I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI)
Vo
l.
9
,
No
.
1
,
Ma
r
ch
2
0
2
0
,
p
p
.
1
~
10
I
SS
N:
2
2
5
2
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
9
.i
1
.
p
p
1
-
10
ļ²
1
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
A f
u
zz
y
neighbo
r
ho
o
d roug
h set
metho
d f
o
r
ano
m
a
l
y
det
ec
tion
in larg
e sca
le dat
a
E
l
M
ez
ia
t
i
M
a
ro
ua
ne
1
,
Z
iy
a
t
i El
ho
u
s
s
a
ine
2
1
A
s
so
c
iate
d
m
e
m
b
e
r
L
RI
T
Un
iv
e
rsity
M
o
h
a
m
m
e
d
V
,
Ra
b
a
t,
M
o
r
o
c
c
o
2
A
s
so
c
iate
d
m
e
m
b
e
r
L
RI
T
Un
iv
e
rsity
H
a
ss
a
n
2
ES
T
C,
RIT
M
LA
B
,
Ra
b
a
t,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
2
0
,
2
0
1
9
R
ev
i
s
ed
Dec
2
5
,
2
0
1
9
A
cc
ep
ted
J
an
6
,
2
0
2
0
Re
c
e
n
t
re
se
a
r
c
h
stu
d
ies
o
n
o
u
tl
ie
r
d
e
tec
ti
o
n
h
a
v
e
f
o
c
u
se
d
o
n
e
x
a
m
in
in
g
th
e
n
e
a
re
st
n
e
ig
h
b
o
r
str
u
c
tu
re
o
f
a
d
a
ta
o
b
jec
t
to
m
e
a
su
re
it
s
o
u
tl
iern
e
ss
d
e
g
re
e
.
M
o
re
o
v
e
r,
p
o
p
u
lar
o
u
tl
ier d
e
tec
ti
o
n
m
e
th
o
d
s
re
q
u
ire
t
h
e
p
a
irw
ise
c
o
m
p
a
riso
n
o
f
o
b
jec
ts
t
o
c
o
m
p
u
te
th
e
n
e
a
re
st
n
e
ig
h
b
o
rs.
T
h
is
q
u
a
d
ra
ti
c
p
ro
b
l
e
m
is
n
o
t
sc
a
lab
le
to
larg
e
d
a
ta
se
ts,
m
a
k
in
g
m
u
lt
id
im
e
n
sio
n
a
l
o
u
tl
ier
d
e
tec
ti
o
n
f
o
r
b
ig
d
a
ta
stil
l
a
n
o
p
e
n
c
h
a
ll
e
n
g
e
.
In
th
is
a
rti
c
le,
w
e
p
re
s
e
n
t
a
n
e
w
a
p
p
ro
a
c
h
f
o
r
o
u
tl
ier
d
e
tec
ti
o
n
,
b
a
se
d
o
n
h
ig
h
ly
sc
a
lab
le
a
p
p
ro
a
c
h
to
c
o
m
p
u
te
th
e
n
e
a
re
st
n
e
ig
h
b
o
rs
o
f
o
b
jec
ts
u
sin
g
f
u
z
z
y
r
o
u
g
h
se
t
th
e
o
ry
.
A
t
th
e
sa
m
e
ti
m
e
,
th
e
o
u
tl
ier
ra
n
k
in
g
p
ro
c
e
ss
is
a
c
c
e
lera
ted
b
y
u
sin
g
a
h
ig
h
-
p
e
rf
o
rm
a
n
c
e
a
n
d
a
p
a
ra
ll
e
l
c
o
m
p
u
tatin
g
u
sin
g
m
a
p
re
d
u
c
e
f
ra
m
e
w
o
rk
.
K
ey
w
o
r
d
s
:
B
ig
d
ata
C
lu
s
ter
i
n
g
Fu
zz
y
n
ei
g
h
b
o
r
h
o
o
d
Ma
p
R
ed
u
ce
Ou
tlier
d
etec
tio
n
R
o
u
g
h
s
et
T
im
e
co
m
p
le
x
it
y
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
:
E
l M
ez
iati M
ar
o
u
an
e
,
Ass
o
ciate
d
m
e
m
b
er
L
R
I
T
,
Un
i
v
er
s
it
y
Mo
h
a
m
m
ed
V,
R
a
b
at,
Mo
r
o
cc
o
.
E
m
ail:
el
m
ez
iati.
m
ar
o
u
a
n
e
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Ou
tlier
s
ar
e
th
e
u
n
u
s
u
al,
u
n
e
x
p
ec
ted
p
atter
n
s
in
th
e
o
b
s
er
v
e
d
w
o
r
ld
.
Ou
tlier
s
ex
i
s
t e
x
te
n
s
i
v
el
y
i
n
r
ea
l
w
o
r
ld
,
an
d
t
h
e
y
ar
e
g
e
n
er
ated
f
r
o
m
d
i
f
f
er
en
t
s
o
u
r
ce
s
:
a
h
ea
v
il
y
tailed
d
is
tr
ib
u
tio
n
o
r
er
r
o
r
s
in
i
n
p
u
tt
in
g
t
h
e
d
ata.
W
h
ile
th
er
e
is
n
o
s
in
g
le,
g
en
e
r
all
y
ac
ce
p
ted
,
f
o
r
m
al
d
ef
in
iti
o
n
o
f
an
o
u
tlier
,
Ha
w
k
i
n
s
ā
d
ef
in
itio
n
ca
p
tu
r
es
th
e
s
p
ir
it:
ā
an
o
u
tlier
i
s
an
o
b
s
er
v
atio
n
t
h
at
d
ev
iate
s
s
o
m
u
ch
f
r
o
m
o
th
er
o
b
s
er
v
atio
n
s
a
s
to
ar
o
u
s
e
s
u
s
p
icio
n
s
t
h
at
it
w
a
s
g
e
n
er
ated
b
y
a
d
if
f
er
e
n
t
m
ec
h
a
n
is
m
ā
[
1
]
.
A
n
o
m
al
y
d
etec
tio
n
is
an
i
m
p
o
r
tan
t
p
r
o
b
le
m
th
at
h
as
b
ee
n
r
esear
ch
ed
w
it
h
i
n
d
iv
er
s
e
r
esear
ch
ar
ea
s
an
d
ap
p
licatio
n
d
o
m
ain
s
s
u
c
h
as
f
r
au
d
d
etec
tio
n
[
2
]
,
in
tr
u
s
io
n
d
is
co
v
e
ry
[
3
]
,
v
id
eo
s
u
r
v
eilla
n
ce
,
p
h
ar
m
ac
eu
tical
tes
t
an
d
w
e
ath
er
p
r
ed
ictio
n
.
T
h
er
e
ar
e
d
if
f
er
en
t
s
u
r
v
e
y
s
ab
o
u
t
class
ical
o
u
tlier
s
an
d
ab
n
o
r
m
al
d
etec
tio
.
T
h
ey
v
ar
y
b
et
w
e
en
d
en
s
it
y
b
ased
ap
p
r
o
ac
h
es
[
3
]
,
s
tatis
tical
[
4
]
,
d
is
tan
ce
-
b
ased
[
5
]
,
n
eu
r
al
n
et
w
o
r
k
s
an
d
m
ac
h
i
n
e
lear
n
in
g
te
ch
n
iq
u
es.
R
ec
en
t
r
esear
ch
s
t
u
d
ies
o
n
o
u
t
lier
d
etec
tio
n
h
av
e
f
o
c
u
s
ed
o
n
ex
a
m
in
i
n
g
t
h
e
n
ea
r
est
n
ei
g
h
b
o
r
s
tr
u
ct
u
r
e
o
f
a
d
ata
o
b
j
ec
t
to
m
ea
s
u
r
e
its
o
u
tlier
n
e
s
s
d
eg
r
ee
[
6
-
7]
.
S
u
c
h
tec
h
n
iq
u
es
ar
e
b
ased
o
n
t
h
e
k
e
y
a
s
s
u
m
p
tio
n
t
h
at
i
n
s
ta
n
ce
s
o
f
n
o
r
m
al
d
ata
o
cc
u
r
in
d
e
n
s
e
n
ei
g
h
b
o
r
h
o
o
d
s
,
w
h
ile
o
u
tlier
s
o
cc
u
r
f
ar
a
w
a
y
f
r
o
m
t
h
eir
clo
s
est
n
eig
h
b
o
r
s
[
8
]
.
P
o
p
u
lar
o
u
tlier
d
etec
tio
n
m
et
h
o
d
s
r
eq
u
ir
e
t
h
e
p
air
w
i
s
e
co
m
p
ar
is
o
n
o
f
o
b
jects
to
co
m
p
u
te
t
h
e
n
ea
r
est
n
ei
g
h
b
o
r
s
.
T
h
is
q
u
ad
r
atic
p
r
o
b
lem
is
n
o
t
s
ca
lab
le
to
l
ar
g
e
d
ata
s
ets,
m
a
k
i
n
g
o
u
tl
ie
r
d
etec
tio
n
f
o
r
lar
g
e
s
ca
le
d
ata
s
til
l
an
o
p
en
ch
al
len
g
e.
T
h
is
p
ap
er
p
r
o
p
o
s
es
a
f
ast
o
u
tlier
d
etec
tio
n
m
et
h
o
d
f
o
r
la
r
g
e
s
ca
le
d
atase
ts
,
w
h
ic
h
co
n
s
is
t
s
o
f
t
w
o
s
tep
s
:
a
g
r
an
u
latio
n
o
f
th
e
u
n
i
v
er
s
e
i
n
t
o
p
ar
ts
w
i
th
th
e
s
a
m
e
p
r
o
p
er
t
ies
th
e
n
th
e
co
m
p
u
t
in
g
o
f
th
e
d
e
g
r
ee
o
f
o
u
tlier
n
es
s
c
alled
Fu
zz
y
n
ei
g
h
b
o
r
h
o
o
d
r
o
u
g
h
s
e
t
o
u
tlier
f
ac
to
r
(
FNR
O
F)
f
o
r
ea
ch
g
r
an
u
le
f
o
r
m
ed
.
Gr
an
u
latio
n
o
f
t
h
e
o
b
esev
ab
le
u
n
iv
er
s
e
in
v
o
lv
e
s
g
r
o
u
p
in
g
o
f
s
i
m
ilar
ele
m
en
ts
in
to
g
r
an
u
le
s
.
