I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
24
,
No
.
2
,
N
o
v
em
b
e
r
2
0
2
1
,
p
p
.
1
1
4
1
~
1
1
4
8
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
24
.i
2
.
p
p
1
1
4
1
-
1
1
4
8
1141
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
A simple,
ef
fectiv
e distance
and de
nsity
bas
ed outlie
r det
e
ction
a
lg
o
rithm
Sa
j
idh
a
S.
A.
1
,
Uda
i A
g
a
rwa
l
2
,
P
rut
hv
ira
j
R.
P
.
3
,
Sp
a
rsh
Ag
a
rwa
l
4
,
Nis
ha
V.
M
.
5
,
Am
it
K
um
a
r
T
y
a
g
i
6
1,
2,
3,
4
,
5,
6
S
c
h
o
o
l
o
f
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
in
e
e
ri
n
g
,
Ve
ll
o
re
In
st
it
u
te o
f
Tec
h
n
o
lo
g
y
,
Ch
e
n
n
a
i,
I
n
d
i
a
6
Ce
n
tre fo
r
Ad
v
a
n
c
e
d
Da
ta S
c
ien
c
e
,
Ve
ll
o
re
In
stit
u
te o
f
Tec
h
n
o
lo
g
y
,
Ch
e
n
n
a
i,
In
d
ia
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
11
,
2
0
2
0
R
ev
is
ed
Au
g
19
,
2
0
2
1
Acc
ep
ted
Sep
9
,
2
0
2
1
Ou
tl
iers
a
re
e
c
c
e
n
tri
c
d
a
ta
p
o
in
t
s
with
a
n
o
m
a
lo
u
s
n
a
tu
re
.
Clu
ste
rin
g
with
o
u
tl
iers
h
a
s
re
c
e
iv
e
d
a
l
o
t
o
f
a
tt
e
n
ti
o
n
i
n
t
h
e
d
a
ta
p
ro
c
e
ss
in
g
c
o
m
m
u
n
it
y
.
Bu
t
,
th
e
y
i
n
o
rd
i
n
a
tely
a
ffe
c
t
t
h
e
q
u
a
li
t
y
o
f
t
h
e
re
su
lt
s
o
b
tain
e
d
in
c
a
se
o
f
p
o
p
u
lar
c
lu
ste
rin
g
a
l
g
o
rit
h
m
s
d
u
ri
n
g
t
h
e
p
ro
c
e
ss
o
f
fin
d
in
g
a
n
o
p
ti
m
a
l
so
lu
ti
o
n
.
In
th
is
wo
rk
,
we
p
ro
p
o
se
a
n
o
v
e
l
m
e
th
o
d
t
o
c
las
sify
th
e
d
a
ta
p
o
in
ts
wit
h
g
ro
u
p
i
n
g
c
h
a
ra
c
teristics
a
s
e
it
h
e
r
a
n
o
u
t
li
e
r
o
r
n
o
t.
We
u
se
b
o
th
d
is
tan
c
e
a
n
d
d
e
n
sity
o
f
a
p
a
rti
c
u
lar
d
a
ta
p
o
in
t
with
re
sp
e
c
t
to
t
h
e
re
st
o
f
t
h
e
d
a
ta
p
o
i
n
ts
fo
r
th
is
p
ro
c
e
ss
.
Dista
n
c
e
s
a
re
u
se
d
t
o
fi
n
d
th
e
p
o
i
n
ts
a
t
th
e
e
x
trem
it
ies
wh
il
e
t
h
e
d
e
n
siti
e
s
a
re
u
se
d
to
i
d
e
n
ti
fy
t
h
e
d
a
ta
p
o
i
n
ts
a
t
th
e
sp
a
rse
st
sp
a
c
e
s.
F
u
rt
h
e
r,
e
v
e
ry
d
a
ta
m
o
d
e
l
h
a
s
to
tak
e
i
n
to
a
c
c
o
u
n
t
th
e
a
sp
e
c
t
o
f
g
e
n
e
ra
l
iz
a
ti
o
n
i
n
o
rd
e
r
to
w
o
rk
ro
b
u
st
ly
e
v
e
n
i
n
o
u
t
o
f
th
e
b
o
x
sit
u
a
ti
o
n
s.
He
n
c
e
,
o
u
r
a
p
p
ro
a
c
h
p
ro
v
id
e
s
a
g
e
n
e
ra
li
z
a
ti
o
n
a
sp
e
c
t
t
o
t
h
e
m
o
d
e
l.
Th
e
a
c
c
u
ra
c
y
o
f
th
e
p
ro
p
o
se
d
wo
rk
is
m
e
a
su
re
d
u
sin
g
a
re
a
u
n
d
e
r
c
u
rv
e
(
AUC
)
wa
s
f
o
u
n
d
th
e
h
ig
h
e
st
fo
r
c
a
rd
io
to
d
a
ta
se
t
-
AU
C
v
a
lu
e
-
0
.
9
0
a
n
d
se
c
o
n
d
h
i
g
h
e
st
AU
C
v
a
lu
e
wa
s
o
b
tai
n
e
d
f
o
r
S
p
a
m
b
a
se
d
a
ta
se
t
-
0
.
5
2
a
n
d
se
v
e
ra
l
o
t
h
e
r
d
a
tas
e
ts
a
re
u
se
d
t
o
d
e
m
o
n
stra
te t
h
e
u
sa
g
e
o
f
th
e
m
o
d
e
l
p
ro
p
o
se
d
.
K
ey
w
o
r
d
s
:
An
ti
-
n
eig
h
b
o
u
r
s
Ar
ea
u
n
d
er
cu
r
v
e
Den
s
ity
Dis
tan
ce
s
Ou
tlier
s
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Sajid
h
a
S.
A
.
Sch
o
o
l o
f
C
o
m
p
u
ter
Scien
ce
a
n
d
E
n
g
in
ee
r
in
g
Vello
r
e
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
Van
d
alu
r
-
Kela
m
b
a
k
k
a
m
R
o
ad
,
C
h
en
n
ai
-
6
0
0
1
2
7
,
I
n
d
ia
E
m
ail:
am
itk
r
ty
ag
i0
2
5
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
An
o
u
tlier
is
an
o
b
s
er
v
atio
n
t
h
at
is
s
tr
ik
in
g
ly
f
ar
f
r
o
m
th
e
clu
s
ter
ce
n
ter
s
.
I
t
is
an
ec
ce
n
tr
ic
v
alu
e
r
elativ
e
to
th
e
d
ata.
No
r
m
ally
,
co
m
p
u
te
d
v
alu
es
s
u
ch
as
av
er
ag
e
o
r
least
s
q
u
ar
e
lin
es
ca
n
b
e
d
r
am
atica
lly
af
f
ec
ted
b
y
s
u
c
h
v
al
u
es
[
1
]
.
Hen
ce
,
m
eth
o
d
s
to
d
etec
t
o
u
tlier
s
an
d
to
m
o
d
er
ate
th
eir
ef
f
ec
ts
ar
e
n
ee
d
ed
.
T
h
e
im
p
o
r
ta
n
ce
o
f
d
en
s
ity
an
d
d
is
tan
ce
o
f
d
ata
p
o
in
ts
wh
ile
id
en
tify
in
g
th
e
in
itial seed
p
o
in
ts
f
o
r
-
m
ea
n
s
f
o
r
n
u
m
er
ical
d
ata,
-
m
o
d
es
f
o
r
ca
teg
o
r
ical
d
ata
a
n
d
m
i
x
ed
d
atase
ts
u
s
in
g
m
o
d
if
ied
-
m
ea
n
s
alg
o
r
ith
m
,
is
elu
cid
ated
in
th
e
wo
r
k
[
2
]
-
[
4
]
in
wh
ich
th
e
in
itial
s
ee
d
p
o
i
n
ts
wer
e
ef
f
ec
tiv
ely
id
en
tifie
d
.
On
e
o
f
th
e
m
ajo
r
d
r
awb
ac
k
s
o
f
t
h
e
p
ar
titi
o
n
b
ased
clu
s
ter
in
g
alg
o
r
ith
m
is
t
h
at
th
ey
ca
n
n
o
t
d
etec
t
th
e
p
r
esen
ce
o
f
o
u
tlier
s
.
Fo
cu
s
in
g
o
n
r
em
o
v
in
g
th
e
o
u
tlier
s
f
r
o
m
th
e
d
ataset
f
u
r
th
er
o
p
tim
izin
g
th
e
clu
s
ter
in
g
m
e
th
o
d
o
lo
g
ies
to
g
iv
e
ac
cu
r
ate
r
esu
lts
.
T
h
e
a
u
th
o
r
s
o
f
[
1
]
s
h
o
we
d
h
o
w
th
e
b
iv
ar
iate
d
ata
r
ep
r
esen
ted
in
th
e
f
o
r
m
o
f
b
o
x
p
lo
ts
,
w
h
en
g
en
er
alize
d
,
g
iv
es
o
u
t
th
e
b
a
g
p
lo
ts
wh
ich
in
t
u
r
n
ca
n
b
e
u
s
ed
to
s
k
im
o
u
t
t
h
e
o
u
tlier
s
.
T
h
is
asp
ec
t
o
f
th
e
b
in
ar
y
p
r
o
p
er
t
y
was
f
u
r
th
er
m
en
tio
n
ed
in
[
5
]
as
elu
cid
at
ed
.
I
n
[
6
]
,
th
e
o
u
tlier
d
etec
tio
n
u
s
in
g
in
-
d
eg
r
ee
n
u
m
b
er
(
ODI
N
)
al
g
o
r
ith
m
,
-
m
ea
n
s
an
d
‘
Ou
tly
in
g
n
ess
f
ac
to
r
’
h
av
e
b
ee
n
u
s
ed
to
r
em
o
v
e
o
n
e
o
r
m
o
r
e
d
ata
p
o
in
ts
(
o
u
tlier
s
)
an
d
g
et
n
o
n
-
o
v
er
lap
p
in
g
clu
s
ter
s
.
I
n
[
7
]
p
r
o
p
o
s
ed
th
e
ap
p
r
o
ac
h
o
f
f
in
d
in
g
th
e
cl
u
s
ter
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
24
,
No
.
2
,
No
v
em
b
er
2
0
2
1
:
1
1
4
1
-
1
1
4
8
1142
f
o
llo
wed
b
y
th
e
l
o
ca
l
o
u
tlie
r
s
an
d
f
in
ally
lo
o
k
in
g
f
o
r
t
h
e
g
lo
b
al
o
u
tlier
s
.
