Presentation 2005/3/21
A Language Processing Neural Network with Inference Ability
Keitaro KATAOKA, Masafumi HAGIWARA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we propose a language processing neural network with inference ability. At first input sentences are decomposed to triples representation by morphological analysis and syntax analysis. The network consists of a memory part and a control layer. The memory part consists of three layers. The memory part memorizes both triples representations and sentences. The control layer memorizes the result of inference and controls inference. From the computer simulation, the followings have been confirmed : 1) The network can memorize several sentences ; 2) The network can learn the difference between subject and object ; 3) The network can carry out some kinds of inferences.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) neural network / area representation / triples-representation / memory / natural language
Paper # NC2004-168
Date of Issue

Conference Information
Committee NC
Conference Date 2005/3/21(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Language Processing Neural Network with Inference Ability
Sub Title (in English)
Keyword(1) neural network
Keyword(2) area representation
Keyword(3) triples-representation
Keyword(4) memory
Keyword(5) natural language
1st Author's Name Keitaro KATAOKA
1st Author's Affiliation Department of Information and Computer Science, Faculty of Science and Technology, Keio University()
2nd Author's Name Masafumi HAGIWARA
2nd Author's Affiliation Department of Information and Computer Science, Faculty of Science and Technology, Keio University
Date 2005/3/21
Paper # NC2004-168
Volume (vol) vol.104
Number (no) 758
Page pp.pp.-
#Pages 6
Date of Issue