W
ith
g
r
an
u
lated
v
ie
w
s
,
w
e
d
e
al
w
i
t
h
ap
p
r
o
x
i
m
a
tio
n
s
o
f
co
n
ce
p
ts
,
r
ep
r
esen
ted
b
y
s
u
b
s
et
s
o
f
t
h
e
u
n
iv
er
s
e,
i
n
ter
m
s
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
9
,
No
.
1
,
Ma
r
ch
20
20
:
1
ā
10
2
g
r
an
u
les
[
9
].
T
h
e
r
em
ai
n
d
er
o
f
th
i
s
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
.
I
n
t
h
e
n
e
x
t
s
ec
t
io
n
,
w
e
p
r
esen
t
s
o
m
e
p
r
eli
m
in
ar
ie
s
o
f
r
o
u
g
h
s
e
t th
eo
r
y
t
h
at
ar
e
r
elev
an
t to
th
is
p
ap
e
r
an
d
d
is
cu
s
s
io
n
o
f
th
e
g
r
an
u
l
ar
it
y
o
f
k
n
o
w
led
g
e
in
co
n
n
ec
tio
n
w
i
th
r
o
u
g
h
a
n
d
f
u
zz
y
s
et
s
.
I
n
Sectio
n
3
,
w
e
p
r
o
p
o
s
e
an
e
f
f
icien
t
p
ar
allel
co
m
p
u
ti
n
g
s
y
s
te
m
b
ased
o
n
Ma
p
R
ed
u
ce
in
o
r
d
er
to
i
m
p
r
o
v
e
th
e
s
p
ee
d
o
f
co
m
p
u
tat
io
n
an
d
t
h
e
al
g
o
r
ith
m
p
r
o
p
o
s
ed
th
at
d
ea
l
w
it
h
m
o
r
e
co
m
p
le
x
o
u
tl
ier
d
etec
tio
n
p
r
o
b
le
m
s
f
o
r
lar
g
e
s
ca
le
d
ata.
2.
RO
UG
H
S
E
T
S (
RST
)
R
o
u
g
h
s
e
t
th
eo
r
y
R
ST
[
10
-
1
1
]
is
a
n
e
w
m
at
h
e
m
atica
l
ap
p
r
o
ac
h
to
i
m
p
er
f
ec
t
k
n
o
w
led
g
e.
T
h
e
th
eo
r
y
h
as a
t
tr
ac
ted
atten
t
io
n
o
f
m
an
y
r
esear
ch
er
s
an
d
p
r
ac
titi
o
n
er
s
all
o
v
er
t
h
e
w
o
r
ld
,
w
h
o
co
n
tr
i
b
u
ted
ess
e
n
tiall
y
to
its
d
ev
elo
p
m
e
n
t
a
n
d
ap
p
licatio
n
s
.
T
h
e
m
ai
n
ad
v
an
tag
e
o
f
r
o
u
g
h
s
et
t
h
eo
r
y
i
n
d
ata
an
al
y
s
is
is
t
h
at
it
d
o
es
n
o
t
n
ee
d
an
y
p
r
eli
m
i
n
ar
y
o
r
ad
d
iti
o
n
al
i
n
f
o
r
m
atio
n
ab
o
u
t
d
ata.
R
o
u
g
h
s
e
t
t
h
eo
r
y
i
s
a
p
o
p
u
lar
a
n
d
p
o
w
er
f
u
l
m
ac
h
in
e
lear
n
in
g
to
o
l.
I
t
is
esp
ec
ially
s
u
itab
le
f
o
r
d
ea
lin
g
w
it
h
in
f
o
r
m
a
tio
n
s
y
s
te
m
s
th
at
ex
h
ib
it
in
co
n
s
is
te
n
cies.
I
n
r
o
u
g
h
s
et
th
eo
r
y
,
an
i
n
f
o
r
m
atio
n
tab
l
e
is
d
ef
in
ed
as
a
tu
p
le
T
=
(
U,
A
)
w
h
er
e
U
an
d
A
ar
e
t
w
o
f
i
n
ite,
n
o
n
-
e
m
p
t
y
s
ets
w
it
h
U
th
e
u
n
iv
er
s
e
o
f
p
r
i
m
iti
v
e
o
b
j
ec
ts
an
d
A
th
e
s
et
o
f
attr
ib
u
tes.
E
ac
h
attr
ib
u
te
o
r
f
ea
t
u
r
e
a
ā
A
is
ass
o
ciate
d
w
it
h
a
s
e
t
V
a
o
f
it
s
v
a
lu
e,
ca
ll
ed
th
e
d
o
m
ain
o
f
a.
W
e
m
a
y
p
ar
titi
o
n
th
e
a
ttrib
u
te
s
e
t
A
in
to
t
w
o
s
u
b
s
et
s
C
a
n
d
D,
ca
lled
co
n
d
itio
n
an
d
d
ec
is
io
n
attr
ib
u
tes,
r
esp
ec
ti
v
el
y
.
L
et
P
ā
A
b
e
a
s
u
b
s
et
o
f
attr
ib
u
te
s
.
T
h
e
in
d
is
ce
r
n
ib
il
it
y
r
elatio
n
,
d
en
o
ted
b
y
:
I
ND
(
ī²
)
=
{
(
ī
,
ī
)
ā
ī·
2
/
ā
ī½
ā
ī²
,
ī½
(
ī
)
=
ī½
(
ī
)
}
(
1
)
W
h
er
e
a(
x
)
d
en
o
tes th
e
v
alu
e
o
f
f
ea
t
u
r
e
o
f
o
b
j
ec
t x
.
I
f
(
x
,
y
)
ā
I
ND
(
P
)
,
x
a
n
d
y
ar
e
s
aid
to
b
e
in
d
i
s
ce
r
n
ib
le
w
i
th
r
esp
ec
t
to
P
.
T
h
e
f
a
m
il
y
o
f
all
eq
u
iv
ale
n
ce
cla
s
s
e
s
o
f
I
ND
(
P
)
,
r
ef
er
r
in
g
to
a
p
ar
ti
tio
n
o
f
U
d
eter
m
in
ed
b
y
P
,
is
d
en
o
ted
b
y
U/I
ND(
P
)
.
E
ac
h
el
e
m
en
t
i
n
U/I
ND
(
P
)
is
a
s
et
o
f
in
d
is
ce
r
n
ib
le
o
b
j
ec
ts
w
it
h
r
esp
ec
t
to
P
.
T
h
e
f
a
m
il
y
o
f
all
eq
u
i
v
ale
n
ce
clas
s
es
o
f
I
ND
(
P
)
,
r
ef
er
r
in
g
to
a
p
ar
titi
o
n
o
f
U
d
eter
m
i
n
ed
b
y
P
,
is
d
en
o
ted
b
y
U/I
ND
(
P
)
.
W
h
er
e
ī£
ā
ī¤
=
{
īŗ
ā©
Y/X
ā
A
,
Y
ā
B
,
X
ā©
ī»
ā
ā
}
(
2
)
Fo
r
an
y
co
n
ce
p
t
X
ā
U,
X
co
u
ld
b
e
ap
p
r
o
x
i
m
ated
b
y
th
e
P
-
lo
w
er
ap
p
r
o
x
i
m
atio
n
an
d
P
-
u
p
p
er
a
p
p
r
o
x
i
m
atio
n
u
s
i
n
g
th
e
k
n
o
w
led
g
e
o
f
P
.
T
h
e
lo
w
er
ap
p
r
o
x
i
m
atio
n
o
f
X
is
t
h
e
s
et
o
f
o
b
j
ec
ts
o
f
U
th
at
ar
e
s
u
r
el
y
in
X
:
ī²
(
īŗ
)
=
ā
{
ī§
ā
U/I
ND
(
ī²
)
:
ī§
ā
īŗ
}
(
3
)
T
h
e
u
p
p
er
ap
p
r
o
x
im
a
tio
n
o
f
X
is
th
e
s
et
o
f
o
b
j
ec
ts
o
f
U
th
at
ar
e
p
o
s
s
ib
ly
i
n
X,
d
ef
i
n
ed
as:
ī²
(
īŗ
)
=
ā
{
ī§
ā
U/I
ND
(
ī²
)
:
ī§
ā©
īŗ
ā
ā
}
(
4
)
T
h
e
co
n
ce
p
t
d
ef
in
i
n
g
th
e
s
et
o
f
o
b
j
ec
ts
th
at
ca
n
p
o
s
s
ib
ly
,
b
u
t
n
o
t
ce
r
tain
l
y
,
b
e
class
i
f
ied
in
a
s
p
ec
if
ic
w
a
y
i
s
ca
lled
th
e
b
o
u
n
d
ar
y
r
eg
io
n
,
wh
ich
i
s
d
ef
i
n
ed
as:
B
N
(
ī²
)
=
ī²
(
īŗ
)
ā
ī²
(
īŗ
)
as sh
o
w
n
i
n
Fi
g
u
r
e
1
.
Fig
u
r
e
1
.
R
ep
r
esen
tatio
n
o
f
t
h
e
d
ata
p
ar
titi
o
n
in
g
f
o
r
a
s
u
b
s
et
X
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
ļ²
A
fu
z
z
y
n
eig
h
b
o
r
h
o
o
d
r
o
u
g
h
s
et
meth
o
d
fo
r
a
n
o
ma
ly
d
etec
ti
o
n
in
la
r
g
e
s
ca
le
d
a
ta
(
E
l Mezia
ti Ma
r
o
u
a
n
e
)
3
2
.
1
.
Ro
ug
h set
a
nd
f
uzzy
dis
cr
et
iza
t
io
n
T
h
e
ex
tr
ac
tio
n
o
f
k
n
o
w
led
g
e
f
r
o
m
a
h
u
g
e
v
o
lu
m
e
o
f
d
at
a
u
s
i
n
g
r
o
u
g
h
s
e
t
m
et
h
o
d
s
r
eq
u
ir
es
th
e
tr
an
s
f
o
r
m
atio
n
o
f
co
n
tin
u
o
u
s
v
alu
e
attr
ib
u
te
s
to
d
is
cr
ete
i
n
te
r
v
als,
i
n
o
r
d
er
to
f
o
r
m
a
g
r
id
s
tr
u
ct
u
r
e
an
d
t
h
e
n
f
o
r
m
cl
u
s
ter
s
f
r
o
m
t
h
e
ce
lls
i
n
th
e
g
r
id
s
tr
u
c
tu
r
e.