I
n
[
8
]
,
l
o
ca
l
k
er
n
el
d
en
s
ity
e
s
tim
atio
n
is
d
o
n
e.
In
[
9
]
ex
p
lain
th
e
Gau
s
s
ian
u
n
if
o
r
m
m
i
x
tu
r
e
m
o
d
el
(
Gu
MM
)
m
et
h
o
d
o
lo
g
y
wh
ich
f
its
a
clu
s
ter
to
id
en
tify
o
u
tlier
s
th
at
d
o
n
o
t f
all
in
to
a
n
y
o
f
th
e
n
o
r
m
al
clu
s
ter
s
.
An
o
th
er
wo
r
k
in
[
1
0
]
p
r
o
p
o
s
ed
an
g
le
-
b
ased
o
u
tlier
d
etec
ti
o
n
(
AB
OD
)
alg
o
r
ith
m
ass
ig
n
s
an
an
g
le
-
b
ased
o
u
tlier
f
ac
to
r
(
AB
OF
)
to
ea
ch
p
o
in
t
in
th
e
d
ata
b
ase
an
d
r
etu
r
n
s
a
s
o
r
ted
lis
t
o
f
p
o
in
ts
b
ased
o
n
AB
OF
v
alu
es.
Fas
tAB
OD
[
1
0
]
ap
p
r
o
x
im
ates
AB
OF
b
ased
o
n
a
s
am
p
le
o
f
th
e
d
atab
ase.
T
h
e
wo
r
k
in
[
1
1
]
,
th
e
d
i
s
t
a
n
c
e
s
o
f
a
l
l
t
h
e
d
a
t
a
p
o
i
n
t
s
f
r
o
m
t
h
e
i
r
c
l
o
s
e
s
t
n
e
a
r
e
s
t
n
e
i
g
h
b
o
r
s
a
r
e
m
e
a
s
u
r
e
d
.
I
f
t
h
e
s
u
m
o
f
a
l
l
t
h
e
d
i
s
t
a
n
c
e
s
i
s
g
r
e
a
t
e
r
t
h
a
n
t
h
e
t
h
r
e
s
h
o
l
d
v
a
l
u
e
,
t
h
e
n
t
h
e
p
o
i
n
t
i
s
c
o
n
s
i
d
e
r
e
d
a
n
o
u
t
l
i
e
r
.
I
n
t
h
e
w
o
r
k
d
i
s
c
u
s
s
e
d
i
n
[
1
2
]
u
s
e
a
d
en
s
ity
-
b
ased
ap
p
r
o
ac
h
f
o
r
f
in
d
in
g
th
e
o
u
tlier
s
in
th
e
lo
ca
l
o
u
tlier
p
r
o
b
ab
ilit
y
(
L
o
OP
)
alg
o
r
ith
m
.
T
h
e
wo
r
k
in
[
1
3
]
f
o
cu
s
es
o
n
f
in
d
in
g
o
u
tlier
s
in
h
ig
h
d
im
e
n
s
io
n
al
d
ata,
u
s
in
g
d
if
f
er
e
n
t
s
u
b
s
p
ac
es
o
f
th
e
o
r
ig
in
al
s
p
ac
e
(
a
s
u
b
s
et
o
f
co
m
p
lete
f
ea
tu
r
e
s
et)
.
T
h
e
wo
r
k
in
[
1
4
]
p
r
o
p
o
s
es
a
n
o
v
el
s
u
b
s
p
ac
e
s
ea
r
ch
m
eth
o
d
t
h
at
s
elec
ts
h
ig
h
co
n
tr
ast
s
u
b
s
p
ac
es
(
HiC
S
)
f
o
r
d
en
s
ity
-
b
ased
o
u
tlier
r
an
k
in
g
.
T
h
e
o
u
tlier
s
co
r
es
ar
e
b
ased
o
n
lo
ca
l
o
u
tlier
f
ac
t
o
r
(
L
OF
)
.
T
h
e
wo
r
k
in
[
1
5
]
in
tr
o
d
u
ce
d
th
e
co
r
r
elatio
n
o
u
tlier
p
r
o
b
a
b
ilit
ies
(
C
O
P
)
m
e
t
h
o
d
t
h
a
t
g
e
n
e
r
a
l
i
z
e
s
t
h
i
s
i
d
e
a
b
y
l
o
o
k
i
n
g
f
o
r
t
h
e
a
r
b
i
t
r
a
r
i
l
y
o
r
i
e
n
t
e
d
s
u
b
s
p
a
c
e
s
o
f
h
i
g
h
e
s
t
v
a
r
i
a
n
c
e
a
n
d
f
u
r
t
h
e
r
p
r
o
v
i
d
e
s
a
n
e
r
r
o
r
v
e
c
t
o
r
f
o
r
e
a
c
h
i
d
e
n
t
i
f
i
e
d
o
u
t
l
i
e
r
a
s
a
f
o
r
m
o
f
e
x
p
l
a
n
a
t
i
o
n
.
I
n
t
h
e
w
o
r
k
s
d
i
s
c
u
s
s
e
d
i
n
[
1
6
]
,
[
1
7
]
s
tu
d
ie
d
d
is
tan
ce
b
ased
tech
n
i
q
u
es u
s
ed
f
o
r
class
im
b
alan
ce
d
ata.
I
n
th
e
wo
r
k
[
1
8
]
co
m
p
ar
es
te
n
d
if
f
er
e
n
t
m
eth
o
d
o
lo
g
ies
an
d
th
eir
p
er
f
o
r
m
an
ce
s
o
v
er
n
i
n
e
r
ea
l
-
tim
e
d
atasets
.
T
h
e
m
eth
o
d
o
lo
g
ies
lik
e
L
OF,
ODI
N
an
d
NN.
ar
e
co
m
p
ar
e
d
.
I
n
[
1
9
]
th
e
au
th
o
r
s
c
o
m
b
in
e
th
e
s
tatis
t
ical
m
eth
o
d
s
o
f
m
ea
n
an
d
s
tan
d
ar
d
d
e
v
iatio
n
(
MS
D
)
with
th
e
K
-
m
ea
n
s
clu
s
ter
in
g
alg
o
r
ith
m
wh
ile
d
etec
tin
g
th
e
o
u
tlier
s
.
I
n
[
2
0
]
u
s
es
clu
s
ter
b
o
u
n
d
to
f
in
d
t
h
e
s
u
s
p
ec
ted
o
u
tlier
in
s
tan
ce
.
I
f
th
e
av
er
ag
e
d
is
tan
c
e
o
f
th
e
s
u
s
p
ec
ted
o
u
tlier
is
g
r
ea
ter
th
an
th
e
av
e
r
ag
e
d
is
ta
n
ce
o
f
th
e
n
ei
g
h
b
o
r
h
o
o
d
p
o
i
n
ts
th
en
it
ca
n
b
e
co
n
s
id
er
ed
as a
n
elig
i
b
le
o
u
tli
er
.
I
n
[
2
1
]
th
e
au
th
o
r
s
h
av
e
e
n
s
em
b
le
3
f
am
o
u
s
clu
s
ter
in
g
alg
o
r
ith
m
s
f
o
r
o
u
tlier
d
etec
tio
n
f
r
o
m
wh
ich
it
ca
n
s
ee
n
th
at
en
s
em
b
le
m
eth
o
d
o
u
tp
er
f
o
r
m
s
in
d
i
v
id
u
al
alg
o
r
ith
m
s
.
I
n
[
2
2
]
th
e
au
t
h
o
r
s
u
s
e
Neig
h
b
o
r
e
n
tr
o
p
y
lo
ca
l
o
u
tlier
f
ac
to
r
(
NE
L
OF
)
to
r
ed
u
ce
th
e
tim
e
tak
e
n
to
s
c
an
th
e
d
ata
s
et
as
c
o
m
p
ar
e
d
t
o
L
OF.
I
n
[
2
3
]
th
e
au
th
o
r
s
h
av
e
p
r
o
p
o
s
ed
a
two
s
tag
e
th
r
esh
o
ld
in
g
m
eth
o
d
wh
ich
o
v
e
r
co
m
es
th
e
b
iasi
n
g
p
r
o
b
l
em
s
in
s
tatis
tical
m
eth
o
d
s
.
I
n
o
u
r
wo
r
k
we
in
tr
o
d
u
ce
a
f
ea
s
ib
le
n
o
v
el
s
im
p
le
o
u
tlier
d
etec
tio
n
alg
o
r
ith
m
th
at
u
s
es
b
o
th
d
is
tan
ce
an
d
d
en
s
ity
to
id
en
tif
y
th
e
o
u
t
lier
s
an
d
d
o
es
n
o
t
u
s
e
an
y
u
s
e
r
d
ef
in
ed
p
ar
am
eter
s
.
T
h
e
d
e
n
s
ity
o
f
ea
ch
p
o
in
t
is
esti
m
ated
o
n
ly
o
n
ce
a
n
d
is
u
s
ed
in
th
e
p
r
o
ce
s
s
o
f
d
etec
tin
g
th
e
o
u
tlier
s
.
Als
o
,
th
e
p
r
o
p
o
s
e
d
alg
o
r
ith
m
en
s
u
r
es
in
id
en
tify
i
n
g
t
h
e
s
am
e
s
et
o
f
o
u
tlier
s
ev
er
y
tim
e
th
e
alg
o
r
ith
m
is
ex
ec
u
ted
,
w
h
ich
is
ess
en
tial
in
th
e
r
ea
l
lif
e
p
r
o
b
lem
s
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
2
.
1
.
P
ro
po
s
ed
o
utlier
det
ec
t
io
n t
ec
hn
iqu
e
W
e
p
r
o
p
o
s
e
a
d
is
tan
ce
an
d
d
en
s
ity
-
b
ased
o
u
tlier
d
etec
t
io
n
m
eth
o
d
o
lo
g
y
th
at
ca
n
b
e
u
s
ed
f
o
r
id
en
tify
in
g
g
l
o
b
al
an
d
co
n
te
x
tu
al
o
u
tlier
s
in
d
ata.
