Skip to main content

Table 4 The classification results of multiple view features using MKL with one kind of kernel

From: Automatic plankton image classification combining multiple view features via multiple kernel learning

Datasets

C

Gaussian

Polynomial

Linear

  

R

1−P

F measure

R

1−P

F measure

R

1−P

F measure

WHOI

1

88.48%

11.35%

0.886

89.58%

10.21%

0.897

88.55%

11.18%

0.887

 

10

88.75%

11.04%

0.889

89.67%

10.16%

0.898

89.12%

10.65%

0.892

 

100

88.58%

11.2%

0.887

89.39%

10.44%

0.895

88.42%

11.41%

0.885

ZooScan

1

83.32%

11.74%

0.857

83.94%

12.61%

0.856

81.78%

15.53%

0.831

 

10

86.26%

10.01%

0.881

86.74%

11.76%

0.875

83.98%

15.28%

0.843

 

100

86.6%

9.92%

0.883

86.79%

11.63%

0.876

84.86%

19.13%

0.828

Kaggle

1

78.46%

17.41%

0.805

80.39%

16.76%

0.818

78.09%

19.24%

0.794

 

10

82.95%

16.42%

0.833

82.6%

15.62%

0.835

81.32%

17.66%

0.818

 

100

82.97%

16.84%

0.831

82.11%

15.82%

0.831

79.68%

19.1%

0.803

  1. The entries in boldface indicate the best classification results with the highest F measure