si (KR); and Dasol Han, Suwon-si (KR)
Assigned to Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed by Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed on Mar. 16, 2022, as Appl. No. 17/696,354.
Claims priority of application No. 10-2021-0064266 (KR), filed on May 18, 2021; and application No. 10-2021-0091360 (KR), filed on Jul. 13, 2021.
Prior Publication US 2022/0383623 A1, Dec. 1, 2022
Int. Cl. G06V 10/776 (2022.01); G06V 10/764 (2022.01); G06V 10/77 (2022.01); G06V 10/82 (2022.01)

CPC G06V 10/776 (2022.01) [G06V 10/764 (2022.01); G06V 10/7715 (2022.01); G06V 10/82 (2022.01)]

17 Claims

1. A processor-implemented method of training a neural network model, the method comprising:

receiving input data and target data;

pooling, by a neural network model, on each of plural classes represented in a feature map, extracted from the input data, based on respective probabilities for each of the plural classes represented in the feature map;

generating output data by inputting the input data to the neural network model;

determining a first loss based on a comparing of the output data and the target data, and an auxiliary loss based on a result of the pooling; and

training the neural network model based on the first loss and the auxillary loss.