T
h
is
p
r
o
ce
s
s
is
es
s
en
tial
f
o
r
v
ar
io
u
s
p
r
ed
ictio
n
an
d
class
if
icatio
n
p
u
r
p
o
s
es
as
th
es
e
o
u
tlier
s
ca
n
ca
u
s
e
a
m
ajo
r
d
ev
iatio
n
in
th
e
r
esu
lts
lead
in
g
to
f
alse
o
u
tco
m
es.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
is
i
n
d
ep
en
d
en
t
o
f
u
s
er
i
n
p
u
t
an
d
h
e
n
ce
p
r
o
v
id
es
a
co
n
s
is
ten
t
o
u
tp
u
t
ev
er
y
tim
e.
T
h
e
r
esu
lts
ar
e
g
en
er
ated
in
a
f
ix
e
d
n
u
m
b
e
r
o
f
iter
atio
n
s
.
2
.
1
.
1
.
Alg
o
rit
hm
1
Ste
p 1
.
I
n
itialize
=
T
o
tal
n
u
m
b
er
o
f
d
ata
p
o
in
ts
Ste
p 2
.
Fin
d
th
e
d
is
tan
ce
m
atr
ix
E
u
clid
ea
n
d
is
tan
ce
(
E
D)
f
o
r
th
e
g
iv
en
p
o
in
ts
u
s
in
g
E
D.
Fo
r
ea
ch
d
ata
p
o
in
t
,
∶
=
1
Fo
r
ea
ch
d
ata
p
o
in
t
,
∶
=
1
C
alcu
late
(
,
)
=
(
,
)
Ste
p
3
.
C
alcu
late
th
e
r
o
w
s
u
m
f
o
r
ea
ch
r
o
w
in
m
atr
ix
(
f
o
r
f
i
n
d
in
g
th
e
ex
tr
em
ities
in
th
e
d
a
taset)
an
d
s
to
r
e
it in
Fo
r
ea
ch
∶
=
1
(
)
=
[
∑
(
,
)
,
=
1
]
(
1
)
So
r
t RS
in
d
escen
d
in
g
o
r
d
er
w
.
r
.
t r
o
w
s
u
m
St
ep
4
.
C
alcu
late
̅
(
i.e
.
th
r
esh
o
ld
r
ad
iu
s
f
o
r
f
in
d
in
g
th
e
co
u
n
t
o
f
an
t
i
-
n
ei
g
h
b
o
r
in
g
p
o
in
ts
)
c
al
cu
late
̅
=
2
|
|
(
|
|
−
1
)
∑
∑
(
,
)
=
+
1
−
1
=
0
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
s
imp
le,
effec
tive
d
is
ta
n
ce
a
n
d
d
en
s
ity
b
a
s
ed
o
u
tlier
d
etec
tio
n
a
lg
o
r
ith
m
(
S
a
jid
h
a
S
.
A
.
)
1143
Ste
p
5
.
C
alcu
late
t
h
e
c
o
u
n
t
o
f
an
ti
-
n
eig
h
b
o
r
s
f
o
r
ea
ch
d
ata
p
o
in
t
as
(
f
o
r
f
in
d
i
n
g
th
e
d
at
a
p
o
i
n
ts
in
t
h
e
s
p
ar
s
est r
eg
io
n
s
)
Fo
r
ea
ch
d
ata
p
o
in
t
∶
=
1
I
n
itialize
co
u
n
t =
0
Fo
r
ea
ch
∶
=
1
If
(
,
)
>
̅
I
n
cr
em
en
t c
o
u
n
t
b
y
1
(
)
=
[
,
]
(
3
)
So
r
t
in
d
escen
d
i
n
g
o
r
d
er
w.
r
.
t
co
u
n
t
Ste
p 6
.
Fin
d
g
ap
v
alu
es in
an
d
as
_
an
d
_
(
f
o
r
f
in
d
in
g
th
e
v
ar
i
atio
n
in
ad
jace
n
t
v
alu
es)
Fo
r
ea
ch
d
ata
p
o
in
t
∶
=
1
−
1
(
)
=
[
(
,
0
)
−
(
+
1
,
0
)
,
]
(
4
)
(
)
=
[
(
,
0
)
−
(
+
1
,
0
)
,
]
(
5
)
So
r
t
_
in
d
escen
d
i
n
g
o
r
d
er
w.
r
.
t
g
ap
s
in
r
o
w
s
u
m
s
So
r
t
_
in
d
escen
d
i
n
g
o
r
d
er
w.
r
.
t
g
ap
s
in
co
u
n
ts
Ste
p 7
.
I
d
en
tif
y
in
g
v
ar
io
u
s
p
o
s
s
ib
le
o
u
tlier
s
ets with
r
esp
ec
t
to
d
if
f
er
e
n
t ɑ v
alu
es
R
ep
ea
t th
e
f
o
llo
win
g
s
tep
s
f
o
r
≔
0
.
01
0
.
09
(
b
y
i
n
cr
em
en
ts
o
f
0
.
01
)
Ste
p 7
.
1
.
I
n
itialize
=
1
(
wh
er
e
n
i
s
th
e
n
u
m
b
e
r
o
f
g
ap
s
to
b
e
co
n
s
id
er
ed
in
f
u
r
th
er
ca
lcu
latio
n
s
)
If
×
>
1
=
(
×
)
{T
ak
in
g
th
e
l
o
wer
b
o
u
n
d
in
te
g
er
o
f
t
h
e
p
r
o
d
u
ct}
Ste
p
7
.
2
.
T
ak
in
g
th
e
m
ea
n
o
f
f
ir
s
t
in
d
ices
in
_
an
d
_
as
_
an
d
_
.
(
I
n
d
ices
s
to
r
ed
with
th
e
g
ap
s
in
_
an
d
_
ar
r
ay
s
ar
e
eq
u
al
to
th
e
n
u
m
b
er
o
f
p
o
in
ts
co
n
s
id
er
ed
as
o
u
tlier
s
if
th
at
g
ap
is
tak
en
as
th
e
d
if
f
er
en
tiatin
g
b
o
u
n
d
f
o
r
n
o
r
m
al
an
d
o
u
tlier
d
at
a
p
o
in
ts
,
s
in
ce
th
e
in
d
ices
ar
e
th
e
in
itial
p
o
s
itio
n
wh
er
e
th
e
g
a
p
s
o
cc
u
r
r
ed
in
t
h
e
s
o
r
ted
an
d
ar
r
ay
s
.
)
=
(
∑
_
(
,
1
)
=
1
)
(
6
)
=
(
∑
_
(
,
1
)
=
1
)
(
7
)
Ste
p
7
.
3
.
I
n
d
ex
cl
o
s
est
to
_
in
_
(
_
)
is
tak
en
as
th
e
d
if
f
er
e
n
tiatio
n
b
o
u
n
d
in
_
.
I
n
d
ex
clo
s
est to
_
in
_
(
_
)
is
tak
en
as th
e
d
if
f
er
en
tiatin
g
b
o
u
n
d
in
_
.
I
n
itialize
_
,
_
=
0
I
n
itialize
_
=
_
(
0
,
1
)
I
n
itialize
_
=
_
(
0
,
1
)
Fo
r
ea
ch
g
a
p
i
∶
=
0
If
(
_
−
_
(
,
1
)
<
min
_
)
_
=
_
−
_
(
,
1
)
(
8
)
_
=
_
(
,
1
)
+
1
if
(
_
−
_
(
,
1
)
<
min
_
)
_
=
_
−
_
(
,
1
)
(
9
)
_
=
_
(
,
1
)
+
1
Ste
p
7
.
4
.
Data
p
o
in
ts
co
r
r
esp
o
n
d
in
g
to
in
d
ices
0
to
idx
_
RS
in
a
n
d
d
ata
p
o
i
n
ts
co
r
r
esp
o
n
d
i
n
g
to
i
n
d
ices
0
to
_
in
CT
ar
r
ay
ar
e
c
o
n
s
id
er
ed
as
_
.
I
n
itialize
_
as a
n
em
p
ty
ar
r
ay
o
f
s
ize
_
.
Fo
r
ea
ch
≔
0
_
_
(
)
=
(
,
1
)
(
1
0
)
I
n
itialize
_
as a
n
em
p
ty
a
r
r
ay
Fo
r
ea
ch
≔
0
_
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
24
,
No
.
2
,
No
v
em
b
er
2
0
2
1
:
1
1
4
1
-
1
1
4
8
1144
_
(
)
=
(
,
1
)
(
1
1
)
Ste
p 7
.
5
.
T
h
e
f
in
al
lis
t o
f
o
u
tlier
s
ar
e
g
en
er
ate
d
b
y
tak
in
g
t
h
e
u
n
io
n
o
f
_
an
d
_
.
=
(
_
)
∪
(
_
)
(
1
2
)
Ste
p
8
.
R
etu
r
n
to
ca
s
e
wit
h
m
ax
im
u
m
AUC
(
i.e
.
Ac
cu
r
ac
y
m
etr
ic
u
s
ed
in
th
e
s
tu
d
y
)
alo
n
g
with
co
r
r
esp
o
n
d
in
g
α
v
alu
e
.
2
.
1
.
2
.
Descript
io
n o
f
t
he
a
lg
o
rit
hm
I
n
th
e
alg
o
r
ith
m
1
m
e
n
tio
n
ed
ab
o
v
e,
we
co
n
s
id
er
b
o
t
h
d
is
tan
ce
s
(
i.e
.
in
th
e
f
o
r
m
o
f
r
o
w
s
u
m
)
an
d
d
en
s
ity
(
i.e
.
in
th
e
f
o
r
m
o
f
co
u
n
ts
)
(
i.e
.
Step
s
3
,
4
,
5
)
.
T
h
ese
ca
lcu
latio
n
s
ar
e
d
o
n
e
o
n
ly
a
f
ter
th
e
co
m
p
u
tatio
n
o
f
a
d
is
tan
ce
m
atr
i
x
E
D
t
h
at
t
ak
es
(
2
)
tim
e
(
i.e
.
Step
2
)
.
On
e
o
f
th
e
m
ain
b
en
e
f
its
o
f
u
s
in
g
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
is
th
e
co
n
s
is
ten
cy
o
f
th
e
r
esu
lts
.
T
h
er
e
is
an
u
n
c
h
an
g
ed
1
0
-
e
p
o
ch
p
r
o
ce
s
s
th
at
is
ex
ec
u
ted
o
n
ce
f
o
r
ea
ch
d
ataset.
Hen
ce
,
a
r
ea
s
o
n
a
b
le
tim
e
is
tak
en
b
y
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
t
o
p
ar
s
e
o
v
er
th
e
d
at
asets
.
2
.
1
.
3
.
I
llu
s
t
ra
t
io
n o
f
t
he
pro
po
s
ed
o
utlier
det
ec
t
io
n a
lg
o
rit
hm
A
d
a
t
as
e
t
is
g
e
n
e
r
a
t
e
d
s
y
n
t
h
e
tic
a
l
l
y
t
o
i
ll
u
s
t
r
at
e
t
h
e
p
r
o
c
e
s
s
i
s
g
i
v
e
n
i
n
T
a
b
l
e
1
.
T
h
e
d
a
t
as
e
t
c
o
n
s
i
s
t
s
o
f
1
5
d
a
t
a
p
o
i
n
t
s
w
i
t
h
2
at
t
r
i
b
u
t
es
(
)
.
A
t
t
r
i
b
u
te
L
a
b
e
l
0
i
d
e
n
t
i
f
ie
s
a
d
a
t
a
p
o
i
n
t
as
a
n
o
u
t
li
e
r
.
T
h
e
p
lo
t
o
f
t
h
e
d
a
t
a
p
o
i
n
ts
h
a
s
b
e
en
s
h
o
w
n
i
n
F
i
g
u
r
e
1
.
A
d
d
i
ti
o
n
a
l
l
y
,
5
d
a
t
a
p
o
i
n
ts
{
(
0
.
6
,
0
.
3
)
,
(
0
.
8
,
0
.
1
)
,
(
0
.
1
)
,
(
1
,
0
.
4
)
,
(
0
.
04
)
}
h
a
v
e
b
e
e
n
a
d
d
e
d
a
s
o
u
t
l
ie
r
s
i
n
t
h
e
d
a
t
as
e
t
.
T
h
e
p
l
o
t
wi
t
h
a
d
d
i
ti
o
n
a
l
p
o
i
n
t
s
h
a
s
b
e
e
n
s
h
o
w
n
i
n
Fi
g
u
r
e
2
.
So
r
ted
r
o
w
s
u
m
v
alu
es
al
o
n
g
with
co
r
r
esp
o
n
d
in
g
o
b
ject
in
d
ices
ar
e
s
h
o
wn
in
T
ab
le
2
.
T
ab
le
1
.
Data
p
o
in
ts
co
n
s
id
er
e
d
f
o
r
illu
s
tr
atio
n
O
b
j
e
c
t
I
n
d
e
x
X
Y
La
b
e
l
0
0
.
3
8
8
0
.
4
3
2
0
1
0
.
1
1
1
0
.
1
0
1
0
2
0
.
3
0
5
0
.
0
8
4
0
3
0
.
5
2
7
0
.
6
4
4
0
4
0
.
5
5
5
0
.
5
7
6
0
5
0
.
1
9
8
0
.
0
6
7
0
6
0
.
8
3
3
0
.
8
9
8
0
7
0
.
5
8
3
0
.
7
7
9
0
8
0
.
8
0
5
0
.
8
1
3
0
9
0
.
0
5
5
0
.
0
5
0
0
10
0
.
4
1
6
0
.
6
9
4
0
11
0
.
1
9
4
0
.
4
2
3
0
12
0
.
1
6
6
0
.
0
8
4
0
13
0
.
5
0
.
6
6
1
0
14
0
.
5
2
7
0
.
5
5
9
0
15
0
.
6
0
.
3
0
16
0
.
8
0
.
1
0
17
0
.
0
1
.
0
0
18
1
.
0
0
.
4
0
19
0
.
0
0
.
4
0
T
ab
le
2
.
R
o
w
s
u
m
s
o
f
all
d
ata
p
o
in
ts
s
o
r
ted
in
d
escen
d
in
g
o
r
d
er
I
n
d
e
x
R
o
w
S
u
m
O
b
j
e
c
t
I
n
d
e
x
0
2
2
.
8
4
0
17
1
2
1
.
1
8
6
18
2
2
0
.
5
6
0
16
3
2
0
.
1
7
8
19
4
1
8
.
6
4
9
15
5
1
5
.
5
5
6
6
6
1
4
.
5
4
5
9
7
1
4
.
4
0
2
8
8
1
3
.
4
8
8
1
9
1
3
.
2
2
7
5
10
1
3
.
1
9
7
12
11
1
2
.
8
3
1
2
12
1
2
.
3
2
3
7
13
1
1
.
6
8
7
11
14
1
1
.
3
5
9
10
15
1
0
.
9
6
8
13
16
1
0
.
9
2
2
3
17
1
0
.
8
1
8
4
18
1
0
.
7
4
2
0
19
1
0
.
6
8
1
14
T
o
f
in
d
th
e
t
h
r
esh
o
ld
f
o
r
co
n
s
id
er
in
g
a
p
o
in
t
as
a
n
o
u
tlier
,
g
ap
s
b
etwe
en
ad
jace
n
t
v
alu
es
o
f
r
o
w
s
u
m
s
in
T
ab
le
2
ar
e
ca
lc
u
lated
an
d
s
o
r
ted
i
n
d
escen
d
i
n
g
o
r
d
er
ar
e
s
h
o
wn
in
T
ab
le
3
.
T
h
e
m
ea
n
o
f
f
ir
s
t
Den
s
ity
I
n
d
ex
as
s
h
o
wn
in
T
a
b
le
3
,
f
o
r
ea
c
h
v
al
u
e
o
f
=0
.
1
t
o
0
.
9
is
co
m
p
u
ted
an
d
f
in
d
in
g
th
e
d
en
s
ity
in
d
ex
clo
s
est
to
th
e
m
ea
n
g
iv
es
th
e
in
d
ex
v
alu
e
c
o
n
s
id
er
ed
(
f
o
r
T
a
b
le
3
th
e
id
en
tifie
d
in
d
e
x
is
4
i.e
.
th
e
f
ir
s
t
v
alu
e)
.
T
h
e
d
ata
p
o
in
ts
b
ef
o
r
e
th
is
in
d
ex
(
i.e
.
in
d
ex
4
in
T
ab
le
2
)
ar
e
co
n
s
id
er
ed
as o
u
tlier
s
.
F
i
n
d
t
h
e
c
o
u
n
t
o
f
a
n
t
i
-
n
e
i
g
h
b
o
r
i
n
g
p
o
i
n
t
s
f
o
r
e
a
c
h
p
o
i
n
t
u
s
i
n
g
r
a
d
i
a
l
d
is
t
a
n
c
e
(
̅
)
a
s
e
x
p
l
a
i
n
e
d
i
n
S
e
c
t
i
o
n
3
(
i
.
e
.
i
n
(
1
)
)
.
T
h
e
s
e
v
a
l
u
e
s
a
r
e
s
o
r
t
e
d
i
n
d
es
c
e
n
d
i
n
g
o
r
d
e
r
a
s
s
h
o
w
n
i
n
T
a
b
l
e
4
,
w
e
g
e
t
t
h
e
d
a
t
a
p
o
i
n
ts
t
h
a
t
a
r
e
f
u
r
t
h
e
s
t
f
r
o
m
f
o
r
m
i
n
g
a
c
l
u
s
t
e
r
.
T
o
f
i
n
d
a
t
h
r
e
s
h
o
l
d
s
i
m
i
l
a
r
t
o
d
i
f
f
e
r
e
n
ti
a
t
e
o
u
t
li
e
r
s
f
r
o
m
n
o
r
m
a
l
p
o
i
n
t
s
t
h
e
g
a
p
s
b
e
t
w
e
e
n
a
d
j
a
c
e
n
t
v
a
lu
e
s
o
f
c
o
u
n
t
s
i
s
c
o
m
p
u
t
e
d
a
n
d
s
o
r
t
e
d
i
n
d
e
s
c
e
n
d
i
n
g
o
r
d
e
r
a
s
s
h
o
w
n
i
n
T
a
b
l
e
4
.
S
o
r
t
e
d
c
o
u
n
t
g
a
p
s
a
l
o
n
g
wi
t
h
th
e
i
n
d
e
x
(
i
.
e
.
p
o
s
i
ti
o
n
)
w
h
e
r
e
th
e
y
o
c
c
u
r
r
e
d
i
n
T
a
b
l
e
4
a
r
e
s
h
o
w
n
i
n
T
a
b
l
e
5
.
I
d
en
tify
in
g
th
e
ap
p
r
o
p
r
iate
i
n
d
ex
f
r
o
m
t
h
e
s
o
r
ted
g
ap
s
to
c
alcu
late
th
e
v
alu
es
o
f
(
)
(
i.e
.
n
u
m
b
er
o
f
g
ap
s
to
b
e
co
n
s
id
er
ed
f
r
o
m
th
e
s
tar
tin
g
o
f
s
o
r
ted
g
a
p
s
ar
r
ay
)
is
d
o
n
e
as
m
en
tio
n
ed
i
n
Step
7
.
1
Fo
r
T
a
b
le
5
th
e
id
en
tifie
d
in
d
e
x
is
4
i.e
.
th
e
f
ir
s
t
v
alu
e.
T
h
er
ef
o
r
e,
t
h
e
d
ata
p
o
in
ts
f
r
o
m
in
d
e
x
0
4
in
clu
s
iv
e
as
s
h
o
wn
in
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
s
imp
le,
effec
tive
d
is
ta
n
ce
a
n
d
d
en
s
ity
b
a
s
ed
o
u
tlier
d
etec
tio
n
a
lg
o
r
ith
m
(
S
a
jid
h
a
S
.
A
.
)
1145
T
ab
le
4
,
ar
e
id
en
tifie
d
as
o
u
tlier
s
(
wh
en
u
s
in
g
co
u
n
ts
o
n
ly
)
.
T
h
u
s
,
th
e
f
in
al
lab
els
o
f
d
ata
p
o
in
ts
f
r
o
m
0
to
1
4
ar
e
0
an
d
f
r
o
m
1
5
to
1
9
ar
e
f
o
u
n
d
as 1
w
h
ich
ar
e
th
e
f
in
al
o
u
tlier
s
.
T
h
is
is
s
h
o
wn
in
Fig
u
r
e
3
.
Un
io
n
o
f
o
u
tlier
s
id
en
tifie
d
with
r
o
w
s
u
m
s
an
d
th
at
with
th
e
co
u
n
ts
is
tak
en
as th
e
o
u
tlier
s
et.
Ou
tlier
s
ets
ar
e
g
en
er
ated
f
o
r
ea
ch
v
a
lu
e
o
f
α
.
AUC
s
co
r
es
f
o
r
th
e
d
ataset
w.
r
.
t
ea
ch
o
u
tlier
s
et
is
ca
lcu
lated
an
d
th
e
o
u
tlier
s
et
g
en
er
atin
g
th
e
b
est
(
i.e
.