C
l
u
s
ter
s
co
r
r
esp
o
n
d
to
r
eg
io
n
s
th
at
ar
e
d
en
s
er
in
d
ata
p
o
in
t
s
th
an
t
h
eir
s
u
r
r
o
u
n
d
in
g
s
.
T
h
e
g
r
ea
t
ad
v
an
tag
e
o
f
g
r
id
-
b
as
ed
clu
s
ter
in
g
is
a
s
ig
n
i
f
ica
n
t
r
ed
u
ctio
n
in
ti
m
e
co
m
p
le
x
it
y
,
esp
ec
iall
y
f
o
r
v
er
y
lar
g
e
d
ata
s
et
s
.
T
h
e
co
n
ce
p
ts
o
f
r
ea
l
r
o
u
g
h
s
p
ac
e,
it
is
w
el
l
k
n
o
w
n
t
h
at
o
n
e
o
f
th
e
r
esear
ch
p
r
em
is
es
i
n
th
e
class
ical
r
o
u
g
h
s
ets
th
eo
r
y
is
th
e
in
f
o
r
m
a
tio
n
o
r
th
e
d
ata
to
b
e
d
is
cr
ete.
Dis
cr
etiza
tio
n
ca
n
b
e
v
ie
w
ed
as
a
d
ata
r
ed
u
ctio
n
tech
n
iq
u
e
w
h
ic
h
r
ed
u
ce
s
th
e
r
a
n
g
e
o
f
v
a
lu
es
o
f
a
co
n
tin
u
o
u
s
v
alu
e
s
attr
ib
u
te
in
to
a
m
i
n
i
m
u
m
n
u
m
b
er
o
f
d
i
s
cr
ete
in
ter
v
als.
T
h
e
n
u
m
b
er
s
o
f
c
u
t
-
p
o
in
t
s
ca
n
d
eter
m
in
e
t
h
e
lev
el
o
f
d
ata
r
ed
u
ct
io
n
.
T
h
e
f
e
w
er
t
h
e
n
u
m
b
er
o
f
cu
t
-
p
o
i
n
ts
th
e
m
o
r
e
t
h
e
d
ata
w
ill
b
e
r
e
d
u
ce
d
an
d
h
e
n
ce
a
g
en
er
alize
d
clas
s
i
f
ier
w
ill
b
e
p
o
s
s
ib
le.
T
h
e
ter
m
ā
c
u
t
-
p
o
in
tā
r
ef
er
s
to
a
r
ea
l
v
al
u
e
w
it
h
in
t
h
e
r
an
g
e
o
f
co
n
ti
n
u
o
u
s
v
alu
e
s
t
h
at
d
i
v
id
es
t
h
e
r
an
g
e
i
n
to
in
ter
v
al
s
.
C
u
t
-
p
o
in
t
i
s
also
k
n
o
w
n
as
s
p
lit
-
p
o
in
t.
T
h
e
g
r
ea
t
ad
v
an
ta
g
e
o
f
g
r
id
-
b
ased
clu
s
ter
in
g
is
a
s
ig
n
i
f
ica
n
t
r
ed
u
ctio
n
in
ti
m
e
co
m
p
lex
i
t
y
,
esp
ec
iall
y
f
o
r
v
er
y
lar
g
e
d
at
a
s
ets.
B
u
t
d
u
r
in
g
th
e
d
is
cr
etiza
tio
n
p
r
o
ce
s
s
,
if
t
h
e
d
is
cr
etiza
tio
n
i
s
to
o
r
o
u
g
h
,
m
u
c
h
u
s
ef
u
l
i
n
f
o
r
m
atio
n
m
a
y
b
e
lo
s
t.
A
n
d
i
f
th
e
d
is
cr
etiza
tio
n
is
to
o
ex
ac
t,
it
w
il
l
tak
e
a
lo
t
o
f
ti
m
e
co
m
p
le
x
it
y
.
So
,
it
ca
n
b
e
s
aid
th
at
th
e
d
is
ad
v
an
ta
g
es
o
f
class
ical
r
o
u
g
h
s
et
s
ar
e
to
o
m
u
c
h
d
ep
en
d
in
g
o
n
g
o
o
d
o
r
b
ad
o
f
th
e
d
is
cr
etiza
tio
n
m
e
t
h
o
d
s
an
d
th
e
li
m
i
ted
ap
p
licatio
n
d
o
m
ai
n
.
L
et
īŗ
=
(
ī
1
,
ī
2
,
.
.
,
ī
īÆ”
)
b
e
a
p
r
o
v
id
ed
d
ataset
h
av
i
n
g
n
o
b
j
ec
ts
an
d
A
attr
ib
u
tes,
ī
mi
nj
=m
in
(
ī
īÆ
)
,
v
maxj
=m
ax
(
ī
īÆ
)
b
e
th
e
m
in
i
m
u
m
a
n
d
m
a
x
i
m
u
m
v
al
u
es
o
f
attr
ib
u
tes
i.
E
ac
h
attr
ib
u
te
[
īø
mi
ni
,
V
m
a
xi
]
is
eq
u
all
y
d
iv
id
ed
in
to
M
in
ter
v
als
ī
īÆ
=
(
ī
maxi
-
v
mi
ni
)
/M
.
T
h
e
s
et
o
f
all
in
itial
in
ter
v
al
o
f
an
attr
ib
u
te
i
is
s
h
o
w
n
as:
ī«
ī
īīī
ī
īÆ
=
{
ī
īÆ
īÆ
īÆ”
īÆ
,
(
ī
īÆ
īÆ
īÆ”
īÆ
+
w
īÆ
)
,
(
ī
īÆ īÆ
īÆ”
īÆ
+2
*
w
īÆ
)
, ...,
v
īÆ īÆīÆ«
īÆ
}
2
.
2
.
F
uzzy
ro
ug
h set
s
Fu
zz
y
r
o
u
g
h
s
et
th
eo
r
y
e
x
ten
d
s
r
o
u
g
h
s
et
th
eo
r
y
to
d
ata
w
it
h
co
n
tin
u
o
u
s
attr
ib
u
tes,
a
n
d
d
etec
ts
d
eg
r
ee
s
o
f
in
co
n
s
i
s
ten
c
y
i
n
t
h
e
d
ata.
K
e
y
to
th
is
i
s
tu
r
n
i
n
g
th
e
i
n
d
is
ce
r
n
ib
ilit
y
r
elatio
n
i
n
to
a
g
r
ad
u
al
r
elatio
n
.
T
h
e
f
u
zz
y
s
et
is
ac
t
u
all
y
a
f
u
n
d
a
m
e
n
tall
y
b
r
o
ad
er
s
et
co
m
p
ar
ed
w
i
th
th
e
clas
s
ical
o
r
cr
is
p
s
et.
T
h
e
class
ical
s
et
o
n
l
y
co
n
s
id
er
s
a
li
m
ited
n
u
m
b
er
o
f
d
eg
r
ee
s
o
f
m
e
m
b
er
s
h
ip
s
u
c
h
as
ā
0
ā
o
r
ā
1
ā
,
o
r
a
r
an
g
e
o
f
d
ata
w
it
h
li
m
ited
d
eg
r
ee
s
o
f
m
e
m
b
er
s
h
ip
as
s
h
o
w
n
i
n
Fi
g
u
r
e
2
.
Def
i
n
itio
n
1
: (
Fu
zz
y
Sets
)
A
f
u
zz
y
s
et,
F,
d
ef
i
n
ed
o
v
er
u
n
iv
er
s
e
X
is
a
f
u
n
c
tio
n
d
ef
i
n
ed
as
:
īØ
=
{
(
ī
,
ī¤
(
ī
)
)
|
ī¤
(
ī
)
ā
[
0
,
1
]
,
ā
ī
ā
īŗ
}
(
5
)
Fu
n
ctio
n
ī¤
(
ī
)
is
ca
lled
t
h
e
m
e
m
b
e
r
s
h
ip
f
u
n
ctio
n
,
w
h
ic
h
m
ap
s
o
b
j
ec
t
x
to
th
e
m
e
m
b
er
s
h
ip
s
p
a
ce
.
T
h
e
r
o
u
g
h
m
e
m
b
er
s
h
ip
f
u
n
ctio
n
ex
p
r
es
s
e
s
co
n
d
itio
n
al
p
r
o
b
ab
ilit
y
th
at
x
b
elo
n
g
s
to
X
g
iv
en
P
an
d
ca
n
b
e
in
ter
p
r
eted
as
a
d
eg
r
ee
th
at
x
b
elo
n
g
s
to
X.
On
e
o
f
th
e
m
o
s
t
i
m
p
o
r
tan
t
c
o
n
c
ep
ts
in
f
u
zz
y
s
et
th
eo
r
y
a
n
d
ap
p
licatio
n
s
is
th
e
Ī±
-
cu
t
d
ec
o
m
p
o
s
itio
n
t
h
eo
r
e
m
d
ev
elo
p
ed
b
y
Z
ad
e
h
in
1
9
7
1
u
n
d
er
th
e
n
a
m
e
r
eso
lu
tio
n
id
en
tit
y
.
T
h
ese
cu
t
s
ar
e
cr
is
p
s
ets as
s
o
ciate
d
w
it
h
ce
r
ta
in
lev
e
ls
Ī±
th
at
r
ep
r
esen
t d
is
ti
n
ct
g
r
ad
es o
f
m
e
m
b
er
s
h
ip
.
Def
i
n
it
io
n
2
:
(
FS
Ī±
-
cu
t)
g
i
v
en
a
n
u
m
b
er
ī
ā
[
0
,
1
]
,
a
Ī±
-
c
u
t o
r
Ī±
-
lev
e
l se
t,
o
f
a
f
u
zz
y
s
e
t F
is
d
ef
in
ed
b
y
:
īØ
ī°
=
{
(
ī
,
ī¤
īÆ«
)
|
ī¤
īÆ«
ā„
ī
,
ā
ī
ā
[
0
,
1
]
}
ī ī
ī
0
<
ī
1
,
īØ
ī°
0
ā
īØ
ī°
1
(
6
)
Fig
u
r
e
2
.
(
A
lp
h
a,
B
eta)
-
c
u
ts
o
f
f
u
zz
y
s
et
F
W
e
d
ef
in
e
th
e
m
e
m
b
er
s
h
ip
f
u
n
ctio
n
o
f
t
h
e
I
n
ter
s
ec
tio
n
o
f
t
w
o
f
u
zz
y
s
ets
A
(
x
,
ī¤
ī®ŗ
ī“
(
ī
)
)
an
d
B
(
x
,
ī¤
ī®»
ī“
(
ī
)
)
as:
(
ī£
ā©
ī¤
)
ī°
=
(
ī
,
ī¤
ī®ŗ
ī“
ā©
ī®»
ī“
(
ī
)
=
1
2
ā
(
ī¤
ī®ŗ
ī“
(
ī
)
+
ī¤
ī®»
ī“
(
ī
)
)
)
x
ā
īŗ
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
9
,
No
.
1
,
Ma
r
ch
20
20
:
1
ā
10
4
2.
3
.
Ro
ug
h set
s
:
neig
hb
o
rho
o
d sy
s
t
e
m
s
T
h
e
co
n
ce
p
t
o
f
in
f
o
r
m
atio
n
g
r
an
u
la
tio
n
w
as
f
ir
s
t
i
n
tr
o
d
u
ce
d
b
y
Z
ad
eh
i
n
th
e
co
n
te
x
t
o
f
f
u
z
z
y
s
ets
i
n
1
9
7
9
[
1
2
]
.
T
h
e
b
asic
id
ea
s
o
f
i
n
f
o
r
m
atio
n
g
r
an
u
lat
io
n
h
a
v
e
ap
p
ea
r
ed
in
f
ie
ld
s
,
s
u
ch
a
s
in
ter
v
al
a
n
al
y
s
i
s
,
q
u
an
tizat
io
n
,
r
o
u
g
h
s
et
t
h
eo
r
y
an
d
m
a
n
y
o
t
h
er
s
.
T
h
er
e
is
a
f
ast
g
r
o
w
i
n
g
a
n
d
r
en
e
w
ed
i
n
t
e
r
est
in
t
h
e
s
tu
d
y
o
f
in
f
o
r
m
atio
n
g
r
an
u
latio
n
an
d
co
m
p
u
tatio
n
s
u
n
d
er
th
e
ter
m
o
f
Gr
an
u
lar
C
o
m
p
u
ti
n
g
(
Gr
C
)
.
[
1
3
]
Gr
an
u
latio
n
o
f
a
u
n
i
v
er
s
e
in
v
o
lv
e
s
g
r
o
u
p
in
g
o
f
s
i
m
ilar
ele
m
e
n
t
s
p
ar
ts
,
o
r
th
e
g
r
o
u
p
in
g
o
f
in
d
i
v
id
u
al
e
le
m
e
n
ts
o
r
o
b
j
ec
ts
in
to
a
f
a
m
il
y
o
f
d
is
j
o
i
n
t
s
u
b
s
et
s
,
b
ased
o
n
av
ailab
le
in
f
o
r
m
atio
n
an
d
k
n
o
w
led
g
e.
T
h
e
co
m
b
in
a
tio
n
o
f
to
p
o
lo
g
ical
s
p
ac
es
an
d
r
o
u
g
h
s
et
s
an
d
t
h
e
p
r
o
p
e
r
ties
o
f
to
p
o
lo
g
ical
r
o
u
g
h
s
p
ac
es
ar
e
d
is
cu
s
s
ed
[
1
4
]
u
s
ed
n
ei
g
h
b
o
r
h
o
o
d
s
y
s
te
m
s
a
n
d
to
p
o
lo
g
ical
co
n
c
ep
t
in
t
h
e
s
t
u
d
y
o
f
ap
p
r
o
x
i
m
atio
n
s
.
Ne
ig
h
b
o
r
h
o
o
d
s
y
s
te
m
is
a
m
a
th
e
m
atica
l
s
tr
u
ct
u
r
e
o
f
g
r
an
u
lar
co
m
p
u
t
i
n
g
to
m
o
d
el
g
r
an
u
le
s
,
an
d
ca
n
b
e
u
s
ed
to
co
m
p
u
te
s
tr
u
ctu
r
e
o
f
g
r
an
u
le
s
an
d
/o
r
b
et
w
ee
n
g
r
an
u
le
s
.
A
n
eig
h
b
o
r
h
o
o
d
s
y
s
te
m
at
a
p
o
in
t
is
a
f
r
am
e
w
o
r
k
to
ca
p
tu
r
e
th
e
co
n
ce
p
t
o
f
ān
ea
r
ā
o
b
j
ec
ts
,
an
d
an
y
s
u
b
s
et
o
f
o
b
j
ec
ts
ca
n
b
e
ap
p
r
o
x
i
m
ated
b
y
a
s
e
t
o
f
n
eig
h
b
o
r
h
o
o
d
s
.
A
n
ei
g
h
b
o
r
h
o
o
d
s
y
s
te
m
d
ef
in
e
s
a
s
et
o
f
b
in
ar
y
r
elat
io
n
s
,
a
n
d
a
s
et
o
f
b
in
ar
y
r
elatio
n
s
h
ip
s
ca
n
b
e
u
s
ed
to
d
ef
in
e
a
n
ei
g
h
b
o
r
h
o
o
d
s
y
s
te
m
.
Def
i
n
itio
n
3
(
n
eig
h
b
o
r
h
o
o
d
o
f
o
b
j
ec
t
x
i
)
:
Giv
e
n
a
n
ar
b
itra
r
y
ī
īÆ
ā
ī·
an
d
P
ā
ī„
,
t
h
e
n
ea
r
est
n
ei
g
h
b
o
r
h
o
o
d
ī
ī°”
īÆ
(
ī
īÆ
)
o
f
x
i
i
n
f
ea
t
u
r
e
s
p
ac
e
P
is
d
ef
i
n
ed
as:
ī
īƦ
īÆ
(
ī
īÆ
)
=
{
ī
īÆ
|
ī
īÆ
(
ī
īÆ
,
ī
īÆ
)
ā¤
ī
,
ī
ā
ā
+
}
(
7
)
W
h
er
e
ī
:
ī·
Ć
ī·
ā
ī“
+
,
a
d
is
tan
ce
(
s
i
m
ilar
it
y
)
f
u
n
ct
io
n
an
d
R
+
is
t
h
e
s
et
o
f
n
o
n
-
n
e
g
ati
v
e
r
ea
l
n
u
m
b
er
.
ī
īƦ
īÆ
:
T
h
e
n
eig
h
b
o
r
h
o
o
d
in
f
o
r
m
atio
n
g
r
an
u
le
i
n
cl
u
d
ed
o
b
j
ec
ts
x
i
a
n
d
th
e
s
ize
o
f
th
e
n
e
ig
h
b
o
r
h
o
o
d
d
ep
en
d
s
o
n
th
r
es
h
o
ld
ī
.
Fo
r
ea
ch
v
alu
e
o
f
ī
ā
ī“
+
,
w
e
p
r
o
p
o
s
e
th
e
f
o
llo
w
in
g
n
e
i
g
h
b
o
r
h
o
o
d
s
y
s
te
m
a
s
th
e
co
llect
io
n
o
f
all
n
eig
h
b
o
r
h
o
o
d
s
o
f
x
ā
U
as:
ī°
īƦ
īÆ
(
ī
)
=
{
ī
īƦ
īÆ
(
ī
)
|
ī
ā
ā
+
,
ī²
ā
ī„
}
(
8
)
W
h
er
e
s
is
a
s
lid
i
n
g
w
i
n
d
o
w
s
f
o
r
o
v
er
lap
p
in
g
co
m
p
u
tatio
n
:
s
<M
.
T
h
eo
r
em
1
:
Fo
r
ea
ch
ī²
1
ā
ī£
,
ī²
2
ā
ī£
:
ī°
īƦ
īÆ
(
ī
)
is
a
n
eig
h
b
o
r
h
o
o
d
r
elatio
n
in
d
u
ce
d
in
f
ea
tu
r
e
s
u
b
s
p
ac
e
P
.
W
e
h
av
e:
ī°
īƦ
īÆ
1
āŖ
īÆ
2
(
ī
)
=
ī°
īƦ
īÆ
1
(
ī
)
ā©
ī°
īƦ
īÆ
2
(
ī
)
ī ī
ī£
=
ā
ī²
īÆ
īÆ
s
o
ī°
īƦ
(
ī
)
=
ā
ī°
īƦ
īÆ
ī³
(
ī
)
īÆ
(
9
)
Giv
e
n
a
s
et
o
f
o
b
j
ec
ts
U
an
d
a
n
eig
h
b
o
r
h
o
o
d
s
y
s
te
m
N
s
o
v
er
U,
w
e
ca
ll
<
U,
N
s
>
a
n
ei
g
h
b
o
r
h
o
o
d
ap
p
r
o
x
i
m
atio
n
s
p
ac
e.
T
h
e
lo
w
er
an
d
u
p
p
er
ap
p
r
o
x
i
m
atio
n
s
(
ī°īŗ
,
ī°īŗ
)
o
f
X
in
<U
,
Ns>,
ar
e
d
ef
in
ed
as
:
()
()
s
N
x
X
N
X
N
x
s
ļ
ļ½
ī°īŗ
=
ā
ī°
īƦ
(
ī
)
īÆ
ī³
(
īÆ«
)
ā©
īÆ
ā
ā
Ob
v
io
u
s
l
y
,
ī°īŗ
ā
īŗ
ā
ī°īŗ
.
T
h
e
b
o
u
n
d
ar
y
r
eg
io
n
o
f
X
in
t
h
e
ap
p
r
o
x
i
m
at
io
n
s
p
ac
e
is
d
ef
i
n
ed
as:
ī¤
ī°īŗ
=
ī°īŗ
ā
ī°īŗ
.
T
h
e
s
ize
o
f
b
o
u
n
d
ar
y
r
e
g
io
n
r
ef
lects
t
h
e
d
eg
r
ee
o
f
r
o
u
g
h
n
ess
o
f
s
et
X
i
n
th
e
ap
p
r
o
x
i
m
a
tio
n
s
p
ac
e
<U
,
N
s
>.
Ass
u
m
in
g
X
is
th
e
s
a
m
p
le
s
u
b
s
et
w
it
h
a
d
ec
is
io
n
lab
el;
g
e
n
er
all
y
s
p
ea
k
in
g
,
w
e
h
o
p
e
th
e
b
o
u
n
d
ar
y
r
e
g
io
n
o
f
th
e
d
ec
is
io
n
s
h
o
u
ld
b
e
as s
m
a
l
l
as
p
o
s
s
ib
le
f
o
r
d
ec
r
ea
s
in
g
u
n
ce
r
tain
t
y
i
n
d
ec
i
s
io
n
.
T
h
e
s
ize
o
f
b
o
u
n
d
ar
y
r
e
g
io
n
d
ep
en
d
s
o
n
X,
attr
ib
u
tes to
d
escr
ib
e
U.
Fo
r
a
f
ix
ed
p
air
o
f
n
u
m
b
er
s
(
ī
0
,
ī
1
)
ā
[
0
,
1
]
Ć[
0
,
1
]
,
w
e
o
b
tain
a
s
u
b
m
o
d
el
i
n
w
h
ich
a
cr
i
s
p
s
et
īØ
ī°
is
ap
p
r
o
x
im
a
ted
in
a
cr
is
p
ap
p
r
o
x
im
a
tio
n
s
p
ac
e
ī½ī
ī
ā
ī°
0
=
(
ī·
,
ī¤
ā
ī°
0
)
T
h
e
r
esu
lt
is
a
r
o
u
g
h
s
et
(
(
ī½īī
ā
ī°
0
(
īØ
)
,
ī½īī
ā
ī°
0
(
īØ
)
)
w
it
h
t
h
e
r
ef
er
e
n
ce
s
et
īØ
.
E
ac
h
g
r
an
u
le
i
n
f
u
zz
y
s
ets
īØ
is
a
n
e
ig
h
b
o
r
h
o
o
d
o
f
an
ele
m
e
n
t
o
f
th
e
u
n
i
v
er
s
e.
T
h
e
ap
p
r
o
x
i
m
atio
n
is
d
ef
i
n
ed
b
y
s
h
o
w
in
Fi
g
u
r
e
3
:
ī°ī£
=
ā
ī°
īƦ
(
ī
)
īÆ
ī³
(
īÆ«
)
ā
ī®æ
ī“
1
,
f
o
r
F
ī°
1
ā
īØ
ī°
0
ā
ī·
(
1
0
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
ļ²
A
fu
z
z
y
n
eig
h
b
o
r
h
o
o
d
r
o
u
g
h
s
et
meth
o
d
fo
r
a
n
o
ma
ly
d
etec
ti
o
n
in
la
r
g
e
s
ca
le
d
a
ta
(
E
l Mezia
ti Ma
r
o
u
a
n
e
)
5
ī°ī£
=
ā
ī°
īƦ
(
ī
)
īÆ
ī³
(
īÆ«
)
ā
ī®æ
ī“
0
(
1
1
)
I
n
th
i
s
ca
s
e,
th
e
s
u
b
s
e
t
īØ
ī°
1
(
lo
w
er
ap
p
r
o
x
im
at
io
n
)
co
n
tai
n
s
t
w
o
clu
s
ter
s
C
1
(
g
r
id
2
)
an
d
C
2
(
g
r
id
3
)
īØ
ī°
1
=
ī„
1
āŖ
ī„
2
.
Fig
u
r
e
3
.
Fu
zz
y
r
o
u
g
h
s
et
ap
p
r
o
x
i
m
at
io
n
T
h
e
r
o
o
t g
r
id
Gr
id
0
(
ī
ī
ī
īī
ī
ī
ī
ī·
)
w
it
h
t
h
e
co
ar
s
est g
r
an
u
lar
it
y
co
v
er
s
t
h
e
en
tire
d
atasets
,
w
h
ic
h
co
n
tai
n
s
o
n
e
s
u
b
g
r
id
s
:
g
r
id
s
1
(
ī
ī
ī
īī
ī½ī
ī
ī
īī
ī
ī
ī½īī
ī
ī
:
īØ
ī°
0
)
at
le
v
el
1
also
co
n
tai
n
s
t
w
o
s
u
b
g
r
id
s
at
le
v
el
2
(
īī
ī
īī
ī½ī
ī
ī
īī
ī
ī
ī½īī
īī
.
:
īØ
ī°
1
)
2.
4
.
F
uzzy
neig
hb
o
rho
o
d r
o
ug
h set
o
utlier
f
a
ct
o
r
(
F
NRO
F
)
I
n
th
is
p
ap
er
,
a
n
e
w
m
et
h
o
d
f
o
r
r
an
k
in
g
o
u
tl
ier
w
h
ich
is
p
r
o
p
o
s
ed
b
ased
o
n
f
u
zz
y
r
o
u
g
h
s
et
d
en
o
ted
ā
F
u
zz
y
n
eig
h
b
o
r
h
o
o
d
r
o
u
g
h
s
et
o
u
tlier
f
ac
to
r
ā
FNR
OF.
Af
t
er
d
iv
id
in
g
ea
ch
d
i
m
e
n
s
io
n
i
n
t
o
in
ter
v
als
o
f
eq
u
al
len
g
th
M,
th
e
d
en
s
it
y
d
i
s
tr
ib
u
ti
o
n
o
f
ea
ch
ce
ll
(
in
f
o
r
m
atio
n
g
r
an
u
lar
i
t
y
)
ca
n
b
e
d
ef
in
ed
as
t
h
e
r
atio
o
f
its
d
en
s
it
y
an
d
th
e
av
er
a
g
e
d
en
s
it
y
o
f
its
k
n
ei
g
h
b
o
r
in
g
ce
ll
s
.
ā
īÆ
īÆ
=
ā
(
īÆ
ī³
īÆ
ī³
)
īÆ”
j=1
*
lo
g
(
īÆ
ī³
īÆ
ī³
)
=
ā
(
īÆ”
ī³
īÆ”
ī³
)
īÆ”
j=1
*
lo
g
(
īÆ”
ī³
īÆ”
ī³
)
(
1
2
)
P
r
o
o
f
:
īÆ
ī³
īÆ
ī³
=
īÆ”
ī³
īÆ
ī³
ā
īÆ
ī³
īÆ”
ī³
=
īÆ”
ī³
īÆ”
ī³
A
n
o
r
m
alize
d
s
co
r
e
o
f
ī
īÆ
īÆ
is
g
i
v
e
n
as
f
o
llo
w
:
ī
īÆ
īÆ
=
1
ā
ā
īÆ
īÆ
ā
ā
īÆ
īÆ
īÆ”
īÆ
ā
īÆ īÆīÆ«
īÆ
ī³
ī³
ī³
ī³
0
ā¤
ī
īÆ
īÆ
ā¤
1
I
tā
s
v
ie
w
ed
as
th
e
r
elati
v
e
d
en
s
it
y
m
ea
s
u
r
e
o
f
īæ
ī
īÆ
(
d
i)
w
i
th
r
esp
ec
t
to
th
e
d
en
s
it
y
o
f
n
s
u
r
r
o
u
n
d
in
g
n
e
ig
h
b
o
r
ā
s
ce
ll.
W
h
en
t
h
e
p
r
o
b
ab
ilit
y
is
u
n
i
f
o
r
m
l
y
d
i
s
tr
ib
u
ted
,
w
e
ar
e
m
o
s
t
u
n
ce
r
tain
ab
o
u
t
th
e
o
u
t
co
m
e,
th
e
e
n
tr
o
p
y
(
s
co
r
e)
is
th
e
h
ig
h
e
s
t
in
t
h
is
ca
s
e.
On
th
e
o
th
er
h
a
n
d
,
w
h
e
n
th
e
d
ata
p
o
in
ts
h
av
e
a
h
i
g
h
l
y
p
r
o
b
ab
ilit
y
m
a
s
s
f
u
n
ctio
n
,
w
e
k
n
o
w
t
h
at
t
h
e
v
a
r
iab
le
is
lik
el
y
to
f
all
w
it
h
i
n
a
s
m
al
l
s
et
o
f
o
u
tco
m
e
s
s
o
t
h
e
u
n
ce
r
tai
n
t
y
a
n
d
th
e
en
tr
o
p
y
(
s
co
r
e)
ar
e
lo
w
.
T
h
e
s
ize
o
f
in
ter
v
al
m
u
s
t
b
e
ca
r
ef
u
l
l
y
s
elec
ted
.
I
f
th
e
in
ter
v
al
s
ize
is
to
o
s
m
all,
t
h
er
e
w
il
l b
e
m
an
y
ce
lls
s
o
th
at
t
h
e
av
er
ag
e
n
u
m
b
er
o
f
p
o
in
ts
i
n
ea
ch
ce
ll c
an
b
e
to
o
s
m
all.
O
n
t
h
e
o
th
er
h
a
n
d
,
if
t
h
e
in
ter
v
a
l
s
ize
is
to
o
lar
g
e,
w
e
m
a
y
n
o
t
b
e
a
b
le
to
ca
p
tu
r
e
th
e
d
if
f
er
en
ce
s
in
d
en
s
it
y
i
n
d
if
f
er
en
t
r
eg
io
n
s
o
f
th
e
s
p
ac
e.
Un
f
o
r
tu
n
atel
y
,
w
it
h
o
u
t
k
n
o
w
i
n
g
t
h
e
d
is
tr
ib
u
tio
n
o
f
t
h
e
d
ata
s
ets,
it
is
d
i
f
f
ic
u
lt
to
e
s
t
i
m
ate
t
h
e
m
i
n
i
m
al
av
er
ag
e
n
u
m
b
er
o
f
p
o
in
ts
r
eq
u
ir
ed
in
ea
ch
ce
ll to
h
a
v
e
th
e
co
r
r
ec
t r
esu
lt.
Def
i
n
itio
n
4
:
Dir
ec
tl
y
d
e
n
s
it
y
-
r
ea
ch
ab
le:
A
ce
ll
īæ
ī
īÆ
is
d
ir
ec
tl
y
d
en
s
it
y
-
r
ea
ch
ab
le
f
r
o
m
a
ce
ll
īæ
ī
īÆ
i
f
o
n
l
y
i
f
,
ī
īÆ
īÆ
ā„
ī
an
d
īæ
ī
īÆ
ā
ī°
(
īæ
ī
īÆ
)
ī
ā
īī
ī
ī
īÆ
(
īæ
ī
īÆ
,
īæ
ī
īÆ
)
=
ī
īÆ
īÆ
ā
ī
īÆ
īÆ
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
9
,
No
.
1
,
Ma
r
ch
20
20
:
1
ā
10
6
T
h
at
is
,
īæ
ī
īÆ
is
a
co
r
e
ce
ll a
n
d
īæ
ī
īÆ
is
i
n
its
n
ei
g
h
b
o
r
h
o
o
d
.
Def
i
n
itio
n
5
:
Den
s
it
y
-
co
n
n
ec
t
ed
.
A
ce
ll
īæ
ī
īÆ
is
d
en
s
it
y
-
co
n
n
ec
te
d
to
a
ce
ll
īæ
ī
īÆ
if
th
er
e
is
a
ce
ll
īæ
ī
īÆ
s
u
c
h
th
at
b
o
th
īæ
ī
īÆ
an
d
īæ
ī
īÆ
ar
e
d
en
s
it
y
-
r
ea
ch
ab
le
f
r
o
m
īæ
ī
īÆ
as sh
o
w
n
i
n
Fi
g
u
r
e
4
.
Fig
u
r
e
4
.
T
h
e
co
n
ce
p
t o
f
d
en
s
it
y
-
r
ea
ch
ab
ilit
y
an
d
d
en
s
it
y
-
co
n
n
ec
t
iv
i
t
y
to
f
o
r
m
clu
s
ter
s
as
co
n
tig
u
o
u
s
d
en
s
e
r
eg
io
n
s
i
n
lo
w
er
ap
p
r
o
x
i
m
atio
n
2.
5
.
A
no
v
el
a
pp
ro
a
ch:
A
hig
h
-
perf
o
r
m
a
nce
pa
ra
llel a
nd
dis
t
ribute
d
co
m
p
uta
t
io
n
us
i
ng
m
a
pre
du
ce
I
n
o
r
d
er
to
co
m
p
u
te
a
n
o
p
ti
m
al
s
et
o
f
cu
t
-
p
o
in
ts
,
m
o
s
t
o
f
d
is
cr
etiza
tio
n
al
g
o
r
ith
m
s
p
er
f
o
r
m
an
iter
ati
v
e
s
ea
r
ch
in
th
e
s
p
ac
e
o
f
ca
n
d
id
ate
d
is
cr
e
tizatio
n
s
,
u
s
in
g
d
if
f
er
en
t
t
y
p
es
o
f
s
co
r
in
g
f
u
n
ctio
n
s
f
o
r
ev
al
u
ati
n
g
a
d
is
cr
etiza
tio
n
,
th
at
ta
k
e
a
lo
t
o
f
ti
m
e.
I
n
t
h
is
p
ap
er
,
w
e
p
r
o
p
o
s
e
a
p
ar
allel
p
r
o
ce
s
s
o
f
d
is
cr
etiza
tio
n
b
ased
o
n
Ma
p
R
ed
u
ce
u
s
i
n
g
s
lid
i
n
g
g
r
id
.
A
s
lid
in
g
g
r
id
is
s
p
ec
if
ied
b
y
d
ef
in
i
n
g
it
s
r
an
g
e
M
an
d
s
lid
e
S.
T
h
e
r
an
g
e
M
is
an
i
n
ter
v
al
o
f
d
is
cr
etiza
t
io
n
wh
ile
t
h
e
s
lid
e
S sp
ec
i
f
ie
s
th
e
p
o
r
tio
n
o
f
t
h
e
g
r
id
t
h
at
i
s
m
o
v
e
d
f
o
r
w
ar
d
.
A
s
l
id
in
g
w
i
n
d
o
w
is
s
p
ec
i
f
ied
as
a
t
u
p
le
(
M,
s
)
.
A
s
m
o
o
t
h
s
lid
in
g
s
p
ec
if
icatio
n
i
s
h
ig
h
l
y
d
es
ir
ed
w
h
er
e
th
e
s
lid
e
S
is
s
m
all
r
elativ
e
to
th
e
r
an
g
e
M.
w
h
er
e
īµ
<
īÆ
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
b
ased
o
n
Ma
p
R
ed
u
ce
co
m
p
u
ted
f
o
r
ea
ch
n
o
d
e
i
(
ī²
īÆ
ā
ī£
)
is
a
p
ar
allel
p
r
o
ce
s
s
th
at
co
n
s
is
ts
o
f
t
h
r
ee
s
tep
s
:
m
ap
,
s
h
u
f
f
l
e,
an
d
r
ed
u
ce
as sh
o
w
n
in
F
ig
u
r
e
5
.
Fig
u
r
e
5
.
Fra
m
e
w
o
r
k
Ma
p
R
ed
u
ce
p
r
o
p
o
s
ed
E
x
a
m
p
le:
S=<U
,
A
={
C
1
,
C
2
,
C
3
,
C
4
,
C
5
}>
P
1
=
{C1
,
C
2
}
P
2
=
{C2
,
C
3
}
P
3
=
{C3
,
C
4
,
C
5
}
(
ī²
īÆ
ā
ī£
an
d
ī£
=
ī²
1
āŖ
ī²
2
āŖ
ī²
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
ļ²
A
fu
z
z
y
n
eig
h
b
o
r
h
o
o
d
r
o
u
g
h
s
et
meth
o
d
fo
r
a
n
o
ma
ly
d
etec
ti
o
n
in
la
r
g
e
s
ca
le
d
a
ta
(
E
l Mezia
ti Ma
r
o
u
a
n
e
)
7
A
t
n
o
d
e
1
(
P1
)
:
E
ac
h
w
o
r
k
er
n
o
d
e
t
h
at
ap
p
lies
th
e
m
ap
f
u
n
ctio
n
r
elate
d
to
ea
ch
g
r
id
d
ef
in
ed
b
y
tu
p
le
{(
M,
s
1
)
,
(
M,
s
2
)
,
(
M
,
s
3
)
,
(
M,
s
4
)
,
(
M,
s
5
)
}
I
n
m
ap
p
h
a
s
e,
f
o
r
ea
ch
g
r
id
g
i
v
en
tu
p
le
(
M,
s
)
,
w
e
g
e
n
er
ates
a
lis
t
(
k
e
y
=
īæ
ī
īÆ
,
v
alu
e=
ī
īÆ
īÆ
)
w
h
er
e
ī
īÆ
īÆ
is
a
s
co
r
e
o
f
īæ
ī
īÆ
.
I
n
s
h
u
f
f
le
p
h
ase,
t
h
e
o
u
tp
u
t
p
air
s
ar
e
p
ar
titi
o
n
ed
an
d
th
en
tr
an
s
f
er
r
ed
to
r
ed
u
ce
r
s
.
I
n
r
ed
u
ce
p
h
ase,
p
air
s
w
it
h
t
h
e
s
a
m
e
k
e
y
ar
e
g
r
o
u
p
ed
to
g
eth
er
as (
īæ
ī
īÆ
,
lis
t(
ī
īÆ
īÆ
))
as sh
o
w
n
in
Fi
g
u
r
e
6
.
Fig
u
r
e
6
.
I
llu
s
tr
ates
h
o
w
th
e
c
ell
o
v
er
lap
s
w
h
e
n
th
e
g
r
id
m
o
v
e
T
h
en
th
e
r
ed
u
ce
f
u
n
ctio
n
g
e
n
er
ates
t
h
e
f
i
n
al
o
u
tp
u
t
p
air
s
lis
t
(
īæ
ī
īÆ
,
ī
īÆ
īÆ
)
f
o
r
ea
ch
f
u
zz
y
ap
p
r
o
x
i
m
atio
n
.
T
h
e
w
h
o
le
p
r
o
ce
s
s
ca
n
b
e
s
u
m
m
ar
ized
as f
o
llo
w
s
:
Ma
p
:
(
M,
s
)
ļØ
(
īæ
ī
īÆ
,
ī
īÆ
īÆ
)
R
ed
u
ce
: (
īæ
ī
īÆ
,
lis
t(
ī
īÆ
īÆ
))
ļØ
(
īæ
ī
īÆ
,
ī
īÆ
īÆ
)
A
p
ar
allel
co
m
p
u
ti
n
g
o
f
FNR
OF a
n
d
its
te
m
p
late
i
m
p
le
m
e
n
tatio
n
Ma
s
ter
:
Get
{
īæ
ī
īÆ
,
ī
īÆ
īÆ
,
ī®ī
ī
īī„
ī
īÆ
}
f
r
o
m
t
h
e
r
es
u
lt q
u
e
u
e
I
f
(
ī
īÆ
īÆ
ā„
ī
)
f
o
r
ea
ch
ce
ll
īæ
ī
īÆ
in
ī®ī
ī
īī„
ī
īÆ
i
f
(
ī
īÆ
īÆ
ā„
ī
an
d
īæ
ī
īÆ
is
n
o
t la
b
eled
)
{
ā
īÆ
īÆ
īÆØ
īƦīƧ
īÆ
ī®½
=
ā
īÆ
īÆ
īÆØ
īƦ
īƧ
īÆ
ī®½
āŖ
īæ
ī
īÆ
L
ab
el
(
īæ
ī
īÆ
)
=
clu
s
tI
D
p
u
t
īæ
ī
īÆ
in
t
h
e
ca
n
d
id
ate
q
u
e
u
e
}
Slav
e:
Get
C
ell
īæ
ī
īÆ
f
r
o
m
t
h
e
ca
n
d
id
ate
q
u
eu
e
ī®ī
ī
īī„
ī
īÆ
=
ī
īī
ī
ā
ī¾
īī
ā
īī
ī
(
īæ
ī
īÆ
)
p
u
t {
īæ
ī
īÆ
,
ī
īÆ
īÆ
,
ī®ī
ī
īī„
ī
īÆ
}
in
th
e
r
es
u
lt q
u
e
u
e
A
l
g
o
r
ith
m
MR
-
F
NR
O
F
:
Fa
s
t
o
u
tlier
d
etec
tio
n
a
lg
o
r
it
h
m
b
ased
o
n
f
u
zz
y
n
eig
h
b
o
r
h
o
o
d
r
o
u
g
h
a
n
d
a
p
ip
elin
e
p
ar
allelis
m
b
et
w
ee
n
m
a
s
ter
an
d
s
lav
e
m
o
d
u
le.
clu
s
tI
D
=0
f
o
r
ea
ch
ce
ll
īæ
ī
īÆ
in
g
r
id
d
atab
ase
{
if
(
īæ
ī
īÆ
is
n
o
t la
b
eled
)
{
I
f
(
ī
īÆ
īÆ
ā¤
ī
)
{
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
9
,
No
.
1
,
Ma
r
ch
20
20
:
1
ā
10
8
ī°ī§
ī©
=
ī°ī§
ī©
āŖ
īæ
ī
īÆ
L
ab
el
(
īæ
ī
īÆ
)
=n
o
is
e
}
else if
(
ī
īÆ
īÆ
ā¤
ī
)
{
ī°ī§
ī©
=
ī°ī§
ī©
āŖ
īæ
ī
īÆ
L
ab
el
(
īæ
ī
īÆ
)
=b
o
u
n
d
ar
y
}
else {
W
h
ile
(
th
er
e
ar
e
p
en
d
in
g
r
esu
lts
)
{
Ma
s
ter
in
(
n
ei
g
h
b
o
r
īæ
ī
īÆ
)
o
u
t c
an
d
id
ate
īæ
ī
īÆ
P
ar
allel:
Slav
e
in
(
ca
n
d
id
ate
īæ
ī
īÆ
)
o
u
t (
n
ei
g
h
b
o
r
īæ
ī
īÆ
)
}
clu
s
tI
D
=
cl
u
s
tI
D
+ 1
}
E
x
a
m
p
le
o
f
co
m
p
u
tat
io
n
(
A
t a
s
in
g
le
n
o
d
e
P
)
:
Ma
p
p
h
ase:
C
ell
i:
īæ
ī
īÆ
20
40
10
70
90
100
80
99
102
ā
īÆ
īÆ
=(
9
0
/2
0
)
*
lo
g
(
9
0
/2
0
)
+(
9
0
/4
0
)
*
lo
g
(
9
0
/4
0
)
+(
9
0
/1
0
)
*
lo
g
(
9
0
/1
0
)
+(
9
0
/7
0
)
*
lo
g
(
9
0
/7
0
)
+(
9
0
/1
0
0
)
*
lo
g
(
9
0
/1
0
0
)
+
(
9
0
/8
0
)
*
lo
g
(
9
0
/8
0
)
+(
9
0
/9
9
)
*
l
o
g
(
9
0
/9
9
)
+
(
9
0
/1
0
2
)
*
lo
g
(
9
0
/1
0
2
)
ā
īÆ
īÆ
=2
8
.
5
3
C
ell
k
:
īæ
ī
īÆ
af
ter
m
o
v
i
n
g
t
h
e
g
r
id
24
44
14
73
94
105
84
103
107
ā
īÆ
īÆ
=(
9
4
/2
4
)
*
lo
g
(
9
4
/2
4
)
+(
9
4
/4
4
)
*
lo
g
(
9
4
/4
4
)
+(
9
4
/1
4
)
*
lo
g
(
9
4
/1
4
)
+(
9
4
/7
3
)
*
lo
g
(
9
4
/7
3
)
+(
9
4
/1
0
5
)
*
lo
g
(
9
4
/1
0
5
)
+(
9
4
/8
4
)
*
lo
g
(
9
4
/8
4
)
+(
9
4
/1
0
3
)
*
lo
g
(
9
4
/1
0
3
)
+
(
9
4
/1
0
7
)
*
lo
g
(
9
4
/1
0
7
)
ā
īÆ
īÆ
==
1
9
.
9
0
ā
īÆ īÆīÆ«
īÆ
80
ā
īÆ
īÆ
īÆ”
īÆ
0
.
3
6
P
i
ļŗ
ļ½
ī
īÆ
īÆ
=
0
.
25
Sh
u
f
f
le
a
n
d
R
ed
u
ce
p
h
a
s
e:
Giv
e
n
a
cu
t p
o
in
t
ī
0
=
0
.
3
ī
īÆ
īÆ
ī³
ā
īÆ
īÆ
ī³
īÆ
=
ī
īÆ
īÆ
ī³
īÆ
=
0
.
36
>
ī
0
ī
īÆ
īÆ
ī³
ā
īÆ
īÆ
ī³
īÆ
=
ī
īÆ
īÆ
ī³
īÆ
=
0
.
25
<
ī
0
ī
īÆ
īÆ
ī³
ā©
īÆ
īÆ
ī³
īÆ
=
1
2
ā
(
ī
īÆ
īÆ
ī³
īÆ
+
ī
īÆ
īÆ
ī³
īÆ
)
ī
īÆ
īÆ
ī³
ā©
īÆ
īÆ
ī³
īÆ
=
1
2
ā
(
0
.
36
+
0
.
25
)
=
0
.
305
>
ī
0
L
o
w
er
ap
p
r
o
x
i
m
atio
n
:
(
īæ
ī
īÆ
ā©
īæ
ī
īÆ
)
ā
īŗ
(
īæ
ī
īÆ
ā
īæ
ī
īÆ
)
ā
īŗ
īŗ
=
īŗ
+
(
īæ
ī
īÆ
ā
īæ
ī
īÆ
)
+
(
īæ
ī
īÆ
ā©
īæ
ī
īÆ
)
Up
p
er
ap
p
r
o
x
i
m
atio
n
:
(
īæ
ī
īÆ
āŖ
īæ
ī
īÆ
)
ā
īŗ
īŗ
=
īŗ
+
(
īæ
ī
īÆ
āŖ
īæ
ī
īÆ
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
ļ²
A
fu
z
z
y
n
eig
h
b
o
r
h
o
o
d
r
o
u
g
h
s
et
meth
o
d
fo
r
a
n
o
ma
ly
d
etec
ti
o
n
in
la
r
g
e
s
ca
le
d
a
ta
(
E
l Mezia
ti Ma
r
o
u
a
n
e
)
9
3.
E
XP
E
R
I
M
E
NT
S AN
D
RE
S
UL
T
S
T
h
e
alg
o
r
ith
m
p
r
o
p
o
s
ed
is
t
ested
w
it
h
s
y
n
t
h
etic
a
n
d
r
e
al
d
ata
co
llected
f
r
o
m
NO
AA
ce
n
ter
.
T
h
e
im
p
le
m
e
n
tatio
n
o
f
t
h
is
w
o
r
k
w
as
r
ea
lized
in
R
u
s
i
n
g
R
S
t
u
d
io
.
Data
s
ets
N
O
AA
:
[
1
5
]
T
h
e
Natio
n
al
C
li
m
atic
Data
C
en
ter
ā
NOAA
:
co
llect
s
a
w
id
e
r
an
g
e
o
f
d
ata;
in
cl
u
d
i
n
g
s
en
s
o
r
s
tr
ea
m
s
w
it
h
te
m
p
o
r
al
in
f
o
r
m
atio
n
,
s
e
n
s
o
r
s
p
atial
in
f
o
r
m
at
io
n
,
te
m
p
er
at
u
r
e,
etc.
3
.
1
.
I
m
pro
v
e
m
ent
in
s
ea
rc
h
t
i
m
e
ef
f
iciency
T
h
e
p
u
r
p
o
s
e
o
f
th
e
ex
p
er
i
m
en
t
w
a
s
to
co
m
p
ar
e
th
e
p
er
f
o
r
m
a
n
ce
b
et
w
ee
n
th
e
a
lg
o
r
it
h
m
p
r
o
p
o
s
ed
MR
-
FNR
O
F
an
d
t
h
e
o
r
ig
i
n
al
L
OF
alg
o
r
ith
m
i
n
ter
m
s
o
f
m
atc
h
i
n
g
d
etec
ted
o
u
tlier
s
a
n
d
ex
ec
u
ti
o
n
ti
m
e.
C
o
m
p
ar
i
n
g
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
to
w
m
e
th
o
d
s
,
it
s
h
o
w
s
th
at
o
u
r
m
et
h
o
d
h
av
e
a
v
er
y
f
ast
p
r
o
ce
s
s
in
g
ti
m
e
w
ith
ac
ce
p
tab
le
tr
ad
e
-
o
f
f
er
r
o
r
s
as s
h
o
w
in
T
ab
le
1
.
T
ab
le
1
.
T
im
e
tak
e
n
an
d
m
atc
h
in
g
d
etec
ted
o
u
tlier
s
ac
co
r
d
i
n
g
to
t
h
e
n
u
m
b
er
o
f
o
b
j
ec
ts
in
th
e
d
ataset
f
o
r
b
o
th
MR
-
FN
R
OF
an
d
L
O
F
m
et
h
o
d
N
u
mb
e
r
o
f
o
b
j
e
c
t
s
T
i
me
t
a
k
e
n
(
se
c
o
n
d
s)
N
u
mb
e
r
o
f
o
u
t
l
i
e
r
s
d
e
t
e
c
t
e
d
MR
-
F
N
R
O
F
M
e
t
h
o
d
(
9
n
o
d
e
s)
L
O
F
me
t
h
o
d
MR
-
F
N
R
O
F
M
e
t
h
o
d
(
9
n
o
d
e
s)
L
O
F
me
t
h
o
d
2
0
2
3
0
.
2
9
5
.
3
7
2
0
3
1
2
3
4
8
4
5
0
.
3
4
1
1
.
3
3
0
2
2
8
4
1
9
7
6
8
1
.
9
5
0
.
2
7
1
3
6
8
8
9
3
8
4
1
9
8
.
4
9
5
2
3
.
4
2
0
2
3
1
9
8
7
3
.
2
.
P
er
f
o
rm
a
nce
o
f
M
R
-
F
N
RO
F
a
cc
o
rding
t
o
nu
m
ber
o
f
wo
rk
er
s
no
de
s
T
h
e
s
ec
o
n
d
ex
p
er
i
m
en
t
s
h
o
ws
th
at
r
ed
u
ctio
n
o
f
t
h
e
r
is
k
o
f
a
T
y
p
e
I
&
I
I
er
r
o
r
is
p
er
f
o
r
m
ed
b
y
in
cr
ea
s
i
n
g
t
h
e
n
u
m
b
er
o
f
w
o
r
k
er
s
n
o
d
es
as
s
h
o
w
n
in
Fi
g
u
r
e
7
.
W
ith
h
ig
h
n
u
m
b
er
o
f
w
o
r
k
er
s
n
o
d
es,
w
e
ar
e
g
etti
n
g
m
o
r
e
o
u
tlier
d
etec
ted
in
u
p
p
er
ap
p
r
o
x
im
a
tio
n
r
o
u
g
h
s
et
(
less
o
f
t
y
p
e
I
I
er
r
o
r
s
)
.
Fig
u
r
e
7
.
