Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
FUZZY NEURON FOR RECOGNIZING PATTERN
Document Type and Number:
Japanese Patent JP3243313
Kind Code:
B2
Abstract:

PURPOSE: To expand the expressed contents of a reference pattern described by a membership function and to recognize a more complicated pattern or a wider purpose.
CONSTITUTION: In a fuzzy neuron recognizing an input pattern by comparing/ collating a distribution signal obtained from an input pattern by a feature segmenting line with a previously set membership function, passage type and inhibition type membership functions regarding a positive distribution signal as a pattern in a part of the input signal and an inverted distribution signal as a pattern in the other part respectively set. Since the inverted distribution signal also can be compared with the membership function in addition to the comparison of the positive ditribution signal, the fuzzy neuron capable of setting up passage type and inhibition type membership functions in any of the positive and inverted distribution signals respectively regarded as patterns in a part of the input signal and the other can be attained.


Inventors:
Fumiaki Mooka
Conquest of the king
Application Number:
JP1418593A
Publication Date:
January 07, 2002
Filing Date:
January 29, 1993
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
ROHM Co., Ltd.
International Classes:
G06F9/44; G06F15/18; G06G7/12; G06G7/60; G06K9/46; G06K9/66; G06K9/68; G06N3/04; G06N7/02; G06N99/00; G06T7/00; (IPC1-7): G06T7/00; G06K9/46; G06K9/68
Domestic Patent References:
JP367303A
JP3110633A
JP6194133A
JP3219385A
JP2310782A
JP6231259A
Attorney, Agent or Firm:
Masashi Kobori