,
h
ig
h
est)
AUC
v
alu
e
is
r
etu
r
n
ed
as
th
e
f
in
al
s
et
o
f
o
u
tlier
s
.
T
ab
le
6
s
h
o
ws
th
e
d
ata
p
o
i
n
ts
m
ar
k
ed
i
n
o
r
a
n
g
e
ar
e
id
en
tifie
d
as o
u
tlier
s
.
T
a
b
l
e
3
.
G
a
p
s
c
o
r
r
es
p
o
n
d
i
n
g
t
o
r
o
w
s
u
m
i
n
d
e
x
I
n
d
e
x
R
o
w
S
u
m
O
b
j
e
c
t
I
n
d
e
x
3
.
0
9
2
4
3
.
0
9
2
1
.
6
3
3
0
1
.
6
3
3
1
.
5
2
9
3
1
.
5
2
9
1
.
0
1
1
5
1
.
0
1
1
0
.
9
1
4
7
0
.
9
1
4
0
,
6
3
5
12
0
,
6
3
5
0
.
6
2
6
1
0
.
6
2
6
0
.
5
0
8
11
0
.
5
0
8
0
.
3
9
1
14
0
.
3
9
1
0
.
3
8
1
2
0
.
3
8
1
0
.
3
6
6
10
0
.
3
6
6
0
.
3
2
7
13
0
.
3
2
7
0
.
2
6
0
8
0
.
2
6
0
0
.
1
4
2
6
0
.
1
4
2
0
.
1
0
4
16
0
.
1
0
4
0
.
0
7
6
7
0
.
0
7
6
0
.
0
6
1
18
0
.
0
6
1
0
.
0
4
5
15
0
.
0
4
5
0
.
0
2
9
9
0
.
0
2
9
T
ab
le
4
.
C
o
u
n
ts
r
elate
d
to
th
e
d
ata
p
o
in
ts
in
s
o
r
ted
o
r
d
er
C
o
u
n
t
I
n
d
e
x
C
o
u
n
t
s
O
b
j
e
c
t
I
n
d
e
x
0
18
17
1
17
16
2
17
18
3
17
18
4
16
15
5
11
6
6
11
9
7
10
7
8
10
8
9
8
1
10
8
2
11
8
5
12
8
12
13
6
3
14
6
10
15
6
11
16
6
13
17
5
0
18
5
4
19
5
14
T
ab
le
5
.
Gap
s
co
r
r
esp
o
n
d
in
g
t
o
co
u
n
t
v
alu
es
T
ab
le
6
.
Fin
al
lab
els o
f
d
ata
p
o
in
ts
C
o
u
n
t
G
a
p
s
C
o
u
n
t
I
n
d
e
x
5
4
2
8
2
12
1
0
1
3
1
6
1
16
0
1
0
2
0
5
0
7
0
9
0
10
0
11
0
13
0
14
0
15
0
17
0
18
O
b
j
e
c
t
i
n
d
e
x
X
Y
La
b
e
l
0
0
.
3
8
8
0
.
4
3
2
0
1
0
.
1
1
1
0
.
1
0
1
0
2
0
.
3
0
5
0
.
0
8
4
0
3
0
.
5
2
7
0
.
6
4
4
0
4
0
.
5
5
5
0
.
5
7
6
0
5
0
.
1
9
8
0
.
0
6
7
0
6
0
.
8
3
3
0
.
8
9
8
0
7
0
.
5
8
3
0
.
7
7
9
0
8
0
.
8
0
5
0
.
8
1
3
0
9
0
.
0
5
5
0
.
0
5
0
0
10
0
.
4
1
6
0
.
6
9
4
0
11
0
.
1
9
4
0
.
4
2
3
0
12
0
.
1
6
6
0
.
0
8
4
0
13
0
.
5
0
.
6
6
1
0
14
0
.
5
2
7
0
.
5
5
9
0
15
0
.
6
0
.
3
1
16
0
.
8
0
.
1
1
17
0
.
0
1
.
0
1
18
1
.
0
0
.
4
1
19
0
.
0
0
.
4
1
Fig
u
r
e
1
.
Sy
n
th
etic
d
ataset
o
f
1
5
d
ata
p
o
in
ts
Fig
u
r
e
2
.
Data
s
et
af
ter
ad
d
in
g
o
u
tlier
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
24
,
No
.
2
,
No
v
em
b
er
2
0
2
1
:
1
1
4
1
-
1
1
4
8
1146
Fig
u
r
e
3
.
Sy
n
th
etic
d
ataset
with
id
en
tifie
d
o
u
tlier
s
[
(
0
(
b
lu
e)
non
-
o
u
tlier
s
,
1
(
o
r
an
g
e
)
-
o
u
tlier
s
]
3.
RE
SU
L
T
S
A
ND
D
IS
CU
SS
I
O
N
Fo
u
r
d
atasets
wer
e
u
s
ed
f
r
o
m
th
e
UC
I
r
ep
o
s
ito
r
y
[
2
4
]
an
d
E
L
KI
lib
r
ar
y
[
2
5
]
f
o
r
th
e
ev
a
lu
atio
n
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
.
T
ab
le
7
p
r
o
v
id
es
t
h
e
d
etails
o
f
th
e
d
atasets
,
wh
er
e
an
d
r
e
f
er
to
th
e
to
tal
n
u
m
b
er
o
f
d
ata
p
o
i
n
ts
an
d
o
u
tlier
s
,
r
esp
ec
tiv
ely
.
AUC
is
a
n
o
u
tp
u
t
m
etr
ic
f
o
r
class
if
icatio
n
p
r
o
b
lem
s
u
s
in
g
d
if
f
er
en
t
th
r
esh
o
ld
s
ettin
g
s
.
Fo
r
a
d
ataset,
th
e
tr
u
e
p
o
s
itiv
e
r
ate
an
d
th
e
f
alse
p
o
s
itiv
e
r
ate
v
alu
es
o
f
th
e
r
esp
ec
tiv
e
o
u
t
p
u
ts
ar
e
c
alcu
lated
(
i.e
.
u
s
in
g
(
1
3
)
a
n
d
(
1
4
)
)
.
F
u
r
t
h
er
,
th
ey
ar
e
p
lo
tted
as
a
n
R
OC
cu
r
v
e.
Hig
h
e
r
th
e
AUC v
alu
es,
b
etter
th
e
m
e
th
o
d
o
lo
g
y
.
=
+
(
1
3
)
=
+
(
1
4
)
T
ab
le
7
.
R
ea
l tim
e
d
atasets
u
s
ed
in
th
e
e
x
p
er
im
en
t
D
a
t
a
s
e
t
N
o
.
o
f
d
a
t
a
p
o
i
n
t
s
N
(
A
c
t
u
a
l
O
u
t
l
i
e
r
s
-
O)
N
o
.
o
f
A
t
t
r
i
b
u
t
e
s
A
r
r
h
y
t
h
m
i
a
4
5
0
(
2
0
6
)
2
5
9
S
p
a
m
b
a
s
e
4
6
0
1
(
1
8
1
3
)
51
C
a
r
d
i
o
t
o
2
1
2
6
(
4
7
1
)
21
S
t
a
m
p
s
3
4
0
(
3
1
)
9
W
e
h
av
e
c
o
m
p
ar
e
d
o
u
r
p
r
o
p
o
s
ed
al
g
o
r
ith
m
with
o
th
e
r
o
u
tlier
d
etec
tio
n
alg
o
r
ith
m
s
,
n
am
ely
Fas
tA
B
OD
[
1
0
]
,
NN
[
1
1
]
,
k
N
NW
[
1
1
]
,
ODI
N
[
6
]
,
L
OF
[
5
]
,
L
o
OP
[
1
2
]
,
C
OP
[
1
5
]
,
SOD
[
1
3
]
,
Gu
MM
[
9
]
an
d
HiC
S
[
1
4
]
.
Fro
m
t
h
e
AU
C
r
esu
lts
o
f
C
ar
d
io
to
d
ataset
s
h
o
wn
in
T
ab
le
8
an
d
Fig
u
r
e
4
it
ca
n
b
e
o
b
s
er
v
ed
th
at
th
e
p
r
o
p
o
s
ed
o
u
tlier
d
etec
tio
n
alg
o
r
ith
m
o
u
tp
er
f
o
r
m
s
Fas
tA
B
OD
[
1
0
]
,
k
NN
[
1
1
]
,
k
NNW
[
1
1
]
,
ODI
N
[
6
]
,
L
OF
[
5
]
,
L
o
OP
[
1
2
]
,
C
O
P
[
1
5
]
,
SOD
[
1
3
]
,
Gu
MM
[
9
]
an
d
HiC
S
[
1
4
]
alg
o
r
ith
m
s
.
T
h
e
p
r
o
p
o
s
ed
o
u
tlier
d
etec
tio
n
alg
o
r
ith
m
b
ased
o
n
th
e
AUC
v
alu
e
f
o
r
Sp
am
b
ase
d
ataset
o
u
t
p
er
f
o
r
m
s
O
DI
N
[
6
]
,
C
OP
[
1
5
]
,
GUM
M
[
9
]
,
p
er
f
o
r
m
s
s
am
e
a
s
L
OF
[
5
]
a
n
d
s
lig
h
tly
less
co
m
p
ar
ed
to
t
h
e
r
est
o
f
th
e
alg
o
r
ith
m
s
.
T
h
e
AUC
Valu
es
as
s
h
o
wn
in
T
ab
le
8
,
also
s
h
o
w
th
at
o
u
r
p
r
o
p
o
s
ed
o
u
tlier
d
etec
tio
n
alg
o
r
ith
m
p
e
r
f
o
r
m
e
d
b
etter
th
an
GUM
M
[
9
]
an
d
s
lig
h
tly
less
er
th
an
o
th
e
r
m
eth
o
d
s
f
o
r
Ar
r
h
y
t
h
m
ia
d
ataset.
T
ab
le
8
.
C
o
m
p
a
r
is
o
n
o
f
AUC f
o
r
4
r
ea
l
-
tim
e
d
atasets
(
p
r
o
p
o
s
ed
alg
o
r
ith
m
r
esu
lts
ar
e
h
ig
h
e
r
th
an
th
e
b
o
ld
ed
v
alu
es)
A
l
g
o
r
i
t
h
ms
A
r
r
h
y
t
h
m
i
a
S
p
a
m
b
a
s
e
C
a
r
d
i
o
t
o
S
t
a
m
p
s
F
a
st
A
B
O
D
0
.