An
o
m
al
y
d
etec
tio
n
u
s
in
g
s
u
cc
es
s
i
v
el
y
3
,
5
an
d
7
w
o
r
k
er
s
n
o
d
es g
i
v
e
n
(
alp
h
a,
b
eta)
-
cu
ts
=
(
2
0
%,
5
0
%)
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
9
,
No
.
1
,
Ma
r
ch
20
20
:
1
ā
10
10
4.
CO
NCLU
SI
O
N
T
h
e
ai
m
o
f
t
h
i
s
p
ap
er
is
to
p
r
o
p
o
s
e
a
n
e
w
al
g
o
r
ith
m
o
f
o
u
tlie
r
d
etec
tio
n
th
at
r
ed
u
ce
s
t
h
e
co
m
p
u
tat
io
n
ti
m
e
r
eq
u
ir
ed
b
y
u
s
in
g
g
r
a
n
u
l
ar
co
m
p
u
ti
n
g
m
et
h
o
d
an
d
f
u
z
z
y
r
o
u
g
h
s
et
th
o
er
y
.
T
h
e
alg
o
r
ith
m
M
R
-
F
NR
O
F
d
iv
id
es
th
e
u
n
iv
er
s
es
i
n
to
a
s
m
aller
n
u
m
b
er
o
f
g
r
a
n
u
les,
an
d
c
alcu
late
s
th
e
f
ac
to
r
o
f
o
u
tlier
n
e
s
s
f
o
r
ea
ch
g
r
an
u
le.
T
o
ex
a
m
in
e
t
h
e
ef
f
ec
ti
v
en
e
s
s
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
,
s
ev
er
al
ex
p
er
i
m
en
ts
in
co
r
p
o
r
atin
g
d
if
f
er
en
t
p
ar
a
m
eter
s
w
er
e
co
n
d
u
cted
.
T
h
e
p
r
o
p
o
s
e
d
m
et
h
o
d
MR
-
FN
R
OF,
d
e
m
o
n
s
tr
ated
a
s
i
g
n
i
f
ica
n
t
co
m
p
u
ta
tio
n
ti
m
e
r
ed
u
ct
io
n
.
Mo
r
eo
v
er
,
i
t c
an
also
b
e
ef
f
ec
t
iv
el
y
u
s
ed
f
o
r
r
ea
l
-
ti
m
e
o
u
tlier
d
etec
tio
n
.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
e
au
th
o
r
s
ar
e
v
er
y
m
u
ch
t
h
a
n
k
f
u
l
to
t
h
e
u
n
a
n
i
m
o
u
s
r
ev
ie
wer
s
o
f
t
h
e
p
ap
er
an
d
ed
ito
r
s
o
f
t
h
e
j
o
u
r
n
al
f
o
r
th
eir
co
n
s
tr
u
cti
v
e
an
d
h
elp
f
u
l c
o
m
m
en
ts
t
h
at
i
m
p
r
o
v
ed
t
h
e
q
u
alit
y
o
f
t
h
e
p
ap
er
RE
F
E
R
E
NC
E
S
[1
]
Ha
w
k
in
s,
D.:
Id
e
n
ti
f
ica
ti
o
n
s
o
f
Ou
tl
iers
,
(C
h
a
p
m
a
n
a
n
d
Ha
ll
,
L
o
n
d
o
n
,
1
9
8
0
).
[2
]
A
n
d
re
a
Da
l
P
o
z
z
o
lo
,
G
iac
o
m
o
Bo
ra
c
c
h
i,
ā
Cre
d
it
Ca
rd
F
ra
u
d
De
tec
t
io
n
:
A
Re
a
li
stic
M
o
d
e
li
n
g
a
n
d
a
No
v
e
l
L
e
a
rn
in
g
S
trate
g
y
ā
-
S
e
p
te
m
b
e
r
2
0
1
7
.
[3
]
Ja
b
e
z
J,
B.
M
u
th
u
k
u
m
a
r,
ā
In
tru
sio
n
De
tec
ti
o
n
S
y
ste
m
(IDS)
:
A
n
o
m
a
l
y
De
te
c
ti
o
n
u
sin
g
Ou
tl
ier
De
tec
t
io
n
A
p
p
ro
a
c
h
ā
in
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
In
t
e
ll
ig
e
n
t
Co
m
p
u
ti
n
g
,
C
o
m
m
u
n
ica
ti
o
n
&
C
o
n
v
e
rg
e
n
c
e
(ICCC
-
2
0
1
5
)
El
se
v
ier.
[4
]
S
.
Hi
d
o
,
Y.
T
su
b
o
i
,
H.
Ka
sh
im
a
,
M
.
S
u
g
iy
a
m
a
,
a
n
d
T
.
Ka
n
a
m
o
ri,
ā
S
tatisti
c
a
l
o
u
tl
ier
d
e
tec
ti
o
n
u
sin
g
d
irec
t
d
e
n
sity
ra
ti
o
e
stim
a
ti
o
n
,
ā
Kn
o
w
l.
I
n
fo
rm
.
S
y
st
.
,
v
o
l.
2
6
,
n
o
.
2
,
p
p
.
3
0
9
-
3
3
6
,
2
0
1
1
.
[5
]
K.
Bh
a
d
u
ri,
B.
L
.
M
a
tt
h
e
w
s,
a
n
d
C.
G
ian
n
e
ll
a
,
ā
A
lg
o
rit
h
m
s
f
o
r
sp
e
e
d
in
g
u
p
d
istan
c
e
-
b
a
se
d
o
u
t
li
e
r
d
e
tec
ti
o
n
,
ā
i
n
P
r
o
c
.
A
CM
S
IG
KD
D In
t.
Co
n
f
.
KD
D,
Ne
w
Yo
rk
,
NY
,
USA
,
2
0
1
1
,
p
p
.
8
5
9
-
8
6
7
.
[6
]
P
.
Ra
jas
h
e
k
a
r,
ā
Ra
n
k
in
g
o
u
tl
ier
d
e
tec
ti
o
n
f
o
r
h
ig
h
d
im
e
n
sio
n
a
l
d
a
t
a
u
sin
g
sy
m
m
e
tri
c
n
e
ig
h
b
o
rh
o
o
d
re
latio
n
sh
i
p
ā
,
V
o
l
.
8
,
Iss
u
e
2
-
2
0
1
6
.
[7
]
Ja
y
sh
re
e
S
.
G
o
sa
v
i,
V
in
o
d
S
.
W
a
d
n
e
,
ā
Un
su
p
e
rv
ise
d
Dista
n
c
e
-
Ba
se
d
Ou
tl
ier
De
tec
ti
o
n
Us
i
n
g
Ne
a
re
st
Ne
ig
h
b
o
u
rs
A
l
g
o
rit
h
m
o
n
Distri
b
u
ted
A
p
p
ro
a
c
h
:
S
u
rv
e
y
ā
-
V
o
l.
2
,
Iss
u
e
1
2
,
De
c
e
m
b
e
r
2
0
1
4
.
[8
]
S
h
u
c
h
it
a
Up
a
d
h
y
a
y
a
,
Ka
r
a
n
ji
t
S
i
n
g
h
,
ā
Ne
a
re
st
Ne
i
g
h
b
o
u
r
Ba
se
d
Ou
tl
ier
De
tec
ti
o
n
T
e
c
h
n
iq
u
e
sā
,
v
o
lu
m
e
3
Iss
u
e
2
-
2
0
1
2
.
[9
]
G
u
o
y
in
W
a
n
g
,
Ji
e
Y
a
n
g
,
ji
X
u
,
ā
G
ra
n
u
lar
c
o
m
p
u
ti
n
g
:
f
ro
m
g
r
a
n
u
larity
o
p
ti
m
iza
ti
o
n
to
m
u
lt
i
-
g
ra
n
u
larity
jo
in
t
p
ro
b
lem
so
lv
in
g
ā
-
V
o
lu
m
e
2
,
issu
e
3
-
2
0
1
7
.
[1
0
]
P
a
w
lak
Za (1
9
8
2
)
Ro
u
g
h
se
ts.
I
n
t
J
Pa
ra
l
lel
Pro
g
ra
m
.
1
1
(
5
):3
4
1
-
3
5
6
.
[1
1
]
Ka
laiv
a
n
i.
R,
M
.
V.S
u
re
sh
,
N.S
ri
n
iv
a
sa
n
ā
A
S
tu
d
y
o
f
Ro
u
g
h
S
e
ts
T
h
e
o
r
y
a
n
d
it
s
A
p
p
li
c
a
ti
o
n
Ov
e
r
V
a
rio
u
s
F
ield
sā
V
o
l
u
m
e
3
,
No
.
2
,
(
2
0
1
7
).
[1
2
]
Ya
o
,
Y.Y.,
In
f
o
rm
a
ti
o
n
g
ra
n
u
lati
o
n
a
n
d
ro
u
g
h
se
t
a
p
p
r
o
x
im
a
ti
o
n
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
I
n
telli
g
e
n
t
S
y
ste
ms
,
V
o
l
.
1
6
,
N
o
.
1
,
8
7
-
1
0
4
,
2
0
0
1
.
[1
3
]
Ya
o
,
Jin
g
tao
&
V
a
silak
o
s,
A
th
a
n
a
sio
s
&
P
e
d
ry
c
z
,
W
it
o
ld
.
(2
0
1
3
).
G
ra
n
u
lar
Co
m
p
u
ti
n
g
:
P
e
rsp
e
c
ti
v
e
s
a
n
d
Ch
a
ll
e
n
g
e
s.
IEE
E
tra
n
sa
c
ti
o
n
s
o
n
c
y
b
e
rn
e
ti
c
s
.
4
3
.
1
0
.
1
1
0
9
/T
S
M
CC.
2
0
1
2
.
2
2
3
6
6
4
8
.
[1
4
]
T
.
L
in
,
Ne
ig
h
b
o
rh
o
o
d
sy
ste
m
s a
n
d
re
latio
n
a
l
d
a
teb
a
se
,
i
n
:
P
r
o
c
e
e
d
i
n
g
s o
f
CS
C
Ć
8
8
,
1
9
8
8
.
[1
5
]
f
tp
:/
/f
tp
.
n
c
d
c
.
n
o
a
a
.
g
o
v
/p
u
b
/d
a
ta/n
o
a
a
/
Evaluation Warning : The document was created with Spire.PDF for Python.