Title:
NEURAL NETWORK TYPE ADDITIONAL LEARNING DEVICE
Document Type and Number:
Japanese Patent JP3435714
Kind Code:
B2
Abstract:
PURPOSE: To provide the neural network type additional learning device which can prevent deterioration in the learning performance by discriminating the range of an input signal from the input range for learning by a neural net.
CONSTITUTION: In an air conditioner subjected to automatic control, in accordance with an input signal from each environment sensor 30, in the case of an input signal within a prescribed range, it is controlled by a learnable control part 121, and in the case of being outside of the prescribed range, it is controlled by a fixed control part 122. When a vehicle crewman changes an air-- conditioning state with a setting switch in a panel P, a changed value, and an input signal in this case are stored as an additional control pattern in a storage part 14. In accordance with the pattern corresponding to the input signal within the prescribed range in this additional control pattern, the learnable control part 121 executes additional learning.
More Like This:
Inventors:
Tomohisa Yoshimi
Takamasa Kawai
Takamasa Kawai
Application Number:
JP34414092A
Publication Date:
August 11, 2003
Filing Date:
December 24, 1992
Export Citation:
Assignee:
株式会社デンソー
International Classes:
G06F15/18; F24F11/02; G06G7/60; G06N3/00; G06N3/08; G06N99/00; (IPC1-7): G06N3/08; F24F11/02; G06N3/00
Domestic Patent References:
| JP6161982A | ||||
| JP4195667A | ||||
| JP558197A |
Attorney, Agent or Firm:
Hirohiko Usui
Previous Patent: LEARNING PATTERN SELECTION TYPE ADDITIONAL LEARNING DEVICE
Next Patent: INDEX SIGNAL GENERATOR
Next Patent: INDEX SIGNAL GENERATOR