Database Reference
In-Depth Information
where
χ
[
a,b
]
=
0 f
t<a∨ t>b,
1 f
a ≤ t ≤ b,
and
η
and
ε
(
t
) are obtained from a standard normal distribution
N
(0
,
1),
a
b − a
is another integer obtained from another uniform distribution in [32, 96].
The examples are generated evaluating those functions for
is an integer obtained from a uniform distribution in [16, 32] and
t
=1,2
,...,
128.
Figure 3 shows some examples of this data set.
The obtained results are shown in Figure 4(A). Two kinds of graphs
are used. The first one shows the error rate as a function of the number of
literals. The second also shows the error rate, but in this case using early
classification and the maximum number of considered literals. The error
rate is shown as a function of the percentage of the length of the series.
In each graph, two results are plotted, one using point based literals and
another using interval based literals. The results are also shown in Table 5.
For early classification, the results obtained with the 60% of the series
length are very close to those obtained with the full series.
(A) ORIGINAL DATA SET
error/series length percentage
error/number of literals
70
12
points
intervals
points
intervals
60
10
50
8
40
6
30
4
20
2
10
0
0
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
(B) VARIABLE LENGTH DATA SET
error/series length percentage
error/number of literals
70
12
points
intervals
points
intervals
60
10
50
8
40
6
30
4
20
2
10
0
0
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
Fig. 4.
Error graphs for the
CBF
data set.
