US 11,928,595 B2
Method of managing data representation for deep learning, method of processing data for deep learning and deep learning system performing the same
Seung-Soo Yang, Hwaseong-si (KR)
Assigned to SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed by SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed on May 24, 2022, as Appl. No. 17/751,720.
Application 17/751,720 is a continuation of application No. 16/265,221, filed on Feb. 1, 2019, granted, now 11,348,004.
Claims priority of application No. 10-2018-0083807 (KR), filed on Jul. 19, 2018.
Prior Publication US 2022/0292354 A1, Sep. 15, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 9/46 (2006.01); G06F 9/48 (2006.01); G06F 9/50 (2006.01); G06N 3/08 (2023.01); G06N 20/00 (2019.01)
CPC G06N 3/08 (2013.01) [G06F 9/4881 (2013.01); G06F 9/5044 (2013.01); G06F 9/5066 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method of processing data for a deep learning system, the method comprising:
receiving path information indicating a computing path for a task including two or more operations, the computing path including a sequence of the two or more operations and a driving sequence of two or more heterogeneous resources for performing the two or more operations;
performing, using the two or more heterogeneous resources, the task based on the path information and a data representation format for each of the two or more heterogeneous resources; and
when one of the two or more heterogeneous resources is changed, changing the data representation format based on data representation information indicating an optimized data representation format for each of the two or more heterogeneous resources to enable the task to be subsequently performed using the two or more heterogeneous resources based on the changed data representation format.