Title:
DISTRIBUTED DEEP LEARNING SYSTEM
Document Type and Number:
WIPO Patent Application WO/2021/009847
Kind Code:
A1
Abstract:
A distributed deep learning system (100) according to the present invention comprises: M number of distributed processing nodes (102) for performing neural network deep learning in a mutually distributed manner; and N number of aggregation processing nodes (101) that are connected, via a first communication line (103) and second communication line (111), to the M number of distributed processing nodes, respectively, and aggregate, via the first communication line (103), the distributed processing results obtained by the M number of distributed processing nodes (102). This feature enables efficient and stable distributed deep learning processing even when multiple users are simultaneously sharing the distributed deep learning system (100).
Inventors:
ITO TSUYOSHI (JP)
KAWAI KENJI (JP)
KATO JUNICHI (JP)
NGO HUYCU (JP)
ARIKAWA YUKI (JP)
SAKAMOTO TAKESHI (JP)
TANAKA KENJI (JP)
KAWAI KENJI (JP)
KATO JUNICHI (JP)
NGO HUYCU (JP)
ARIKAWA YUKI (JP)
SAKAMOTO TAKESHI (JP)
TANAKA KENJI (JP)
Application Number:
PCT/JP2019/027922
Publication Date:
January 21, 2021
Filing Date:
July 16, 2019
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G06N3/08
Foreign References:
US20190042884A1 | 2019-02-07 | |||
JP2019080232A | 2019-05-23 |
Attorney, Agent or Firm:
YAMAKAWA, Shigeki et al. (JP)
Download PDF: