US 12,229,534 B2
Apparatus and method for developing neural network applications
Chang-Sik Cho, Daejeon (KR); Jae-Bok Park, Daejeon (KR); Kyung-Hee Lee, Daejeon (KR); Ji-Young Kwak, Daejeon (KR); Seon-Tae Kim, Daejeon (KR); and Ik-Soo Shin, Daejeon (KR)
Assigned to ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE, Daejeon (KR)
Filed by ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE, Daejeon (KR)
Filed on Jan. 18, 2023, as Appl. No. 18/098,402.
Claims priority of application No. 10-2022-0041217 (KR), filed on Apr. 1, 2022.
Prior Publication US 2023/0315402 A1, Oct. 5, 2023
Int. Cl. G06F 9/44 (2018.01); G06F 8/10 (2018.01); G06F 8/35 (2018.01); G06N 3/10 (2006.01)
CPC G06F 8/35 (2013.01) [G06F 8/10 (2013.01); G06N 3/10 (2013.01)] 14 Claims
OG exemplary drawing
 
1. An apparatus for developing a neural network application, comprising:
one or more processors; and
executable memory for storing at least one program executed by the one or more processors,
wherein the at least one program is configured to
receive a target specification and an application specification including user requirements,
limits a neural network search space required for searching for a neural network model in a database by performing a search for a backbone architecture,
search for the neural network model, corresponding to the target specification and the application specification, in the database with the limited neural network search space,
build an inference engine for performing a neural network operation used by the neural network model, and
generate a target image for executing the neural network model to be suitable for a target device using the inference engine.