novel feature recognition neural network
Table 1 represents the results of novel feature recognition neural network. The data is the number of firing neurons. The largest data which we have marked classified properly to the related patterns, so the recognition ratio is 100%. It proved that this method was able to correct mistakes made by some of the subpatterns classifiers, while maintaining the correct classifications for the object where there were no confusion in the algorithm outputs. It should be emphasized that Fig. 6(c), Fig. 6(f) and Fig. 6(i) are noisy but still recognized correctly.
 
testing specimen
training
sample
Fig. 5(a)
training sample
Fig. 5(b)
training sample
Fig. 5(c)
training sample
Fig. 5(d)
training sample
Fig. 5(e)
training sample
Fig. 5(f)
training sample
Fig. 5(g)
training sample
Fig. 5(h)
training sample
Fig. 5(i)
training sample
Fig. 5(j)
Fig. 6(a)
13
9
18
23
17
17
18
17
15
17
Fig. 6(b)
12
9
17
21
17
17
18
17
15
17
Fig. 6(c)
7
5
10
16
13
14
15
11
11
11
Fig. 6(d)
11
8
14
19
17
17
21
17
16
17
Fig. 6(e)
12
8
14
19
17
17
22
17
16
17
Fig. 6(f)
6
8
7
7
8
11
15
14
13
12
Fig. 6(g)
12
9
12
16
16
14
17
20
21
20
Fig. 6(h)
12
8
12
15
15
14
15
20
20
19
Fig. 6(i)
11
5
11
11
10
11
12
14
22
13
 Table1 The results of the operation of neural network in training samples and testing specimen.
 
nextindexback