Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
SYSTEM FOR DETERMINING CONTROL PARAMETERS FOR OPTIMIZING ACCELERATOR PERFORMANCE USING REINFORCEMENT LEARNING AND MACHINE LEARNING TECHNIQUES
Document Type and Number:
WIPO Patent Application WO/2019/164125
Kind Code:
A1
Abstract:
A system for determining parameters for accelerator control devices is disclosed. The system for determining parameters for accelerator control devices of the present invention comprises: a plurality of device simulators corresponding to a plurality of respective accelerator control devices, for learning and simulating on the basis of an artificial neural network; and a machine learning processing unit for specifying a control parameter collection set corresponding to at least one collection of control parameters which the plurality of device simulators have, and for learning, through a learning process on the basis of an artificial neural network, changes to the values of the control parameters included in the control parameter collection set and the final output quality of an accelerator according to the changes, and then calculating the optimal values of the control parameters which can maximize the final output quality of the accelerator.

Inventors:
LEE, Geunho (108-102, 25 Yeondong-ro 13-gil,Guro-gu, Seoul, 08360, KR)
KIM, Jinwon (110-1903, 17 Dongnimmungongwon-gil,Seodaemun-gu, Seoul, 03733, KR)
Application Number:
KR2019/000514
Publication Date:
August 29, 2019
Filing Date:
January 14, 2019
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
MOBIIS CO., LTD. (C-203, 253 Pangyo-ro Bundang-gu,,Seongnam-si, Gyeonggi-do, 13486, KR)
International Classes:
G05B13/02; G05B13/04; G06N3/08; G06N20/00
Attorney, Agent or Firm:
KIM, Ham-Kon (2nd Floor, 522 Nonhyeon-ro,Gangnam-gu, Seoul, 06140, KR)
Download PDF: