Information Technology Reference
In-Depth Information
h e main disadvantages of the V-detector approach are as follows:
h e number of detectors is large compared to similar methods
Larger detectors do not necessarily contribute more to the coverage because
they may overlap other detectors
NS Algorithm 7: Generate a negative detector set of V-detectors
Input
S
: set of self samples;
r
s
: self radius
p
: target coverage
ψ
: signifi cant level for hypothesis testing
T
max
: maximum number of detectors
Output:
A set of V-Detectors
←
1
n
max{5/p, 5/(1 − p)} {sample size required for hypothesis testing}
∅
3
repeat
4
t
←
2
D
←
0{counter of “already covered” candidates}
5
N
←
0{counter of valid candidates}
6
C
← ∅
{the collection of valid candidates}
7
x
←
random sample from [0, 1]
n
∞
9
for all
s
i
in
S
do
{censored by self samples}
10
d
8
r
←
←
Euclidean distance between
s
i
and
x
←
11
if
d
r
S
then
12
go to 7
13
else
←
14
r
min {
r
,
d
}
15
end if
16
end for
17
N
1
18
for all
d
i
in
D
←
N
+
=
{
d
i
,
i
=
1, 2, ….}
do
{censored by existing
detectors}
←
19
d
d
Euclidean distance between
d
i
and
x
<
20
if
d
d
r
(
d
i
)
then
{
r
(
d
i
) is the radius of detector
d
i
}
←
+
21
t
t
1
t
np
z
→
22
1
p
np
(
1
p
)
>
23
if z
z then
24
return D
Search WWH ::
Custom Search