7
4
0
.
0
1
0
.
5
5
0
.
0
1
K
N
N
[
1
1
]
0
.
7
5
0
.
6
4
0
.
6
2
0
.
9
3
K
N
N
W
[
1
1
]
0
.
7
5
0
.
6
3
0
.
5
9
0
.
9
0
ODIN
[
6
]
0
.
7
1
0
.
5
1
0
.
5
8
0
.
8
1
LO
F
[
5
]
0
.
7
4
0
.
5
2
0
.
5
9
0
.
9
5
Lo
O
P
[
1
2
]
0
.
7
3
0
.
5
4
0
.
5
7
0
.
6
7
C
O
P
[
1
5
]
0
.
7
0
0
.
5
0
0
.
5
7
0
.
6
9
S
O
D
[
1
3
]
0
.
7
3
0
.
5
5
0
.
5
2
0
.
6
2
G
U
M
M
[
9
]
0
.
4
7
0
.
5
0
0
.
5
1
0
.
8
7
H
i
C
S
[
1
4
]
0
.
7
0
0
.
5
5
0
.
6
0
0
.
9
5
P
r
o
p
o
se
d
0
.
6
7
0
.
5
2
0
.
9
0
0
.
0
4
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
s
imp
le,
effec
tive
d
is
ta
n
ce
a
n
d
d
en
s
ity
b
a
s
ed
o
u
tlier
d
etec
tio
n
a
lg
o
r
ith
m
(
S
a
jid
h
a
S
.
A
.
)
1147
Fig
u
r
e
4
.
AUC
co
m
p
a
r
is
o
n
f
o
r
ca
r
d
io
to
an
d
Sp
a
m
b
ase
d
ata
s
et
T
ab
le
9
s
h
o
ws
th
e
ac
t
u
al
o
u
tl
ier
to
n
o
r
m
al
d
ata
p
o
in
t
r
ati
o
s
with
th
e
tim
e
tak
en
f
o
r
e
x
e
cu
tin
g
th
e
alg
o
r
ith
m
f
o
r
th
e
ab
o
v
e
-
m
e
n
tio
n
ed
d
atasets
.
Fro
m
th
is
we
o
b
s
er
v
e
th
at
th
e
r
atio
f
o
r
C
ar
d
i
o
to
,
Sp
am
b
ase,
a
n
d
Ar
r
h
y
th
m
ia
is
s
ig
n
if
ican
tly
h
ig
h
er
as
co
m
p
a
r
ed
to
th
e
Stam
p
s
d
ataset.
Als
o
,
C
ar
d
io
to
,
Ar
r
h
y
th
m
ia,
an
d
Sp
am
b
ase
d
ataset
attr
ib
u
tes
co
n
tain
n
u
m
er
ical
d
ata.
T
h
is
c
an
b
e
a
co
n
t
r
ib
u
tin
g
f
ac
to
r
f
o
r
d
eter
m
in
i
n
g
th
e
p
er
f
o
r
m
an
ce
o
f
an
alg
o
r
ith
m
.
T
h
e
Sp
am
b
ase
d
a
taset
d
escr
ib
es
th
e
wo
r
d
an
d
ch
a
r
f
r
e
q
u
en
c
y
in
an
em
ail;
s
u
ch
d
ig
ital
f
r
eq
u
en
cies
g
iv
e
v
e
r
y
f
ew
g
r
o
u
p
c
h
ar
ac
ter
is
tics
[
1
8
]
wh
ich
ca
n
also
co
n
tr
ib
u
te
t
o
th
e
r
esu
lts
o
f
o
u
r
p
r
o
p
o
s
ed
o
u
tlier
d
etec
tio
n
alg
o
r
ith
m
.
Fro
m
T
ab
le
8
th
e
AUC
v
alu
es
f
o
r
th
e
Stam
p
s
d
ataset
is
less
er
co
m
p
a
r
ed
t
o
th
e
o
th
er
alg
o
r
ith
m
s
.
As
th
is
d
ataset
al
s
o
h
av
e
v
er
y
f
e
w
g
r
o
u
p
in
g
ch
ar
ac
ter
is
tics
[
1
8
]
th
is
also
co
n
tr
ib
u
tes
to
th
e
co
n
s
id
er
ab
ly
lo
we
r
r
esu
lts
f
r
o
m
o
u
r
p
r
o
p
o
s
ed
alg
o
r
ith
m
.
Fro
m
T
ab
le
9
,
we
o
b
s
er
v
e
th
at
th
e
o
u
tlier
to
n
o
r
m
al
d
ata
p
o
in
ts
r
atio
is
0
.
1
in
ca
s
e
o
f
th
e
Stam
p
s
d
atase
t.
T
h
e
co
m
p
ar
ativ
ely
l
o
wer
AUC
s
co
r
es
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
f
o
r
th
e
Stam
p
s
d
ataset
ca
n
also
b
e
at
tr
ib
u
ted
to
th
e
lo
w
r
atio
o
f
o
u
tlier
s
to
n
o
r
m
al
d
ata
p
o
in
ts
.
Sin
ce
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
u
s
es
d
is
tan
ce
an
d
d
en
s
ity
f
o
r
id
en
ti
f
y
in
g
o
u
tlier
s
f
r
o
m
n
o
r
m
al
d
ata
p
o
in
ts
,
b
etter
r
esu
lts
ca
n
b
e
ac
h
iev
ed
with
d
atasets
h
av
in
g
g
r
o
u
p
ch
ar
ac
ter
is
tics
.
O
n
e
o
f
t
h
e
b
e
n
e
f
i
t
s
o
f
u
s
i
n
g
t
h
e
p
r
o
p
o
s
e
d
a
l
g
o
r
i
t
h
m
i
s
t
h
e
c
o
n
s
i
s
t
e
n
c
y
o
f
t
h
e
r
e
s
u
l
ts
.
Mo
s
t
o
f
t
h
e
m
e
t
h
o
d
o
l
o
g
i
e
s
u
s
e
d
f
o
r
c
o
m
p
a
r
i
s
o
n
,
i
n
T
a
b
l
e
8
u
t
il
i
z
e
u
s
er
-
d
e
f
i
n
e
d
v
a
r
i
a
b
l
es
d
u
r
i
n
g
t
h
e
p
r
o
c
e
s
s
o
f
o
u
tl
i
er
d
e
t
e
c
t
i
o
n
.
T
h
u
s
,
it
c
a
n
b
e
u
n
e
q
u
i
v
o
c
a
l
l
y
p
r
o
v
e
d
t
h
a
t
e
v
e
n
t
h
o
u
g
h
i
n
s
o
m
e
c
a
s
es
t
h
e
p
r
o
p
o
s
ed
m
e
t
h
o
d
o
l
o
g
y
d
o
e
s
n
o
t
g
i
v
e
g
o
o
d
A
UC
v
al
u
e
s
,
i
t
o
n
l
y
t
a
k
e
s
a
d
e
f
i
n
i
t
e
n
u
m
b
e
r
o
f
r
u
n
s
w
it
h
o
u
t
a
n
y
d
e
p
e
n
d
e
n
cy
o
n
a
u
s
e
r
-
d
e
f
i
n
e
d
p
a
r
a
m
e
t
e
r
.
T
h
e
r
e
is
a
n
u
n
c
h
a
n
g
e
d
10
-
e
p
o
c
h
p
r
o
c
e
s
s
t
h
a
t
i
s
e
x
e
c
u
t
e
d
o
n
c
e
f
o
r
e
a
c
h
d
a
t
as
et
.
H
e
n
c
e
,
a
r
e
as
o
n
a
b
l
e
t
i
m
e
is
t
a
k
e
n
b
y
t
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
t
o
p
a
r
s
e
o
v
e
r
t
h
e
d
at
a
s
et
s
as
s
h
o
w
n
i
n
T
a
b
l
e
9
.
I
t
i
s
s
et
s
o
t
o
g
i
v
e
t
h
e
b
e
s
t
d
i
f
f
e
r
e
n
t
i
a
t
i
o
n
b
et
w
e
e
n
o
u
t
li
e
r
s
a
n
d
n
o
r
m
a
l
d
a
t
a
p
o
i
n
t
s
a
n
d
a
v
o
i
d
i
n
g
f
a
l
s
e
-
p
o
s
i
t
i
v
e
o
u
tl
i
e
r
s
.
A
l
s
o
,
t
h
e
f
i
n
a
l
s
e
t
o
f
o
u
t
l
i
e
r
s
i
s
c
o
n
s
is
t
e
n
t
i
n
n
a
t
u
r
e
a
s
o
u
r
p
r
o
p
o
s
e
d
a
l
g
o
r
i
t
h
m
d
o
e
s
n
o
t
u
s
e
a
n
y
u
s
e
r
d
e
f
i
n
e
d
p
a
r
am
e
t
e
r
s
.
T
ab
le
9
.
R
ea
l
-
tim
e
d
atasets
an
d
r
esu
lts
wi
th
tim
e
o
f
ex
ec
u
tio
n
in
m
illi
s
ec
o
n
d
s
D
a
t
a
s
e
t
N
o
.
o
f
D
a
t
a
p
o
i
n
t
s
(
T
o
t
a
l
)
A
c
t
u
a
l
O
u
t
l
i
e
r
s
(
A
c
t
u
a
l
O
u
t
l
i
e
r
s)
/
(
T
o
t
a
l
-
A
c
t
u
a
l
O
u
t
l
i
e
r
s)
Ti
me
(
ms)
A
r
r
h
y
t
h
m
i
a
4
5
0
2
0
6
0
.
8
4
4
0
.
1
S
p
a
m
b
a
s
e
4
6
0
1
1
8
1
3
0
.
6
5
4
.
4
2
C
a
r
d
i
o
t
o
2
1
2
6
4
7
1
0
.
2
8
4
0
.
9
2
S
t
a
m
p
s
3
4
0
31
0
.
1
0
.
0
2
4.
CO
NCLU
SI
O
N
T
h
e
wo
r
k
we
h
av
e
p
r
o
p
o
s
ed
i
s
to
d
etec
t
o
u
tlier
s
in
clu
s
ter
i
n
g
alg
o
r
ith
m
s
wh
ich
is
a
f
ea
s
ib
le
n
o
v
el
s
im
p
le
o
u
tlier
d
etec
tio
n
alg
o
r
i
th
m
th
at
u
s
es
b
o
th
d
is
tan
ce
an
d
d
en
s
ity
to
id
en
tify
th
e
o
u
tlie
r
s
an
d
d
o
es
n
o
t
u
s
e
an
y
u
s
er
d
ef
in
ed
p
ar
a
m
eter
s
.
T
h
e
d
en
s
ity
o
f
ea
ch
p
o
i
n
t
is
esti
m
ated
o
n
ly
o
n
ce
an
d
is
u
s
e
d
in
th
e
p
r
o
ce
s
s
o
f
d
etec
tin
g
th
e
o
u
tlier
s
.
Als
o
,
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
en
s
u
r
es
in
id
en
tify
i
n
g
th
e
s
am
e
s
et
o
f
o
u
tlier
s
ev
er
y
tim
e
th
e
alg
o
r
ith
m
is
ex
ec
u
ted
,
wh
ich
is
ess
en
tial
in
th
e
r
ea
l
life
p
r
o
b
lem
s
.
I
n
o
u
r
p
r
o
p
o
s
ed
n
o
v
el
s
im
p
le
o
u
tlie
r
d
etec
tio
n
alg
o
r
ith
m
,
th
er
e
is
a
n
u
n
ch
a
n
g
ed
1
0
-
e
p
o
ch
p
r
o
ce
s
s
th
at
is
ex
ec
u
ted
o
n
ce
f
o
r
ea
c
h
d
ataset.
Hen
ce
,
a
r
ea
s
o
n
ab
le
tim
e
is
tak
en
b
y
th
e
p
r
o
p
o
s
ed
m
o
d
el
to
p
ar
s
e
o
v
er
th
e
d
atasets
.
I
t
is
s
et
s
o
to
g
iv
e
th
e
b
est
d
if
f
er
en
tiatio
n
b
etwe
en
o
u
tlier
s
an
d
n
o
r
m
al
d
ata
p
o
in
ts
an
d
av
o
id
in
g
f
alse
-
p
o
s
itiv
e
o
u
tlier
s
.
I
t
also
lim
its
th
e
n
u
m
b
er
o
f
iter
atio
n
s
th
r
o
u
g
h
wh
ich
th
e
alg
o
r
ith
m
ex
ec
u
tes.
RE
F
E
R
E
NC
E
S
[1
]
D.
R.
Bril
li
n
g
e
r
.
,
“
Da
ta
An
a
ly
sis
-
Ex
p
l
o
ra
to
r
y
,
”
Ame
ric
a
n
J
o
u
rn
a
l
o
f
Po
l
it
ica
l
S
c
ien
c
e
,
v
o
l.
5
2
,
n
o
.
3
,
p
p
.
7
0
5
-
7
2
2
,
2
0
1
1
.
[
2
]
S
.
S
.
A
z
i
m
u
d
d
i
n
a
n
d
K
.
D
e
s
i
k
a
n
,
“
A
s
i
m
p
l
e
d
e
n
s
i
t
y
w
i
t
h
d
i
s
t
a
n
c
e
b
a
s
e
d
i
n
i
t
i
a
l
s
e
e
d
s
e
l
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
f
o
r
K
-
m
e
a
n
s
a
l
g
o
r
i
t
h
m
,
”
C
I
T
.
J.
C
o
m
p
u
t
.
i
n
f
o
r
m
a
t
i
o
n
.
T
e
c
h
n
o
l
o
g
y
, v
o
l
.
2
5
,
n
o.
4
,
p
p
.
291
-
3
0
0
,
2
0
1
7
,
doi
:
1
0
.
2
0
5
3
2
/
c
i
t
.
2
0
1
7
.
1
0
0
3
6
0
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
24
,
No
.
2
,
No
v
em
b
er
2
0
2
1
:
1
1
4
1
-
1
1
4
8
1148
[3
]
S
.
A.
S
a
ji
d
h
a
,
S
.
P
.
C
h
o
d
n
e
k
a
r
,
a
n
d
K.
De
sik
a
n
,
“
In
it
ial
se
e
d
se
lec
ti
o
n
fo
r
c
l
u
ste
rin
g
:
a
d
istan
c
e
a
n
d
d
e
n
sit
y
b
a
se
d
a
p
p
ro
a
c
h
,
”
J
o
u
rn
a
l
o
f
K
in
g
S
a
u
d
U
n
ive
rs
it
y
-
Co
mp
u
ter
a
n
d
In
f
o
rm
a
ti
o
n
S
c
i
e
n
c
e
s
,
2
0
1
8
,
doi
:
1
0
.
1
0
1
6
/j
.
jk
su
c
i.
2
0
1
8
.
0
4
.
0
1
3
.
[4
]
S
.
A.
S
a
ji
d
h
a
,
K.
De
sik
a
n
,
a
n
d
S
.
P
.
Ch
o
d
n
e
k
a
r,
“
I
n
it
ial
se
e
d
a
lg
o
rit
h
m
fo
r
m
ix
e
d
d
a
ta
u
sin
g
m
o
d
ifi
e
d
K
-
m
e
a
n
s
c
lu
ste
rin
g
a
lg
o
rit
h
m
,
”
Ara
b
ia
n
J
o
u
rn
a
l
o
f
S
c
ien
c
e
a
n
d
E
n
g
i
n
e
e
rin
g
,
v
o
l
.
4
5
,
p
p
.
2
6
8
5
-
2
7
0
3
,
2
0
2
0
,
doi
:
1
0
.
1
0
0
7
/s1
3
3
6
9
-
0
1
9
-
0
4
1
2
1
-
0
.
[5
]
M
.
M
.
B
re
u
n
i
g
,
H.
P
.
Krie
g
e
l,
R.
T.
Ng
,
a
n
d
J.
S
a
n
d
e
r,
“
LOF
:
Id
e
n
ti
fy
i
n
g
d
e
n
sit
y
b
a
se
d
l
o
c
a
l
o
u
tl
iers
,
”
In
t
h
e
Pro
c
e
e
d
in
g
s
o
f
t
h
e
2
0
0
0
AC
M
S
IGM
OID
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
M
a
n
a
g
e
me
n
t
o
f
D
a
ta
,
2
0
0
0
,
p
p
.
93
-
1
0
3
,
doi
:
1
0
.
1
1
4
5
/3
3
5
1
9
1
.
3
3
5
3
8
8
.
[6
]
V.
Ha
u
tam
ä
k
i,
S
.
C
h
e
re
d
n
ich
e
n
k
o
,
I
.
Kä
rk
k
ä
i
n
e
n
,
T
.
Kin
n
u
n
e
n
,
a
n
d
P
.
F
rä
n
ti
,
“
Im
p
ro
v
in
g
k
-
m
e
a
n
s
b
y
o
u
tl
ier
re
m
o
v
a
l
,
”
S
c
a
n
d
i
n
a
v
i
a
n
C
o
n
fer
e
n
c
e
o
n
Ima
g
e
An
a
lys
is
,
S
p
ri
n
g
e
r,
B
e
rli
n
,
He
id
e
l
b
e
rg
,
2
0
0
5
,
p
p
.
9
7
8
-
9
8
7
.
[7
]
Y.
Zh
o
u
,
H.
Y
u
a
n
d
X.
Ca
i,
“
A
n
o
v
e
l
k
-
m
e
a
n
s
a
lg
o
rit
h
m
f
o
r
c
l
u
ste
rin
g
a
n
d
o
u
tl
ier
d
e
tec
ti
o
n
,
”
2
nd
IEE
E
in
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
F
u
t
u
re
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
a
n
d
M
a
n
a
g
e
me
n
t
En
g
in
e
e
rin
g
,
2
0
0
9
,
p
p
.
4
7
6
-
4
8
0
,
doi
:
1
0
.
1
1
0
9
/F
IT
M
E.
2
0
0
9
.
1
2
5
.
[8
]
B.
Tan
g
a
n
d
H.
He
,
“
A
lo
c
a
l
d
e
n
sity
-
b
a
se
d
a
p
p
ro
a
c
h
f
o
r
o
u
tl
ier
d
e
tec
ti
o
n
,
”
Ne
u
r
o
c
o
mp
u
ti
n
g
,
v
o
l
.
2
4
1
.
pp.
1
7
1
-
1
8
0
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j.
n
e
u
c
o
m
.
2
0
1
7
.
0
2
.
0
3
9
.
[9
]
E.
S
c
h
u
b
e
rt
Zi
m
e
k
a
n
d
H.
P
.
Kri
e
g
e
l,
“
A
su
rv
e
y
o
n
u
n
su
p
e
r
v
ise
d
o
u
tl
ier
d
e
tec
ti
o
n
i
n
h
i
g
h
d
ime
n
si
o
n
a
l
n
u
m
e
rica
l
d
a
ta
,
”
S
ta
t
isti
c
a
l
A
n
a
lys
is
a
n
d
D
a
ta
M
in
in
g
:
T
h
e
AS
A
D
a
ta
S
c
ie
n
c
e
J
o
u
rn
a
l
,
v
o
l
.
5
,
n
o.
5
.
p
p
.
363
-
3
8
7
,
2
0
1
2
,
doi
:
1
0
.
1
0
0
2
/sa
m
.
1
1
1
6
1
.
[1
0
]
H.
P
.
Krie
g
e
l,
M
.
S
c
h
u
b
e
rt
,
a
n
d
A.
Zi
m
e
k
,
“
An
g
le
-
b
a
se
d
o
u
tl
ier
d
e
te
c
ti
o
n
i
n
h
i
g
h
d
ime
n
si
o
n
a
l
d
a
t
a
,
”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
1
4
th
AC
M
S
IGKD
D
In
te
rn
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
K
n
o
wl
e
d
g
e
Disc
o
v
e
ry
a
n
d
Da
t
a
M
in
i
n
g
2
0
0
8
,
2
0
0
8
,
p
p
.
4
4
4
-
4
5
2
,
d
o
i
:
1
0
.
1
1
4
5
/1
4
0
1
8
9
0
.
1
4
0
1
9
4
6
.
[1
1
]
T.
T
.
Da
n
g
,
H.
W.
Ng
a
n
,
a
n
d
W.
Li
u
,
“
Dista
n
c
e
b
a
se
d
k
-
n
e
a
re
st
n
e
ig
h
b
o
rs
o
u
tl
ier
d
e
tec
ti
o
n
m
e
th
o
d
i
n
larg
e
sc
a
le
traff
ic
d
a
ta
,
”
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Dig
it
a
l
S
ig
n
a
l
Pr
o
c
e
ss
in
g
,
2
0
1
5
,
p
p
.
