CONSTITUTING METHOD FOR NEURAL CIRCUIT NETWORK AND NEURAL CIRCUIT NETWORK

PURPOSE:To integrate former knowledge with the small amount of computation by determining the load of a branch between neural elements by using an intermediate pattern, which is prepared by using a partial neural circuit network to correspond to the projecting area of an input signal space and conve...

Full description

Saved in:
Bibliographic Details
Main Author KAWAHARA HIDENORI
Format Patent
LanguageEnglish
Published 15.03.1990
Subjects
ANALOGUE COMPUTERS
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:PURPOSE:To integrate former knowledge with the small amount of computation by determining the load of a branch between neural elements by using an intermediate pattern, which is prepared by using a partial neural circuit network to correspond to the projecting area of an input signal space and converting an input pattern for learning, and a value, for which a desired output is inverted to a value in the input side of an output element. CONSTITUTION:By constituting the partial neural circuit network to express the projecting area in an input pattern space to the layer-shaped neural circuit network, loads 10 and 12 of the branch in the neural circuit network and the sensivity of respective neural elements and a threshold value are directly determined by using the example of coupling among knowledge concerning the input pattern, which goes to be a subject, various inputs and an desired output to correspond to the respective inputs. Accordingly, the obtained knowledge concerning the subject can be projected to determination for the branch weight of the neural circuit network, the sensitivity of the element and the threshold value.
Bibliography:Application Number: JP19880228189