Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
HIGH-SPEED DIVIDED NEURAL NETWORK AND METHOD FOR CONSTRUCTING IT
Document Type and Number:
Japanese Patent JPH06203005
Kind Code:
A
Abstract:

PURPOSE: To provide a feed-forward control type divided neural network which operates at a high speed and a method for constructing the network.

CONSTITUTION: A divided network 10 is composed of a plurality of network layers 1-5 piled up upon another in the form of an upward pyramid. Network layers 1-5 are composed of pluralities of auxiliary networks SN12 and a plurality of nodes is arranged on each auxiliary network SN12, and then, the nodes are constituted in a layered neural network array in which the nodes are completely or partially interconnected. The input and output of each auxiliary network SN12 are 1-bit digital value limited to '0' or '1'. Each auxiliary network SN12 is independent from the other auxiliary networks SN12 in a given network layer 1-5 and, therefore, the network layers 1-5 are divided.


Inventors:
SUCHIIBUN EMU BURAIANTO
KENESU EICHI REEBENTAARU
Application Number:
JP23043193A
Publication Date:
July 22, 1994
Filing Date:
September 16, 1993
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
EASTMAN KODAK CO
International Classes:
G06F15/18; G06G7/60; G06N3/04; G06N3/063; G06N99/00; (IPC1-7): G06F15/18; G06G7/60
Attorney, Agent or Firm:
Kenji Yoshida (2 outside)