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.
37
,
No
.
1
,
J
an
u
ar
y
20
25
,
p
p
.
2
5
7
~
2
6
8
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
37.
i
1
.
p
p
257
-
2
6
8
257
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Enha
ncing
f
a
ke p
ro
file
de
tect
io
n
t
h
ro
ug
h superv
ised
a
nd hy
brid
ma
chine learning
:
a
com
pa
ra
tive a
na
ly
sis
I
s
m
a
il B
ens
a
s
s
i
,
O
us
s
a
m
a
N
da
m
a
,
M
o
ha
m
ed
K
o
uis
s
i
,
E
l
M
o
k
hta
r
E
n
-
N
a
im
i
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
a
n
d
Te
c
h
n
o
l
o
g
y
o
f
Ta
n
g
i
e
r
,
A
b
d
e
l
ma
l
e
k
E
ssaa
d
i
U
n
i
v
e
r
si
t
y
o
f
Te
t
u
a
n
,
Ta
n
g
i
e
r
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
28
,
2
0
2
4
R
ev
is
ed
Au
g
28
,
202
4
Acc
ep
ted
Sep
2
,
2
0
2
4
In
m
o
d
e
r
n
ti
m
e
s,
so
c
ial
n
e
tw
o
rk
s
h
a
v
e
b
e
c
o
m
e
u
b
iq
u
it
o
u
s
p
latfo
rm
s
fa
c
il
it
a
ti
n
g
wid
e
sp
re
a
d
i
n
fo
rm
a
ti
o
n
d
isse
m
in
a
ti
o
n
,
re
su
l
ti
n
g
in
s
ig
n
ifi
c
a
n
t
d
a
il
y
d
a
ta
g
e
n
e
ra
ti
o
n
.
T
h
is
i
n
c
re
a
se
in
d
a
ta
p
ro
d
u
c
t
io
n
e
n
c
o
m
p
a
s
se
s
a
wid
e
ra
n
g
e
o
f
u
se
r
-
g
e
n
e
ra
ted
c
o
n
ten
t,
wh
ich
i
n
t
u
rn
p
r
o
m
o
tes
t
h
e
p
r
o
li
f
e
ra
ti
o
n
o
f
fra
u
d
u
le
n
t
u
se
rs
c
re
a
ti
n
g
fa
k
e
p
r
o
fil
e
s
a
n
d
e
n
g
a
g
in
g
in
d
e
c
e
p
ti
v
e
a
c
ti
v
it
ies
.
Th
is
a
rti
c
le
a
ims
to
a
d
d
re
ss
th
is
c
h
a
ll
e
n
g
e
b
y
e
m
p
lo
y
in
g
m
a
c
h
in
e
lea
rn
in
g
a
lg
o
rit
h
m
s
to
a
c
c
u
ra
tely
i
d
e
n
ti
f
y
fa
k
e
p
ro
f
il
e
s.
T
h
e
re
se
a
rc
h
i
n
v
o
lv
e
s
a
th
o
r
o
u
g
h
a
n
a
l
y
sis
o
f
v
a
rio
u
s
u
se
r
b
e
h
a
v
io
rs,
e
n
g
a
g
e
m
e
n
t
m
e
t
rics
,
a
n
d
c
o
n
ten
t
a
tt
rib
u
tes
wit
h
in
so
c
ial
p
latfo
rm
s.
T
h
e
p
rima
ry
g
o
a
l
is
t
o
d
e
v
e
lo
p
ro
b
u
st
m
o
d
e
ls
c
a
p
a
b
le
o
f
e
ffe
c
ti
v
e
ly
d
e
tec
ti
n
g
d
e
c
e
p
ti
v
e
p
ro
fil
e
s
b
y
m
e
ti
c
u
lo
u
sl
y
e
x
a
m
in
in
g
u
se
r
a
c
ti
v
it
ies
a
n
d
c
o
n
ten
t
c
h
a
ra
c
teristics
.
Th
e
stu
d
y
e
x
p
lo
re
s
t
h
e
a
p
p
li
c
a
ti
o
n
o
f
ro
b
u
st
m
e
th
o
d
o
l
o
g
ies
su
c
h
a
s
K
-
m
e
a
n
s
a
n
d
K
-
m
e
d
o
id
s
c
l
u
ste
rin
g
,
a
l
o
n
g
sid
e
su
p
e
r
v
ise
d
m
a
c
h
i
n
e
lea
rn
i
n
g
c
las
sifiers
in
c
lu
d
in
g
K
-
n
e
a
re
st
n
e
i
g
h
b
o
rs
(KN
N),
su
p
p
o
r
t
v
e
c
to
r
m
a
c
h
in
e
(S
VM),
Be
rn
o
u
l
li
Na
ïv
e
Ba
y
e
s
(NB)
,
lo
g
isti
c
re
g
re
ss
io
n
,
a
n
d
li
n
e
a
r
su
p
p
o
rt
v
e
c
to
r
c
las
sifica
ti
o
n
(
S
VC
)
,
sp
e
c
ifi
c
a
ll
y
tailo
re
d
f
o
r
t
h
e
d
e
tec
ti
o
n
o
f
fa
k
e
p
ro
fil
e
s.
K
ey
w
o
r
d
s
:
B
er
n
o
u
lli Na
ïv
e
B
ay
es
Fak
e
p
r
o
f
iles
d
etec
tio
n
K
-
m
ed
o
id
s
L
in
ea
r
SVC
an
d
K
-
m
ea
n
s
L
o
g
is
tic
r
eg
r
ess
io
n
Su
p
er
v
is
ed
an
d
u
n
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
al
g
o
r
ith
m
s
as
KNN,
SVM
User
b
eh
av
io
r
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
:
I
s
m
ail
B
en
s
as
s
i
Facu
lty
o
f
Scien
ce
an
d
T
ec
h
n
o
lo
g
y
o
f
T
a
n
g
ier
,
Ab
d
elm
alek
E
s
s
aa
d
i U
n
iv
er
s
ity
o
f
T
etu
an
B
.
