Search International and National Patent Collections
Some content of this application is unavailable at the moment.
If this situation persists, please contact us atFeedback&Contact
1. (WO1991017501) METHOD OF ADJUSTING AUTOMATICALLY RULE BASE OF EXPERT SYSTEM USING NEURAL NETWORK AND DEVICE THEREOF
Latest bibliographic data on file with the International Bureau

Pub. No.: WO/1991/017501 International Application No.: PCT/JP1991/000576
Publication Date: 14.11.1991 International Filing Date: 26.04.1991
Chapter 2 Demand Filed: 08.11.1991
IPC:
G06N 3/04 (2006.01) ,G06N 5/02 (2006.01)
G PHYSICS
06
COMPUTING; CALCULATING; COUNTING
N
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3
Computer systems based on biological models
02
using neural network models
04
Architecture, e.g. interconnection topology
G PHYSICS
06
COMPUTING; CALCULATING; COUNTING
N
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
5
Computer systems utilizing knowledge based models
02
Knowledge representation
Applicants:
KABUSHIKI KAISHA KOMATSU SEISAKUSHO [JP/JP]; 3-6, Akasaka 2-chome Minato-ku Tokyo 107, JP (AllExceptUS)
YAMAGUCHI, Hiroyoshi [JP/JP]; JP (UsOnly)
MIMURA, Tatsuo [JP/JP]; JP (UsOnly)
Inventors:
YAMAGUCHI, Hiroyoshi; JP
MIMURA, Tatsuo; JP
Agent:
KIMURA, Takahisa; 6F Ginza Daisaku Building 11-2, Ginza 2-chome Chuo-ku Tokyo 104, JP
Priority Data:
2/11346227.04.1990JP
Title (EN) METHOD OF ADJUSTING AUTOMATICALLY RULE BASE OF EXPERT SYSTEM USING NEURAL NETWORK AND DEVICE THEREOF
(FR) PROCEDE DE REGLAGE AUTOMATIQUE DE LA BASE DE REGLES D'UN SYSTEME EXPERT A L'AIDE D'UN RESEAU NEURAL ET DISPOSITIF A CET EFFET
Abstract:
(EN) The rule base of an expert system which comprises a set of production rules including condition parts, action parts and certainties is resolved hierarchically into units of propositions each of which includes no or one logical operator. Then, every proposition, a characteristic list including the relations with propositions of upper hierarchies and certainty is generated automatically. Further, an initial pattern of a neural network in which the connective relations between propositions correspond to the ones between neurons is created automatically according to the generated characteristic list. Thereafter, learning of the created neural network is performed. Thereby, even one who has no expert knowledge can perform effectively the maintenance of the rule base of the expert system.
(FR) La base de règles d'un système expert qui est composé d'un ensemble de règles de production comprenant des parties d'état, des parties d'action et des certitudes est hiérarchiquement résolue en unités de propositions qui ne comprennent chacune aucun opérateur logique ou qu'un seul. Puis, chaque proposition, composée d'une liste catactéristique comprenant les relations avec des propositions des hiérarchies supérieures et la certitude, est automatiquement générée. En outre, une configuration initiale d'un réseau neural, dans lequel les relations de connexcion entre les propositions correspondent à celles ayant lieu entre des neurones, est automatiquement produite selon la liste caractéristique générée. Par la suite, l'apprentissage du réseau neural créé est effectué. Ainsi, même une personne qui n'aurait aucune expertise peut effectuer de manière efficace la maintenance de la base de règles du système expert.
Designated States: KR, US
European Patent Office (AT, BE, CH, DE, DK, ES, FR, GB, GR, IT, LU, NL, SE)
Publication Language: Japanese (JA)
Filing Language: Japanese (JA)
Also published as:
EP0536396