Title:
NEURAL NETWORK LEARNING DEVICE
Document Type and Number:
Japanese Patent JP3031075
Kind Code:
B2
Abstract:
PURPOSE: To provide a high-speed neural network learning device by detecting and deleting unnecessary cells in a neural network part after learning a neural network without restrictions.
CONSTITUTION: A neural network part 11 consists of links and cells. An example storage part 12 holds examples. A detecting part 13 detects unnecessary cells in the neural network part 11. A learning control part 14 controls learning of the neural network part 11 and restricts erasing of cells.
Inventors:
Minoru Asokawa
Application Number:
JP22257692A
Publication Date:
April 10, 2000
Filing Date:
August 21, 1992
Export Citation:
Assignee:
NEC
International Classes:
G06F15/18; G06G7/60; G06N99/00; (IPC1-7): G06F15/18; G06G7/60
Domestic Patent References:
| JP383157A | ||||
| JP4290156A |
Attorney, Agent or Firm:
Naoki Kyomoto (2 outside)