US 11,790,228 B2
Methods and systems for performing tasks on media using attribute specific joint learning
Kunal Swami, Bengaluru (IN); Prasanna Vishnu Bondada, Bengaluru (IN); Pankaj Kumar Bajpai, Bengaluru (IN); Mahesh P J, Bengaluru (IN); and Manoj Kumar, Bengaluru (IN)
Assigned to SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed by SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed on Jan. 21, 2021, as Appl. No. 17/154,580.
Claims priority of application No. 202041002429 (IN), filed on Jan. 20, 2020; and application No. 202041002429 (IN), filed on Jan. 18, 2021.
Prior Publication US 2021/0224652 A1, Jul. 22, 2021
Int. Cl. G06N 3/08 (2023.01); G06V 20/40 (2022.01); G06F 18/214 (2023.01); G06V 10/82 (2022.01); G06T 7/50 (2017.01); G06V 10/772 (2022.01); G06V 10/774 (2022.01)
CPC G06N 3/08 (2013.01) [G06F 18/2155 (2023.01); G06T 7/50 (2017.01); G06V 10/772 (2022.01); G06V 10/7753 (2022.01); G06V 10/82 (2022.01); G06V 20/41 (2022.01)] 20 Claims
OG exemplary drawing
 
1. An electronic device comprising:
a memory; and
a processor coupled to the memory and configured to:
group a plurality of training datasets into a plurality of groups of training datasets based on a plurality of attributes,
obtain a plurality of output batches for the plurality of groups of training datasets by sampling each group of training datasets based on a batch proportion of the plurality of groups of training datasets,
obtain a plurality of loss values of the plurality of groups of training datasets respectively, and
train a learning-based model based on the plurality of loss values.
 
11. A control method of an electronic device comprising:
grouping a plurality of training datasets into a plurality of groups of training datasets based on a plurality of attributes;
obtaining a plurality of output batches for the plurality of groups of training datasets by sampling each group of the plurality of groups of training datasets based on a batch proportion of the plurality of groups of training datasets;
obtaining a plurality of loss values of the plurality of groups of training datasets respectively; and
training a learning-based model based on the plurality of loss values.
 
20. The electronic device of claim 1, wherein the processor is further configured to obtain the plurality of loss values of the plurality of groups of training datasets respectively by applying an attribute specific loss function of the each of the plurality of groups of training datasets on each of the plurality of output batches.