Information Technology Reference
In-Depth Information
Table 6.4
g-Means and Total Learning Time Using SMOTE, AL, and VIRTUAL
g-Means
(%)
Total Learning time (s)
Dataset
Batch
SMOTE
AL
VIRTUAL
SMOTE
AL
VIRTUAL
Reuters
acq
96.19 (3)
96.21 (2)
96.19 (3)
96.54 (1)
2271
146
203
corn
85.55 (4)
89.62 (2)
86.59 (3)
90.60 (1)
74
43
66
crude
88.34 (4)
91.21 (2)
88.35 (3)
91.74 (1)
238
113
129
earn
98.92 (3)
98.97 (1)
98.92 (3)
98.97 (1)
4082
121
163
grain
91.56 (4)
92.29 (2)
91.56 (4)
93.00 (1)
296
134
143
interest
78.45 (4)
83.96 (2)
78.45 (4)
84.75 (1)
192
153
178
money-fx
81.43 (3)
83.70 (2)
81.08 (4)
85.61 (1)
363
93
116
ship
75.66 (3)
78.55 (2)
74.92 (4)
81.34 (1)
88
75
76
trade
82.52 (3)
84.52 (2)
82.52 (3)
85.48 (1)
292
72
131
wheat
89.54 (3)
89.50 (4)
89.55 (2)
90.27 (1)
64
29
48
UCI
Abalone
100 (1)
100 (1)
100 (1)
100 (1)
18
4
6
Breast
98.33 (2)
97.52 (4)
98.33 (2)
98.84 (1)
4
1
1
letter
99.28 (3)
99.42 (2)
99.28 (3)
99.54 (1)
83
5
6
Satimage
83.57 (1)
82.61 (4)
82.76 (3)
82.92 (2)
219
18
17
“Batch” corresponds to the classical SVM learning in batch setting without resampling. The numbers in parentheses denote the rank of the
corresponding method in the dataset.
Search WWH ::
Custom Search