US 12,236,331 B2
Method and system of DNN modularization for optimal loading
Brijraj Singh, Bengaluru (IN); Mayukh Das, Bengaluru (IN); Yash Hemant Jain, Bengaluru (IN); Sharan Kumar Allur, Bengaluru (IN); Venkappa Mala, Bengaluru (IN); and Praveen Doreswamy Naidu, Bengaluru (IN)
Assigned to SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Appl. No. 17/430,644
Filed by SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
PCT Filed Jul. 9, 2021, PCT No. PCT/KR2021/008776
§ 371(c)(1), (2) Date Aug. 12, 2021,
PCT Pub. No. WO2022/035058, PCT Pub. Date Feb. 17, 2022.
Claims priority of application No. 202041034900 (IN), filed on Aug. 13, 2020; and application No. 202041034900 (IN), filed on Mar. 1, 2021.
Prior Publication US 2023/0153565 A1, May 18, 2023
Int. Cl. G06N 3/04 (2023.01); G06N 3/092 (2023.01)
CPC G06N 3/04 (2013.01) [G06N 3/092 (2023.01)] 17 Claims
OG exemplary drawing
 
15. A method of loading an artificial intelligence (AI) driven neural model in an electronic device, the method comprising:
identifying, by the electronic device, at least one main task of an application that is executed on the electronic device;
identifying, by the electronic device, a plurality of individual sub-tasks to be executed for completing the identified at least one main task;
identifying, by the electronic device, a parent AI model to be deployed for performing the identified at least one main task;
identifying, by the electronic device, a plurality of child AI models to be deployed for performing the identified plurality of individual sub-tasks;
determining, by the electronic device, using a pre-trained AI model, one or more of the identified plurality of child AI models to be executed in parallel for completing the identified plurality of individual sub-tasks; and
loading, by the electronic device, the determined one or more of the plurality of child AI models in parallel to execution of the identified at least one main task.