Skip to main content

Table 7 Amount of improvements in the MA

From: Integrating artificial neural network and classical methods for unsupervised classification of optical remote sensing data

Class number

UMCS@-K-means (%)

UMCS@-Kohonen (%)

 

Improvement of UMCS@ over each of K-means and Kohonen

1

−0.98

+2.53

 

2

−1.53

+6.59

 

3

+2.39

+7.90

 

4

−3.08

+3.04

 

5

+0.97

+0.48

 

6

+1.33

+0.59

 

7

+6.57

+1.11

 

8

+0.49

+0.14

 

Average

+0.77

+2.79

 

Improvement of UMCS# over each of K-medians and Kohonen

 

1

+0.15

+4.01

 

2

−2.70

+3.23

 

3

−0.44

+1.82

 

4

+0.68

+6.44

 

5

+1.94

+0.48

 

6

+0.50

+3.90

 

7

+4.83

+1.39

 

8

+1.67

+2.08

 

Average

+0.829

+2.92

 

Improvement of UMCS* over each of K-means, K-medians and Kohonen

Class number

UMCS*-K-means (%)

UMCS*-K-medians (%)

UMCS*-Kohonen (%)

1

+2.64

+2.33

+6.19

2

+1.26

+2.39

+8.32

3

+4.55

+7.80

+10.06

4

+3.34

+4.70

+10.46

5

+2.96

+3.93

+2.47

6

+8.51

+4.37

+8.77

7

+8.36

+6.34

+2.90

8

+2.53

+1.77

+2.18

Average

+4.27

+3.70

+6.41

  1. A: Improvement of UMCS@ over each of K-means and Kohonen. B: Improvement of UMCS# over each of K-medians and Kohonen. C: Improvement of UMCS* over each of K-means, K-medians and Kohonen.