Presentation 1999/3/5
From word representation vectors to phrase representation vectors
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Miikkulainen extended back-propagation to modify not only link weights but also input signals given to neural networks. He trained his neural network, ca11ed FGREP, with this "extened back-propagation" to learn case-role assignment, and showed that FGREP acquired vectorial respresentations of words based on a corpus.The authors modified the output format of FGREP so that the learning becomes faster and more stable. We also added two additional networks so that noun phrases of the form "adjective+noun" are accepted as syntactic constituent of the input sentence, in addition to simple nouns.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) semantic representation form / neural network / FGREP / case-role assignment / auto-association
Paper # TL98-21
Date of Issue

Conference Information
Committee TL
Conference Date 1999/3/5(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Thought and Language (TL)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) From word representation vectors to phrase representation vectors
Sub Title (in English)
Keyword(1) semantic representation form
Keyword(2) neural network
Keyword(3) FGREP
Keyword(4) case-role assignment
Keyword(5) auto-association
1st Author's Name Naoto TAKAHASHI
1st Author's Affiliation Machine Understanding Division, Electrotechnical Laboratory()
2nd Author's Name Minoru MOTOKI
2nd Author's Affiliation Faculty of Engineering Kyushu Sangyo University
Date 1999/3/5
Paper # TL98-21
Volume (vol) vol.98
Number (no) 640
Page pp.pp.-
#Pages 8
Date of Issue