Information Technology Reference
In-Depth Information
1
0.9
0.8
REA
Normal
SMOTE
0.7
SERA
0.6
0.5
0.4
UB
0.3
0.2
10
20
30
40
50
60
70
80
90
100
(a) OA of algorithms in comparison
for SHP dataset under setup 3
1
REA
0.95
0.9
SERA
0.85
Normal
UB
0.8
0.75
0.7
0.65
SMOTE
10
20
30
40
50
60
70
80
90
100
(b) AUC of algorithms in comparison
for SHP dataset under setup 3
Figure 7.12
OA and AUROC for SHP dataset under
setup 3
.
be guaranteed that
H
always maintains the “freshest” subset of the generated
hypotheses in memory.
Figure 7.13 shows the AUROC performance of REA on learning from SHP
datasets with 100 data chunks when the size of
H
, that is,
|
H
|
,isequalto
∞
, 60, 40, and 20, respectively. One can see that REA initially improves its
performance in terms of AUROC when
|
H
|
shrinks. However, when
|
H
|=
20,
the performance of REA deteriorates, which is worse than the case when
|
H
|=
∞
. On the basis of these observations, one can conclude that there exists a trade-
off between
|
H
|
and the REA's performance. A heuristic would be to set
|
H
|
Search WWH ::
Custom Search