Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
TASK SCHEDULING METHOD FOR AUTOMATED MACHINE LEARNING
Document Type and Number:
WIPO Patent Application WO/2023/287109
Kind Code:
A1
Abstract:
Disclosed in this specification is a method for a terminal to schedule tasks for automated machine learning (AutoML), wherein: the ratio of 1) first tasks requiring multiple calculation devices and 2) second tasks requiring a single calculation device is set; third tasks for the AutoML are allocated on the basis of the set ratio; a request for allocating sessions is received from a user; whether it is possible to allocate the sessions is determined on the basis of the ratio of the second tasks; and the sessions can be allocated to a calculation device related to the second tasks on the basis of the ratio of the second tasks when it is possible to allocate the sessions.

More Like This:
JPH02118864INTERRUPTING SYSTEM
WO/2000/036487PRIORITIZED INSTRUCTION SCHEDULING FOR MULTI-STREAMING PROCESSORS
JPS5591033INFORMATION PROCESSOR WHICH REPORTS PROPRIETY OF INTERRUPTION
Inventors:
KIM WAN SOO (KR)
Application Number:
PCT/KR2022/009861
Publication Date:
January 19, 2023
Filing Date:
July 07, 2022
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
RIIID INC (KR)
International Classes:
G06F9/48; G06F9/50; G06N20/00
Foreign References:
KR20170116439A2017-10-19
US20200125545A12020-04-23
KR20210086448A2021-07-08
KR100646383B12006-11-14
KR102384821B12022-04-08
Other References:
KIM WANSU: "Ray: A scalable, high-performance distributed/parallel machine learning framework", RIIID TEAMBLOG, KOREA, 17 December 2020 (2020-12-17), Korea, pages 1 - 12, XP093023952, Retrieved from the Internet [retrieved on 20230215]
Attorney, Agent or Firm:
DODAM IP LAW FIRM (KR)
Download PDF:
View/Download PDF      PDF Help



 
Previous Patent: METHOD AND SYSTEM FOR INDIRECTLY ESTIMATING CALCULATIVE EXECUTION SPEED OF DEEP LEARNING MODEL

Next Patent: LITHIUM SECONDARY BATTERY