Information Technology Reference
In-Depth Information
Tabl e 6 . 3 Best results of MEE-BS and MEE-VLR algorithms with the time per
epoch,
T
,in
10
−
3
sec.
Dataset "Ionosphere" (
n
=351,
c
=2)
Error (Std) L n
h
Epochs
T
T
VLR
/
T
BS
MEE-VLR
12.06(1.11)
-
12
40
16.7
-
MEE-BS
12.27(1.23)
4
8
100
6.4
2.6
MEE-BS
12.00(1.09)
8
8
100
4.8
3.5
Dataset "Olive" (
n
=572,
c
=9)
Error (Std) L n
h
Epochs
T
T
VLR
/
T
BS
MEE-VLR
5.04(0.53)
-
25
200
77.7
-
MEE-BS
5.17(0.51)
4
30
140
17.6
4.4
MEE-BS
5.24(0.70)
8
20
180
12.8
6.1
Dataset "Wdbc" (
n
=569,
c
=2)
Error (Std) L n
h
Epochs
T
T
VLR
/
T
BS
MEE-VLR
2.33(0.37)
-
4
40
38.7
-
MEE-BS
2.31(0.35)
5
10
60
13.6
2.8
MEE-BS
2.35(0.48)
8
10
40
9.6
4.0
Dataset "Wine" (
n
=178,
c
=3)
Error (Std) L n
h
Epochs
T
T
VLR
/
T
BS
MEE-VLR
1.83(0.83)
-
14
40
5.8
-
MEE-BS
1.88(0.80)
4
16
60
3.2
1.8
MEE-BS
1.88(0.86)
8
16
60
2.5
2.3
100
100
90
MEE −BS
MEE −BS (SA)
MEE −BS (RBP)
90
90
80
80
80
70
70
70
60
60
60
50
50
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
0
50
100
150
200
250
300
0
50
100
150
200
250
300
0
50
100
150
200
250
300
Epochs
Fig. 6.9 Training curves for MEE-BS, MEE-BS(SA) and MEE-BS(RBP) algo-
rithms in one experiment with a real-world dataset.
