Pattern recognition apparatus using a neural network system
First Claim
1. A pattern recognition apparatus, comprising:
- (a) pattern input means for inputting pattern data and learning data;
(b) a neural network system including a plurality of neural networks, wherein;
(1) each of said plurality of neural networks receives said pattern data from said pattern input means;
(2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2);
(3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that;
(A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes;
(i) an output value A from said first unit (Uo1) equals a first value V1; and
(ii) an output value B from said second unit (Uo2) equals a second value V2; and
(B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes;
(i) said output value A from said first unit (Uo1) equals said second value V2 and(ii) said output value B from said second unit (Uo2) equals said first value V1; and
(c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks.
1 Assignment
0 Petitions
Accused Products
Abstract
A pattern recognition apparatus includes a pattern input unit inputting pattern data and learning data, and a neural network system including a plurality of neural networks, each of the plurality of neural networks being assigned a corresponding one of a plurality of identification classes and having only two output units of a first unit (Uo1) and a second unit (Uo2). Learning for each of the plurality of neural networks is performed by using the learning data. The image recognition apparatus also includes judgment unit judging which one of the identification classes the pattern data input from the image reading unit belongs to on the basis of output values A and B from the two output units (Uo1) and (Uo2) of all neural networks.
93 Citations
11 Claims
-
1. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks. - View Dependent Claims (2, 10, 11)
-
-
3. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks, wherein said judgment means has means for judging whether said pattern belongs to one of said identification classes, a corresponding one of said plurality of neural networks satisfying the condition of output values A and B being A>
B.
-
-
4. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks, wherein said judgement means has; (1) candidate selection means for selecting at least one of said plurality of said neural networks satisfying the condition of output values A and B being A>
B as a candidate neural network; and(2) means for judging that said pattern data belongs to one of said identification classes corresponding to said candidate neural network when one candidate neural network is selected by said candidate selection means.
-
-
5. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks, wherein said judgment means has; (1) candidate selection means for selecting at least one of said plurality of neural networks satisfying the condition of output values A and B being A>
B as a candidate neural network, and(2) means for judging that said pattern belongs to one of said identification classes corresponding to one of the plurality of neural networks for which the output value A is the greatest when two or more candidate neural networks are selected by said candidate selection means.
-
-
6. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment mean for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks, wherein said judgment means has; (1) candidate selection means for selecting at least one of said plurality of neural networks satisfying the condition of output values A and B being A>
B as a candidate neural network, and(2) output means for outputting rejection information expressing that said pattern data does not belong to any identification class, when a candidate neural network is not selected by said candidate selection means.
-
-
7. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks, wherein said judgment means has; (1) candidate selection means for selecting at least one of said plurality of neural networks satisfying the condition of output values A and B being A>
B as a candidate neural network, and(2) output means for outputting rejection information expressing that said pattern data does not belong to any identification class, when two or more candidate neural networks are selected by said candidate selection means.
-
-
8. A pattern recognition apparatus, comprising:
-
(a) pattern input means for inputting pattern data and learning data; (b) a neural network system including a plurality of neural networks, wherein; (1) each of said plurality of neural networks receives said pattern data from said pattern input means; (2) each of said plurality of neural networks is assigned a corresponding one of a plurality of identification classes and has only two output units consisting of a first unit (Uo1) and a second unit (Uo2); (3) learning for each of said plurality of neural networks is performed by using said learning data belonging to all of said identification classes so that; (A) when each of said neural networks receives said learning data belonging to said corresponding one of the identification classes; (i) an output value A from said first unit (Uo1) equals a first value V1; and (ii) an output value B from said second unit (Uo2) equals a second value V2; and (B) when each of said neural networks receives learning data belonging to one of said identification classes other than said corresponding one of the identification classes; (i) said output value A from said first unit (Uo1) equals said second value V2 and (ii) said output value B from said second unit (Uo2) equals said first value V1; and (c) judgment means for judging which one of the identification classes said pattern data from said pattern input means belongs to, on the basis of output values A and B from said two output units (Uo1) and (Uo2) of all neural networks, wherein said judgment means has; (1) candidate selection means for selecting at least one of said plurality of neural networks satisfying the condition of output values A and B being A>
B as a candidate neural network, and(2) another recognition system that judges that said pattern data is related to one of identification classes corresponding to each candidate neural network, on the basis of said pattern data input from said pattern input means when said two or more candidates have been selected by said candidate selection means. - View Dependent Claims (9)
-
Specification