P 4
1
6
,
p
o
s
tal
co
d
e:
9
0
0
0
0
,
T
an
g
ier
,
Mo
r
o
cc
o
E
m
ail: b
en
s
ass
i.isma
il
@
g
m
ail
.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
r
ap
id
ex
p
a
n
s
io
n
o
f
s
o
cia
l
m
ed
ia
p
latf
o
r
m
s
in
r
ec
e
n
t
y
ea
r
s
h
as
p
r
o
f
o
u
n
d
ly
im
p
ac
t
ed
v
ar
io
u
s
s
o
cieta
l
r
ea
lm
s
,
p
ar
ticu
lar
ly
i
n
m
ar
k
etin
g
a
n
d
ce
leb
r
ity
s
el
f
-
p
r
o
m
o
tio
n
,
wh
ich
h
av
e
s
ig
n
if
ican
tly
b
r
o
ad
en
e
d
th
eir
f
o
llo
wer
a
n
d
f
an
b
ases
.
W
h
ile
th
ese
p
latf
o
r
m
s
o
f
f
er
e
x
ten
s
iv
e
b
e
n
ef
its
s
u
ch
as
wid
esp
r
ea
d
in
f
o
r
m
atio
n
d
is
s
em
in
atio
n
,
th
ey
also
in
tr
o
d
u
ce
c
h
allen
g
es
r
elate
d
to
p
r
iv
ac
y
,
m
is
in
f
o
r
m
atio
n
,
a
n
d
o
n
lin
e
s
ec
u
r
ity
.
A
cr
itical
is
s
u
e
am
o
n
g
th
ese
ch
allen
g
es
is
th
e
wid
esp
r
ea
d
p
r
esen
ce
o
f
f
a
k
e
p
r
o
f
iles
,
wh
ich
p
o
s
e
r
i
s
k
s
b
y
d
is
s
em
in
atin
g
m
is
in
f
o
r
m
atio
n
,
en
g
ag
in
g
in
f
r
au
d
u
len
t
ac
tiv
ities
,
an
d
co
m
p
r
o
m
is
in
g
th
e
p
r
iv
ac
y
o
f
g
en
u
in
e
u
s
er
s
.
Ad
d
r
ess
in
g
th
e
co
m
p
lex
ch
allen
g
es
p
o
s
ed
b
y
f
ak
e
p
r
o
f
iles
r
eq
u
ir
es
a
co
m
p
r
eh
en
s
iv
e
ap
p
r
o
ac
h
in
teg
r
atin
g
tech
n
o
l
o
g
ical
s
o
l
u
tio
n
s
aim
ed
at
d
etec
tin
g
,
p
r
ev
en
tin
g
,
an
d
m
itig
atin
g
th
e
n
eg
ativ
e
im
p
ac
ts
ass
o
ciate
d
with
th
ese
d
ec
ep
ti
v
e
en
titi
es.
T
h
is
s
tu
d
y
f
o
cu
s
e
s
o
n
h
a
r
n
ess
in
g
m
ac
h
i
n
e
lear
n
in
g
alg
o
r
ith
m
s
to
ac
cu
r
ately
d
etec
t
f
ak
e
p
r
o
f
ile
s
o
n
s
o
cial
n
etwo
r
k
s
.
B
y
m
et
icu
lo
u
s
ly
an
aly
zin
g
u
s
er
b
eh
a
v
io
r
s
,
en
g
a
g
em
en
t
m
etr
ics,
an
d
co
n
te
n
t
ch
a
r
ac
ter
is
tics
with
in
th
ese
p
latf
o
r
m
s
,
th
e
g
o
al
is
to
d
ev
elo
p
r
o
b
u
s
t
m
o
d
els
ca
p
a
b
le
o
f
ef
f
ec
tiv
ely
id
en
tif
y
in
g
d
ec
ep
ti
v
e
p
r
o
f
iles
.
E
x
is
tin
g
s
tu
d
ies
ar
e
ca
r
r
ied
o
u
t
with
s
u
p
e
r
v
is
ed
al
g
o
r
ith
m
s
wh
ich
ar
e
g
en
er
ally
m
o
r
e
ac
cu
r
ate
f
o
r
s
p
ec
if
ic
task
s
with
lab
eled
d
at
a,
wh
ile
u
n
s
u
p
er
v
is
ed
alg
o
r
ith
m
s
ar
e
b
etter
f
o
r
d
is
co
v
er
in
g
h
id
d
en
s
tr
u
ctu
r
es
i
n
u
n
lab
eled
d
ata.
I
n
o
u
r
ap
p
r
o
a
ch
,
we
h
av
e
o
p
te
d
f
o
r
a
m
ix
t
u
r
e
o
f
b
o
th
t
y
p
es
o
f
alg
o
r
ith
m
,
wh
ich
allo
ws
u
s
to
tak
e
ad
v
a
n
tag
e
o
f
th
e
s
tr
en
g
t
h
s
o
f
ea
ch
m
eth
o
d
,
u
s
in
g
u
n
s
u
p
er
v
is
ed
alg
o
r
ith
m
s
to
e
x
p
lo
r
e
an
d
p
r
ep
ar
e
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
1
,
J
an
u
ar
y
20
25
:
257
-
2
6
8
258
d
ata
b
ef
o
r
e
ap
p
ly
i
n
g
s
u
p
er
v
is
ed
m
eth
o
d
s
f
o
r
ac
cu
r
ate
p
r
e
d
i
ctio
n
s
.
T
h
is
c
o
m
b
in
atio
n
is
o
f
t
en
u
s
ed
to
o
p
tim
ize
p
er
f
o
r
m
an
ce
in
s
itu
atio
n
s
wh
e
r
e
lab
eled
d
ata
is
s
ca
r
ce
o
r
ex
p
en
s
iv
e
to
o
b
tain
,
a
n
d
also
to
g
et
th
e
b
est
r
esu
lts
in
ter
m
s
o
f
tim
e
an
d
ac
cu
r
ac
y
in
d
etec
tin
g
f
alse p
r
o
f
iles
in
a
b
ig
d
ata
co
n
tex
t.
T
h
is
p
ap
er
’
s
p
r
im
ar
y
co
n
tr
ib
u
tio
n
lies
in
d
ev
elo
p
in
g
a
m
o
d
el
th
at
u
tili
ze
s
v
ar
io
u
s
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
to
an
al
y
ze
lar
g
e
d
atas
ets
an
d
id
en
tif
y
p
atter
n
s
in
d
icativ
e
o
f
f
ak
e
p
r
o
f
ile
s
with
h
eig
h
ten
ed
ac
cu
r
ac
y
.
Ad
d
itio
n
ally
,
th
e
s
tu
d
y
id
en
tifie
s
an
d
ex
am
i
n
es
p
er
tin
en
t
f
ea
tu
r
es
th
at
d
is
tin
g
u
is
h
g
en
u
in
e
f
r
o
m
f
ak
e
p
r
o
f
iles
,
s
u
ch
as
u
s
er
f
o
ll
o
wer
s
,
f
r
ien
d
s
,
f
av
o
r
ites
,
an
d
s
tatu
s
co
u
n
t,
p
r
o
v
id
i
n
g
v
alu
ab
le
in
s
ig
h
ts
in
to
th
e
ch
ar
ac
ter
is
tics
o
f
d
ec
ep
tiv
e
ac
co
u
n
ts
.
T
h
e
r
em
ai
n
d
er
o
f
th
is
p
ap
er
is
s
tr
u
ctu
r
ed
as
f
o
llo
ws:
s
ec
tio
n
1
p
r
o
v
i
d
es
an
ex
ten
s
iv
e
liter
atu
r
e
r
ev
iew
f
o
cu
s
in
g
o
n
d
if
f
er
en
t
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es
e
m
p
lo
y
e
d
in
d
etec
tin
g
f
a
k
e
p
r
o
f
iles
.
Sectio
n
2
elab
o
r
ates
o
n
th
e
m
eth
o
d
o
lo
g
y
u
s
ed
in
th
is
s
tu
d
y
,
d
etailin
g
th
e
d
ataset,
d
ata
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
,
an
d
th
e
ex
tr
ac
ted
f
ea
tu
r
es
f
o
r
an
aly
s
is
.
I
n
s
ec
tio
n
3
p
r
esen
ts
th
e
r
esu
lts
an
d
d
is
cu
s
s
es
f
in
d
in
g
s
,
em
p
h
asizin
g
th
e
ef
f
ec
tiv
e
n
ess
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
.
Fi
n
ally
,
s
ec
tio
n
4
d
r
aws
co
n
cl
u
s
io
n
s
f
r
o
m
th
e
s
tu
d
y
an
d
o
u
tlin
es r
ec
o
m
m
en
d
atio
n
s
f
o
r
f
u
tu
r
e
r
esear
ch
ef
f
o
r
ts
in
t
h
is
f
ield
.
2.
ST
UD
I
E
S AN
D
M
E
T
H
O
D
2
.
1
.
Rela
t
ed
s
t
ud
ies
T
h
e
liter
atu
r
e
r
ev
iew
co
m
p
r
e
h
en
s
iv
ely
ex
am
in
es
p
r
e
v
io
u
s
s
tu
d
ies
an
d
m
eth
o
d
o
lo
g
ies
em
p
lo
y
ed
f
o
r
d
etec
tin
g
f
a
k
e
p
r
o
f
iles
o
n
s
o
cial
m
ed
ia
p
latf
o
r
m
s
.
Var
i
o
u
s
m
eth
o
d
s
ca
teg
o
r
ize
p
r
o
f
iles
b
ased
o
n
ac
c
o
u
n
t
ac
tiv
ity
,
r
esp
o
n
s
e
r
ates
to
r
e
q
u
ests
,
v
o
lu
m
e
o
f
co
m
m
u
n
ic
atio
n
s
,
an
d
o
th
e
r
p
r
o
f
ile
ch
a
r
ac
ter
is
tics
.
So
m
e
m
o
d
els
ar
e
g
r
a
p
h
-
b
ase
d
,
wh
i
le
o
th
er
s
f
o
cu
s
o
n
d
is
tin
g
u
is
h
in
g
b
etwe
en
c
y
b
o
r
g
s
an
d
b
o
ts
u
s
in
g
s
p
ec
if
ic
tech
n
iq
u
es.
Ad
d
itio
n
ally
,
s
p
a
m
d
etec
tio
n
b
ased
o
n
ce
r
tain
k
ey
wo
r
d
s
in
m
ess
ag
es
h
as
b
ee
n
ap
p
lied
to
id
en
tify
f
r
au
d
u
le
n
t
s
o
cial
m
ed
ia
p
r
o
f
iles
.
I
n
2
0
0
8
,
Sy
b
il
R
an
k
was
d
ev
elo
p
e
d
to
m
itig
ate
th
e
im
p
ac
t
o
f
Sy
b
il
attac
k
s
o
n
s
o
cial
m
ed
ia,
r
estrictin
g
r
an
d
o
m
walk
en
co
u
n
ter
s
a
n
d
u
tili
zin
g
Klein
b
er
g
’
s
s
y
n
th
e
tic
s
o
cial
n
etwo
r
k
d
ataset.
C
o
n
cu
r
r
e
n
tly
,
Sy
b
il
R
estrictio
n
em
p
lo
y
ed
a
s
im
ilar
ap
p
r
o
ac
h
b
u
t
em
p
h
asized
r
ap
id
c
o
n
v
er
g
en
ce
o
u
ts
id
e
th
e
Sy
b
il
zo
n
e
u
s
in
g
r
an
d
o
m
v
ar
iab
les
an
d
f
r
eq
u
e
n
cy
-
b
ased
r
an
k
in
g
.
I
n
2
0
0
9
,
Sy
b
il
-
in
f
er
u
tili
ze
d
r
an
d
o
m
ize
d
walk
s
a
n
d
r
ap
i
d
co
n
v
er
g
e
n
ce
o
u
ts
id
e
th
e
n
o
n
-
Sy
b
il
ar
ea
,
em
p
lo
y
in
g
m
o
d
el
-
b
ased
s
am
p
lin
g
,
g
r
ee
d
y
alg
o
r
ith
m
s
,
an
d
B
ay
e
s
ian
n
etwo
r
k
s
with
p
r
o
b
ab
ili
s
tic
th
r
esh
o
ld
s
elec
tio
n
.
Mislo
v
e
’
s
alg
o
r
ith
m
i
n
2
0
1
0
s
elec
ted
p
r
o
f
iles
f
r
o
m
th
e
Face
b
o
o
k
d
ataset
b
ased
o
n
m
etr
ic
-
ad
ju
s
ted
c
o
n
d
u
ctiv
ity
u
s
in
g
g
r
ee
d
y
s
ea
r
ch
.
T
h
e
Face
b
o
o
k
I
m
m
u
n
e
Sy
s
tem
,
in
tr
o
d
u
ce
d
in
2
0
1
1
,
i
n
teg
r
ated
r
an
d
o
m
f
o
r
est,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
,
an
d
b
o
o
s
tin
g
tech
n
iq
u
es
f
o
r
d
etec
tio
n
,
em
p
lo
y
in
g
f
ea
tu
r
e
lo
o
p
s
s
elec
tio
n
o
n
th
e
Face
b
o
o
k
d
ataset.
R
am
alin
g
am
an
d
C
h
in
n
aiah
[
1
]
,
Hajd
u
et
a
l.
[
2
]
in
tr
o
d
u
ce
d
a
r
eg
r
ess
io
n
ap
p
r
o
ac
h
r
an
k
i
n
g
p
r
o
f
iles
b
ased
o
n
in
ter
ac
tio
n
s
,
tag
g
in
g
,
an
d
wal
l
p
o
s
tin
g
s
,
w
h
er
e
l
o
wer
r
an
k
i
n
g
s
in
d
icate
f
alse
ac
co
u
n
ts
co
m
p
ar
ed
to
g
en
u
in
e
o
n
es,
alth
o
u
g
h
its
r
eliab
ilit
y
was
q
u
esti
o
n
ed
.
Swe
an
d
My
o
[
3
]
d
e
v
elo
p
ed
a
b
lac
k
lis
t
d
is
tin
g
u
is
h
in
g
b
etwe
en
f
ak
e
f
ea
tu
r
es
an
d
ac
c
o
u
n
ts
,
wh
ile
an
o
th
er
p
r
o
p
o
s
ed
W
an
d
a
an
d
J
ie
[
4
]
,
u
tili
zin
g
a
s
u
p
er
v
is
ed
lear
n
in
g
alg
o
r
ith
m
with
in
a
d
y
n
am
ic
C
NN
f
r
am
ewo
r
k
f
o
r
d
etec
tio
n
.
Ko
d
ati
et
a
l.
[
5
]
c
o
m
b
in
e
d
S
VM
,
r
an
d
o
m
f
o
r
est,
an
d
Ad
a
B
o
o
s
t
f
o
r
d
etec
tin
g
f
a
k
e
o
n
lin
e
s
o
cial
n
etwo
r
k
(
OSN
)
ac
co
u
n
ts
,
Me
s
h
r
am
et
a
l.
[
6
]
s
p
e
cif
ically
u
s
ed
r
eg
r
ess
io
n
an
aly
s
is
an
d
r
an
d
o
m
f
o
r
est
class
if
ier
s
to
id
e
n
tify
f
ak
e
I
n
s
tag
r
am
ac
co
u
n
ts
.
Var
io
u
s
r
elate
d
wo
r
k
s
by
C
h
ak
r
ab
o
r
ty
et
a
l.
[
7
]
an
d
Ah
m
ed
et
a
l.
[
8
]
h
av
e
also
co
n
tr
ib
u
ted
to
th
is
ev
o
lv
in
g
f
ield
.
2
.
2
.
M
et
ho
d (
m
a
chine le
a
rning
cla
s
s
if
ica
t
io
n a
nd
clus
t
er
i
ng
m
o
dels
)
2
.
2
.
1
.
Su
perv
is
ed
m
a
chine
le
a
rning
a
lg
o
rit
hm
s
A.
T
he
K
-
nea
re
s
t
neig
hb
o
rs
m
a
chine le
a
rning
a
lg
o
rit
hm
:
T
h
e
K
-
n
ea
r
est
n
eig
h
b
o
r
s
(
K
NN
)
alg
o
r
ith
m
is
a
s
tr
aig
h
tf
o
r
war
d
m
eth
o
d
f
o
r
d
ata
class
if
icatio
n
.
I
t
o
p
er
ates
as
a
n
o
n
-
p
ar
am
etr
ic
,
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
u
itab
le
f
o
r
b
o
t
h
class
if
icat
io
n
an
d
r
eg
r
ess
io
n
task
s
.
T
h
e
f
u
n
d
am
en
tal
g
o
al
o
f
KNN
is
to
cla
s
s
if
y
n
ew
d
ata
p
o
in
ts
b
ased
o
n
th
e
attr
ib
u
tes
o
f
n
ea
r
b
y
t
r
ain
in
g
d
ata
p
o
i
n
ts
[
9
]
.
Her
e
’
s
h
o
w
th
e
al
g
o
r
ith
m
ty
p
ically
wo
r
k
s
:
−
Sp
ec
if
y
th
e
n
u
m
b
er
o
f
n
ea
r
est n
eig
h
b
o
r
s
KKK
to
co
n
s
id
er
.
−
Def
in
e
th
e
tr
ain
in
g
d
ataset
D(
y
j)
D(
y
_
j)
D(
y
j)
.
−
C
alcu
late
th
e
d
is
tan
ce
d
(
x
i,
y
j)
d
(
x
_
i,
y
_
j)
d
(
x
i,
y
j
)
b
etwe
en
th
e
n
ew
o
b
s
er
v
atio
n
x
ix
_
ix
i
an
d
ea
c
h
p
o
in
t
y
jy
_
jy
j
i
n
th
e
tr
ain
i
n
g
d
ata.
Var
io
u
s
d
is
tan
ce
m
etr
ics
ca
n
b
e
u
s
ed
,
with
E
u
clid
ea
n
d
is
tan
ce
b
ein
g
th
e
d
ef
au
lt c
h
o
ice.
−
So
r
t th
e
d
is
tan
ce
s
in
ascen
d
in
g
o
r
d
e
r
.
−
Select
th
e
f
ir
s
t K
KK
en
tr
ies w
ith
th
e
s
m
allest d
is
tan
ce
s
.
−
C
las
s
if
y
th
e
o
b
s
er
v
atio
n
x
ix
_
ix
i b
ased
o
n
th
e
m
ajo
r
ity
class
o
f
its
KKK
n
ea
r
est n
eig
h
b
o
r
s
.
KNN
o
f
f
er
s
s
ev
er
al
ad
v
a
n
tag
e
s
:
−
Simp
licity
an
d
ea
s
e
o
f
im
p
le
m
en
tatio
n
:
KNN
is
s
tr
aig
h
tf
o
r
war
d
to
u
n
d
er
s
tan
d
an
d
im
p
le
m
en
t,
m
a
k
in
g
it
ac
ce
s
s
ib
le
ev
en
f
o
r
b
eg
in
n
e
r
s
in
m
ac
h
in
e
lear
n
in
g
.
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:
2
5
0
2
-
4
7
52
E
n
h
a
n
ci
n
g
fa
ke
p
r
o
file d
etec
tio
n
th
r
o
u
g
h
s
u
p
ervis
ed
a
n
d
h
y
b
r
id
…
(
I
s
ma
il B
en
s
a
s
s
i)
259
−
Ver
s
atility
:
i
t
ca
n
b
e
ap
p
lied
to
class
if
icatio
n
,
r
eg
r
ess
io
n
,
an
d
s
ea
r
ch
p
r
o
b
lem
s
.
I
n
clas
s
if
icatio
n
task
s
,
it
p
r
ed
icts
th
e
class
b
ased
o
n
t
h
e
m
o
s
t
p
r
ev
ale
n
t
class
am
o
n
g
its
n
eig
h
b
o
r
s
.
Fo
r
r
eg
r
ess
io
n
,
it
p
r
ed
icts
th
e
v
alu
e
b
ased
o
n
th
e
a
v
er
ag
e
(
o
r
weig
h
ted
av
er
a
g
e)
o
f
its
n
eig
h
b
o
r
s
’
v
alu
es.
Ov
er
all,
KNN
is
v
alu
ed
f
o
r
its
s
im
p
licity
,
v
er
s
atili
ty
ac
r
o
s
s
d
if
f
er
en
t
ty
p
es
o
f
p
r
o
b
lem
s
,
an
d
ef
f
ec
tiv
en
ess
in
s
ce
n
ar
io
s
wh
er
e
th
e
d
ec
is
io
n
b
o
u
n
d
a
r
y
is
n
o
t lin
ea
r
o
r
wh
er
e
th
e
d
ata
d
is
tr
ib
u
tio
n
is
n
o
t k
n
o
wn
in
ad
v
a
n
ce
.
B
.
T
he
s
up
po
rt
v
ec
t
o
r
m
a
chi
ne
m
a
chine le
a
rning
a
lg
o
rit
hm
T
h
e
SVM
is
a
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
u
tili
ze
d
f
o
r
class
if
icatio
n
,
r
e
g
r
e
s
s
io
n
,
an
d
an
o
m
aly
d
etec
tio
n
[1
0
]
.
I
t
o
r
ig
in
ated
in
1
9
9
2
f
r
o
m
a
s
tatis
tical
lear
n
in
g
th
eo
r
y
d
ev
el
o
p
ed
b
y
Vlad
im
ir
Vap
n
ik
,
B
o
s
er
,
an
d
Gu
y
o
n
,
g
ain
in
g
r
a
p
id
ad
o
p
tio
n
d
u
e
to
its
ca
p
ab
ilit
y
to
h
an
d
le
h
ig
h
-
d
im
en
s
io
n
al
d
ata,
th
eo
r
etica
l
u
n
d
e
r
p
in
n
i
n
g
s
,
an
d
p
r
ac
tical
ef
f
ec
tiv
en
ess
.
SVMs
ar
e
ap
p
r
ec
iated
f
o
r
t
h
eir
s
im
p
licity
,
r
eq
u
ir
in
g
f
ew
p
ar
am
ete
r
s
co
m
p
ar
ed
to
o
th
er
m
o
d
els.
I
n
p
r
ac
tice,
SVMs
ar
e
wid
ely
u
s
ed
f
o
r
b
i
n
ar
y
class
if
icatio
n
task
s
[1
1
]
,
im
p
le
m
en
ted
th
r
o
u
g
h
lib
r
ar
ies s
u
ch
as Scik
it
-
lear
n
in
Py
th
o
n
(
s
k
lear
n
.
s
v
m
m
o
d
u
le)
.
T
h
is
allo
ws u
s
er
s
to
s
o
lv
e
b
o
th
class
if
icatio
n
an
d
r
eg
r
ess
io
n
p
r
o
b
lem
s
ef
f
icien
tly
.
T
h
e
SVM
o
p
er
ates
b
y
f
in
d
i
n
g
o
p
tim
al
d
ec
is
io
n
b
o
u
n
d
ar
ies
o
r
h
y
p
er
p
la
n
es
as
s
h
o
wn
in
in
an
N
-
di
m
en
s
io
n
al
s
p
ac
e
[1
2
]
,
w
h
er
e
N
r
e
p
r
esen
ts
th
e
n
u
m
b
er
o
f
f
ea
tu
r
es
in
th
e
d
ataset.
T
h
ese
h
y
p
er
p
lan
es
ac
t
as
d
ec
is
io
n
b
o
u
n
d
ar
ies
to
d
is
tin
ctly
class
if
y
d
ata
p
o
in
ts
[1
3
]
.
T
h
e
p
r
im
ar
y
o
b
jectiv
e
o
f
SVM
is
to
m
ax
im
ize
th
e
m
ar
g
in
b
etwe
en
class
es,
wh
ich
r
ef
er
s
to
th
e
d
is
tan
ce
b
et
wee
n
th
e
n
ea
r
est
d
ata
p
o
in
ts
o
f
d
if
f
er
e
n
t
class
es.
I
n
s
tan
ce
s
o
f
d
ata
th
at
lie
clo
s
est
to
th
ese
s
ep
ar
atin
g
h
y
p
e
r
p
lan
es
ar
e
ter
m
ed
s
u
p
p
o
r
t
v
ec
to
r
s
[1
4
]
,
[
1
5
]
.
Ma
x
im
izin
g
th
is
m
ar
g
in
allo
ws
f
o
r
m
o
r
e
co
n
f
id
e
n
t
class
if
icatio
n
o
f
f
u
tu
r
e
d
ata
p
o
in
t
s
,
c
an
b
e
s
ee
n
in
Fig
u
r
e
1
.
T
h
e
v
er
s
atility
o
f
SVM
ex
ten
d
s
to
v
ar
io
u
s
ap
p
licatio
n
s
,
in
clu
d
in
g
:
−
I
m
ag
e
c
lass
if
icatio
n
:
SVM
s
a
r
e
k
n
o
wn
f
o
r
ac
h
iev
in
g
h
ig
h
ac
cu
r
ac
y
in
task
s
s
u
ch
as
im
a
g
e
class
if
icatio
n
an
d
s
eg
m
en
tatio
n
.
−
T
ex
t
an
d
h
y
p
er
tex
t
ca
te
g
o
r
i
za
tio
n
:
th
ey
ca
n
r
ed
u
ce
th
e
d
ep
en
d
en
cy
o
n
lab
eled
d
at
a,
m
ak
in
g
th
em
ef
f
ec
tiv
e
in
ca
teg
o
r
izin
g
te
x
tu
al
an
d
web
c
o
n
ten
t.
−
Han
d
wr
itten
ch
ar
ac
ter
s
r
ec
o
g
n
itio
n
:
SVMs
ex
h
ib
it
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
in
r
ec
o
g
n
izin
g
h
a
n
d
wr
itten
ch
ar
ac
ter
s
.
−
Pro
tein
s
class
if
icatio
n
in
b
io
lo
g
y
: th
ey
a
r
e
ap
p
lied
to
class
if
y
p
r
o
tein
s
in
b
io
lo
g
ical
r
esear
c
h
.
Ov
er
all,
SVMs
ar
e
r
o
b
u
s
t
to
o
ls
in
m
ac
h
i
n
e
lear
n
in
g
,
v
alu
e
d
f
o
r
t
h
eir
a
b
ilit
y
to
h
an
d
le
c
o
m
p
lex
d
atasets
an
d
d
eliv
er
r
eliab
le
r
esu
lts
ac
r
o
s
s
d
iv
er
s
e
d
o
m
ai
n
s
.
Fig
u
r
e
1
.
Op
tim
al
h
y
p
e
r
p
lan
e
an
d
s
u
p
p
o
r
t v
ec
t
o
r
s
C.
T
he
lo
g
is
t
ic
re
g
re
s
s
io
n
L
o
g
is
tic
r
eg
r
ess
io
n
is
a
s
tat
is
tical
m
eth
o
d
p
r
im
ar
ily
em
p
lo
y
ed
f
o
r
b
in
a
r
y
class
if
icatio
n
task
s
,
p
r
ed
ictin
g
th
e
p
r
o
b
ab
ilit
y
th
at
an
in
s
tan
ce
b
elo
n
g
s
to
a
s
p
ec
if
ic
ca
teg
o
r
y
.
Desp
ite
its
n
am
e,
lo
g
is
tic
r
eg
r
ess
io
n
f
u
n
ctio
n
s
as
a
class
if
icatio
n
alg
o
r
ith
m
r
ath
er
th
a
n
a
r
eg
r
ess
io
n
alg
o
r
ith
m
.
I
t
is
d
esig
n
ed
to
m
o
d
el
th
e
p
r
o
b
a
b
ilit
y
o
f
a
n
in
s
tan
ce
b
elo
n
g
in
g
t
o
a
p
ar
ticu
la
r
class
b
ased
o
n
i
n
p
u
t f
ea
t
u
r
es.
T
h
is
m
eth
o
d
is
p
ar
ticu
lar
ly
well
-
s
u
ited
f
o
r
b
in
ar
y
class
i
f
icatio
n
p
r
o
b
lem
s
wh
er
e
th
e
o
u
tco
m
e
v
ar
iab
le
(
d
e
p
en
d
e
n
t
v
ar
iab
le)
h
as
two
p
o
s
s
ib
le
o
u
tco
m
es.
L
o
g
is
tic
r
eg
r
ess
io
n
is
ch
o
s
en
wh
en
th
e
d
e
p
en
d
e
n
t
v
ar
iab
le
is
d
ich
o
to
m
o
u
s
,
m
ea
n
in
g
it
h
as
two
ca
teg
o
r
ies
o
r
lev
els.
I
t
is
u
s
ed
to
an
aly
ze
an
d
ex
p
lain
th
e
r
elatio
n
s
h
ip
b
etwe
en
a
b
in
ar
y
o
u
tco
m
e
v
ar
ia
b
le
an
d
o
n
e
o
r
m
o
r
e
in
d
ep
e
n
d
en
t
v
a
r
iab
les,
wh
ich
ca
n
b
e
n
o
m
in
al,
o
r
d
in
al,
i
n
ter
v
al,
o
r
r
atio
-
lev
el
v
ar
iab
les
[1
6
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
1
,
J
an
u
ar
y
20
25
:
257
-
2
6
8
260
I
n
ess
en
ce
,
lo
g
is
tic
r
eg
r
ess
io
n
esti
m
ates
th
e
p
r
o
b
ab
ilit
y
o
f
a
b
i
n
ar
y
o
u
tco
m
e
b
ased
o
n
p
r
ed
icto
r
v
ar
iab
les,
p
r
o
v
id
in
g
in
s
ig
h
ts
in
to
th
e
lik
elih
o
o
d
o
f
a
n
in
s
tan
ce
b
elo
n
g
in
g
to
a
s
p
ec
if
ic
class
.
I
t
is
wid
ely
ap
p
lied
in
v
ar
io
u
s
f
ield
s
f
o
r
task
s
s
u
ch
as
p
r
ed
ictin
g
cu
s
to
m
er
ch
u
r
n
,
m
ed
ical
d
iag
n
o
s
es,
cr
ed
it
s
co
r
in
g
,
an
d
m
o
r
e,
wh
er
e
u
n
d
er
s
tan
d
in
g
th
e
p
r
o
b
a
b
ilit
y
o
f
a
n
ev
e
n
t o
cc
u
r
r
in
g
is
cr
u
cial
f
o
r
d
ec
is
io
n
-
m
a
k
in
g
.
D.
T
he
lin
ea
r
s
up
po
rt
v
ec
t
o
r
cla
s
s
if
ica
t
io
n
L
in
ea
r
s
u
p
p
o
r
t
v
ec
to
r
class
if
icatio
n
(
SVC
)
,
s
h
o
r
t
f
o
r
l
in
ea
r
SVC
,
i
s
a
v
ar
ian
t
o
f
th
e
SVM
alg
o
r
ith
m
tailo
r
ed
f
o
r
b
o
th
b
in
a
r
y
an
d
m
u
lticlas
s
clas
s
if
icatio
n
task
s
.
Un
lik
e
tr
ad
itio
n
al
SVMs
th
at
s
ee
k
a
h
y
p
er
p
lan
e
t
o
s
ep
ar
ate
d
ata
in
to
d
is
tin
ct
c
lass
es,
l
in
ea
r
SVC
s
p
ec
if
ical
ly
f
o
cu
s
es
o
n
ac
h
ie
v
in
g
lin
ea
r
s
ep
ar
atio
n
.
I
ts
o
b
jectiv
e
is
to
id
en
tify
a
h
y
p
e
r
p
lan
e
with
in
th
e
f
ea
tu
r
e
s
p
ac
e
th
at
ef
f
ec
tiv
ely
d
iv
id
es
in
s
t
an
ce
s
b
elo
n
g
in
g
to
d
if
f
er
en
t c
lass
es.
T
h
e
p
r
im
ar
y
aim
o
f
l
in
ea
r
SVC
is
to
estab
li
s
h
a
h
y
p
er
p
lan
e
th
at
m
ax
im
izes
th
e
m
ar
g
in
o
f
s
ep
ar
atio
n
b
etwe
en
class
es.
T
h
is
m
ar
g
in
r
ep
r
esen
ts
th
e
d
is
tan
ce
b
etwe
en
th
e
h
y
p
e
r
p
lan
e
an
d
th
e
clo
s
est
d
ata
p
o
in
ts
f
r
o
m
ea
ch
class
,
k
n
o
wn
as
s
u
p
p
o
r
t
v
ec
to
r
s
.
B
y
o
p
tim
izin
g
a
co
n
v
ex
q
u
ad
r
atic
p
r
o
g
r
a
m
m
in
g
p
r
o
b
lem
,
l
in
ea
r
SVC
d
ete
r
m
in
es
th
e
co
ef
f
icien
ts
an
d
b
ias
ter
m
s
th
at
d
ef
in
e
th
e
s
ep
ar
atin
g
h
y
p
er
p
la
n
e.
T
h
is
o
p
tim
izatio
n
p
r
o
ce
s
s
aim
s
to
m
in
im
ize
class
if
icatio
n
er
r
o
r
s
wh
ile
en
s
u
r
in
g
t
h
e
lar
g
est p
o
s
s
ib
le
m
ar
g
in
b
etwe
en
class
es.
I
n
p
r
ac
tical
ter
m
s
,
l
in
ea
r
SVC
is
ad
v
an
tag
eo
u
s
f
o
r
s
c
en
ar
io
s
wh
er
e
th
e
r
elatio
n
s
h
ip
b
etwe
en
f
ea
tu
r
es
an
d
class
es
ca
n
b
e
ef
f
ec
tiv
ely
m
o
d
eled
u
s
in
g
a
lin
ea
r
d
ec
is
io
n
b
o
u
n
d
a
r
y
.
I
t
is
wid
ely
u
s
ed
in
ap
p
licatio
n
s
s
u
ch
as
tex
t
ca
teg
o
r
izatio
n
,
i
m
ag
e
r
ec
o
g
n
itio
n
,
an
d
s
en
tim
en
t
an
aly
s
is
,
wh
er
e
lin
ea
r
s
e
p
ar
atio
n
s
u
f
f
i
ce
s
to
ac
h
iev
e
ac
cu
r
ate
class
if
icatio
n
r
esu
lts
.
