Approximate reasoning apparatus

• US 5,651,100 A
• Filed: 06/12/1991
• Issued: 07/22/1997
• Est. Priority Date: 06/28/1990
• Status: Expired due to Fees

• Current Assignee: Omron Corporation
• Original Assignee: Omron Corporation
• Inventors: Nakajima, Hiroshi, Hayashi, Motoji
• Primary Examiner(s): Davis, George B.
• Application Number:
  US07/714,020
• Time in Patent Office: 2,232 Days
• Field of Search: 395/61, 395/62, 395/68, 395/3, 395/50
• US Class Current: 706/52
• CPC Class Codes: G06N 5/048 Fuzzy inferencing

Abstract

A threshold value is set in advance for each of a number of conclusions. Possibility values of conclusions obtained from an approximate reasoning apparatus are compared with the corresponding threshold values, and a conclusion whose possibility value is equal to or greater than its threshold value is outputted (displayed). When information indicative of an error is entered with regard to a conclusion whose possibility value is equal to or greater than the corresponding threshold value but which actually does not hold true, or a conclusion whose possibility value is less than the corresponding threshold value but which actually does hold true, the threshold value concerning the conclusion that has produced the error is revised dynamically. The weights of conclusions are set. By multiplying the possibility value of a conclusion from an approximate reasoning apparatus by the weight set for this conclusion, a score is calculated for every conclusion. If the order of the obtained scored differs from an order of priority devised by a human being, weight is revised in such a manner that the order of the scores will coincide with the order of priority.

Claims

1. An approximate reasoning apparatus comprising:

an input device for inputting a decision regarding the correctness of a result of reasoning performed based upon knowledge from a plurality of experts;

a selector for selecting which knowledge of an expert among the plurality thereof is to have its weight altered, based upon the decision regarding correctness inputted by said input device;

an arithmetic unit for calculating a new weight for the knowledge of the expert selected by said selector;

a memory unit for storing the new weight for the knowledge of the expert calculated by said arithmetic unit; and

a knowledge re-synthesizing device for automatically re-synthesizing knowledge based upon the new weight by combining the knowledge of the expert with said new weight calculated by said arithmetic unit.