5
0
7
-
5
1
0
,
doi
:
1
0
.
1
1
0
9
/ICDS
P
.
2
0
1
5
.
7
2
5
1
9
2
4
.
[1
2
]
H.
P
.
Krie
g
e
l,
P
.
Krö
g
e
r,
E
.
S
c
h
u
b
e
rt
,
a
n
d
A.
Zi
m
e
k
,
“
Lo
OP
:
L
o
c
a
l
o
u
tl
ier
p
r
o
b
a
b
i
li
ti
e
s
,
”
In
Pro
c
e
e
d
in
g
s
Of
T
h
e
18
th
AC
M
Co
n
fer
e
n
c
e
o
n
In
fo
rm
a
ti
o
n
a
n
d
Kn
o
wled
g
e
M
a
n
a
g
e
me
n
t
,
p
p
.
5
0
7
-
5
1
0
,
2
0
0
9
,
doi
:
1
0
.
1
1
4
5
/1
6
4
5
9
5
3
.
1
6
4
6
1
9
5
.
[1
3
]
H.
P
.
Krie
g
e
l,
E.
S
c
h
u
b
e
rt
,
a
n
d
A.
Zi
m
e
k
,
“
Ou
tl
ier
d
e
tec
ti
o
n
i
n
a
x
is
p
a
ra
ll
e
l
su
b
sp
a
c
e
s
o
f
h
i
g
h
d
i
m
e
n
sio
n
a
l
d
a
ta
,
”
In
P
a
c
if
ic
-
Asia
Co
n
fer
e
n
c
e
o
n
Kn
o
wled
g
e
Disc
o
v
e
ry
a
n
d
Da
t
a
M
in
i
n
g
,
S
p
r
in
g
e
r,
Be
rli
n
,
He
id
e
lb
e
rg
,
2
0
0
9
,
pp.
8
3
1
-
838
,
d
o
i
:
1
0
.
1
0
0
7
/9
7
8
-
3
-
642
-
0
1
3
0
7
-
2
_
8
6
.
[1
4
]
F
.
Ke
ll
e
r,
E.
M
u
ll
e
r
,
a
n
d
K.
Bo
h
m
,
“
Hig
h
c
o
n
tras
t
su
b
s
p
a
c
e
fo
r
d
e
n
sit
y
b
a
se
d
o
u
tl
ier
ra
n
k
i
n
g
,
”
28
th
IEE
E
Nter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Da
t
a
En
g
in
e
e
rin
g
,
p
p
.
1
0
3
7
-
1
0
4
8
,
2
0
1
2
,
d
o
i
:
1
0
.
1
1
0
9
/ICDE.
2
0
1
2
.
8
8
.
[1
5
]
H.
P
.
Krie
g
e
l,
P
.
Kr
ö
g
e
r,
E
.
S
c
h
u
b
e
rt
,
a
n
d
A.
Zi
m
e
k
,
“
Ou
tl
ier
d
e
tec
ti
o
n
i
n
th
e
a
x
is
p
a
ra
ll
e
l
su
b
sp
a
c
e
s
o
f
h
ig
h
d
ime
n
sio
n
a
l
d
a
ta
,
”
In
P
ro
c
e
e
d
in
g
s o
f
PA
KDD
,
p
p
.
8
3
1
-
8
3
8
,
2
0
0
9
,
doi
:
1
0
.
1
0
0
7
/9
7
8
-
3
-
6
4
2
-
0
1
3
0
7
-
2
_
8
6
.
[1
6
]
G
.
Re
k
h
a
,
V.
K.
Re
d
d
y
,
a
n
d
A.
K.
Ty
a
g
i,
“
Cir
u
s
-
c
rit
ica
l
in
sta
n
c
e
s
,
re
m
o
v
a
l
b
a
se
d
u
n
d
e
r
sa
m
p
li
n
g
-
A
so
l
u
ti
o
n
fo
r
c
las
s imb
a
lan
c
e
,
”
IJ
HIS
, v
o
l.
1
6
,
n
o.
2
,
p
p
.
55
-
6
6
,
2
0
2
0
,
d
o
i
:
1
0
.
3
2
3
3
/HIS
-
2
0
0
2
7
9
.
[1
7
]
G
.
Re
k
h
a
,
V.
K.
Re
d
d
y
,
a
n
d
A
.
K.
Ty
a
g
i,
“
An
e
a
rth
m
o
v
e
r’s
d
ist
a
n
c
e
b
a
se
d
u
n
d
e
r
sa
m
p
li
n
g
a
p
p
r
o
a
c
h
fo
r
h
a
n
d
li
n
g
c
las
s
-
imb
a
lan
c
e
d
d
a
ta
,
”
In
ter
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
I
n
telli
g
e
n
t
I
n
fo
r
ma
ti
o
n
a
nd
D
a
ta
b
a
se
S
y
ste
ms
,
v
o
l.
1
3
,
n
o.
2
/
3
/
4
,
2
0
2
0
,
d
o
i
:
1
0
.
1
5
0
4
/IJIIDS
.
2
0
2
0
.
1
0
9
4
6
3
.
[1
8
]
X.
Xu
,
H.
Li
u
,
L.
Li
,
a
n
d
M
.
Ya
o
,
“
A
c
o
m
p
a
riso
n
o
f
o
u
tl
ier
d
e
tec
ti
o
n
tec
h
n
iq
u
e
s
fo
r
h
ig
h
d
i
m
e
n
sio
n
a
l
d
a
ta
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
o
f
Co
mp
u
ta
ti
o
n
a
l
In
telli
g
e
n
c
e
S
y
ste
ms
,
v
o
l
.
1
1
,
n
o.
1
,
p
p
.
6
5
2
-
6
6
2
,
2
0
1
8
,
doi
:
1
0
.
2
9
9
1
/i
jcis.
1
1
.
1
.
5
0
.
[1
9
]
Y.
Wei,
J.
Ja
n
g
-
Ja
c
c
a
rd
,
F
.
S
a
b
ri
n
a
,
a
n
d
T.
M
c
In
to
sh
,
"
M
S
D
-
k
m
e
a
n
s:
A
n
o
v
e
l
a
lg
o
rit
h
m
fo
r
e
ffici
e
n
t
d
e
tec
ti
o
n
o
f
g
lo
b
a
l
a
n
d
lo
c
a
l
o
u
t
li
e
rs,
"
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
,
a
rX
iv p
re
p
rin
t
a
rX
iv: 1
9
1
0
.
0
6
5
8
8
,
2
0
1
9
.
[2
0
]
S
.
Ka
n
jan
a
wa
tt
a
n
a
,
"
A
n
o
v
e
l
o
u
t
li
e
r
d
e
tec
ti
o
n
a
p
p
li
e
d
t
o
a
n
a
d
a
p
t
iv
e
k
-
m
e
a
n
s,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
M
a
c
h
i
n
e
L
e
a
rn
in
g
a
n
d
Co
m
p
u
t
in
g
,
v
ol
.
9,
n
o.
5
,
p
p
.
5
6
9
-
5
7
4
,
2
0
1
9
,
d
o
i
:
1
0
.
1
8
1
7
8
/i
jmlc
.
2
0
1
9
.
9
.
5
.
8
4
1
.
[2
1
]
A.
Ch
a
tt
e
rjee
S
a
h
a
,
S
.
G
h
o
sh
,
N
.
Ku
m
a
r
,
a
n
d
R.
S
a
r
k
a
r
,
"
An
e
n
s
e
m
b
le
a
p
p
ro
a
c
h
to
o
u
tl
ier
d
e
tec
ti
o
n
u
si
n
g
s
o
m
e
c
o
n
v
e
n
ti
o
n
a
l
c
lu
ste
rin
g
a
lg
o
rit
h
m
s
,
"
M
u
lt
ime
d
i
a
T
o
o
ls
a
n
d
Ap
p
li
c
a
t
io
n
s
,
p
p
.
1
-
2
5
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
0
7
/s1
1
0
4
2
-
0
2
0
-
0
9
6
2
8
-
5
.
[2
2
]
P
.
Ya
n
g
,
D.
Wan
g
,
Z.
Wei,
X.
Du
a
n
d
T.
Li
,
"
A
n
o
u
tl
ier
d
e
tec
t
io
n
a
p
p
ro
a
c
h
b
a
se
d
o
n
imp
ro
v
e
d
se
lf
-
o
rg
a
n
izin
g
f
e
a
tu
re
m
a
p
c
lu
ste
rin
g
a
lg
o
rit
h
m
,
"
IE
EE
Acc
e
ss
,
v
o
l
.
7
,
p
p
.
1
1
5
9
1
4
-
1
1
5
9
2
5
,
2
0
1
9
,
doi
:
1
0
.
1
1
0
9
/ACCES
S
.
2
0
1
9
.
2
9
2
2
0
0
4
.
[2
3
]
J.
Ya
n
g
,
S
.
Ra
h
a
r
d
ja
a
n
d
P
.
F
rä
n
ti
,
"
Ou
tl
ier
d
e
tec
ti
o
n
:
h
o
w
to
t
h
re
sh
o
l
d
o
u
t
li
e
r
sc
o
re
s?
,
"
In
Pro
c
e
e
d
in
g
s
o
f
th
e
In
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
Arti
fi
c
ia
l
In
te
ll
i
g
e
n
c
e
,
I
n
f
o
rm
a
ti
o
n
P
ro
c
e
ss
in
g
a
n
d
C
lo
u
d
Co
mp
u
ti
n
g
,
2
0
1
9
,
p
p
.
1
-
6,
doi
:
1
0
.
1
1
4
5
/3
3
7
1
4
2
5
.
3
3
7
1
4
2
7
.
[2
4
]
EL
KI.
[On
li
n
e
]
.
Av
a
il
a
b
le:
h
tt
p
s:/
/elk
i
-
p
r
o
jec
t.
g
i
th
u
b
.
i
o
/
d
a
tas
e
ts/o
u
tl
ier
[2
5
]
UCI rep
o
sito
r
y
.
[O
n
li
n
e
]
.
Av
a
il
a
b
le:
h
tt
p
s://
a
rc
h
iv
e
.
ics
.
u
c
i.
e
d
u
/ml/i
n
d
e
x
.
p
h
p
Evaluation Warning : The document was created with Spire.PDF for Python.