Ov
er
all,
l
in
ea
r
SV
C
p
r
o
v
id
es
a
co
m
p
u
tatio
n
ally
e
f
f
icien
t
ap
p
r
o
ac
h
to
b
in
ar
y
an
d
m
u
lticlas
s
class
if
i
ca
tio
n
p
r
o
b
lem
s
b
y
le
v
er
ag
in
g
th
e
p
r
in
ci
p
les
o
f
SVMs,
em
p
h
asizin
g
s
im
p
licit
y
an
d
ef
f
ec
tiv
e
n
ess
in
s
ep
ar
atin
g
d
ata
p
o
in
ts
ac
r
o
s
s
d
if
f
er
en
t c
lass
es.
E.
T
he
B
er
no
ull
i N
a
iv
e
B
a
y
e
s
(
B
er
no
ull
i N
B
)
T
h
e
B
er
n
o
u
lli
NB
alg
o
r
ith
m
is
a
p
r
o
b
ab
ilis
tic
class
if
icat
io
n
m
eth
o
d
b
elo
n
g
i
n
g
to
th
e
NB
f
am
ily
.
I
t
is
s
p
ec
if
ically
d
esig
n
ed
f
o
r
b
in
ar
y
class
if
icatio
n
task
s
wh
er
e
ea
ch
f
ea
tu
r
e
r
ep
r
esen
ts
a
b
in
ar
y
attr
ib
u
te
.
At
its
co
r
e,
th
e
NB
alg
o
r
ith
m
o
p
er
ates
b
ased
o
n
B
ay
es
’
th
e
o
r
em
[1
7
]
,
w
h
ich
ca
lcu
lates
th
e
p
r
o
b
ab
ilit
y
o
f
a
h
y
p
o
th
esis
g
iv
en
o
b
s
er
v
ed
e
v
id
en
ce
.
T
h
e
“
N
aïv
e”
ass
u
m
p
tio
n
in
NB
is
th
at
f
ea
tu
r
es
ar
e
co
n
d
itio
n
ally
in
d
ep
en
d
en
t g
iv
e
n
th
e
class
lab
el,
s
im
p
lify
in
g
p
r
o
b
ab
ilit
y
ca
lcu
latio
n
s
.
B
er
n
o
u
lli
NB
is
a
s
p
ec
ial
ize
d
v
ar
ian
t
o
f
NB
s
u
ited
f
o
r
d
atasets
wi
th
b
in
ar
y
f
ea
tu
r
es,
wh
er
e
ea
ch
f
ea
tu
r
e
is
a
b
in
ar
y
v
a
r
iab
le
in
d
icatin
g
p
r
esen
ce
o
r
a
b
s
en
ce
(
e.
g
.
,
p
r
esen
ce
o
f
a
ter
m
i
n
a
d
o
cu
m
e
n
t,
s
p
am
o
r
n
o
t
s
p
am
)
.
I
n
B
er
n
o
u
lli
N
B
,
th
e
in
p
u
t
is
a
f
ea
tu
r
e
v
ec
to
r
wh
er
e
ea
ch
elem
en
t
r
ep
r
esen
t
s
a
b
in
ar
y
v
a
r
iab
le.
T
y
p
ically
,
th
e
p
r
esen
ce
o
f
a
f
e
atu
r
e
is
d
en
o
ted
b
y
1
,
an
d
its
ab
s
en
ce
b
y
0
.
I
n
p
r
ac
tice,
B
er
n
o
u
lli
NB
is
ef
f
ec
tiv
e
f
o
r
task
s
s
u
ch
as
tex
t
class
if
icatio
n
(
e.
g
.
,
id
e
n
tif
y
in
g
s
p
am
em
ails
)
,
wh
er
e
f
ea
tu
r
es
ar
e
b
in
ar
y
in
d
icato
r
s
o
f
s
p
ec
if
ic
ter
m
s
o
r
ch
ar
ac
ter
is
tics
.
I
ts
s
im
p
licity
an
d
ef
f
icien
cy
m
ak
e
it
s
u
itab
le
f
o
r
lar
g
e
-
s
ca
le
d
atasets
an
d
r
ea
l
-
tim
e
a
p
p
lic
atio
n
s
,
lev
er
ag
i
n
g
th
e
p
r
o
b
ab
il
is
tic
f
r
am
ewo
r
k
o
f
NB
to
m
ak
e
r
eliab
le
p
r
e
d
ict
io
n
s
b
ased
o
n
b
in
ar
y
f
ea
tu
r
e
i
n
p
u
ts
.
2
.
2
.
2
.
Uns
up
er
v
is
e
d
m
a
chine le
a
rning
a
lg
o
rit
hm
s
A.
T
he
K
-
m
ea
ns
a
lg
o
rit
hm
Data
clu
s
ter
in
g
in
v
o
lv
es g
r
o
u
p
in
g
item
s
s
o
th
at
th
o
s
e
with
in
th
e
s
am
e
g
r
o
u
p
ar
e
m
o
r
e
s
im
ilar
to
ea
ch
o
th
er
th
an
to
th
o
s
e
in
o
th
er
g
r
o
u
p
s
[
1
8
]
.
T
h
e
p
r
im
ar
y
g
o
al
is
to
id
en
tify
h
o
m
o
g
en
e
o
u
s
s
u
b
g
r
o
u
p
s
with
in
th
e
d
ata
th
at
s
h
ar
e
c
o
m
m
o
n
ch
a
r
a
cter
is
tics
[
19
]
,
u
s
in
g
s
im
ilar
i
t
y
m
ea
s
u
r
es
lik
e
E
u
clid
ea
n
d
is
tan
ce
o
r
co
r
r
elatio
n
-
b
ased
m
etr
ics.
Un
lik
e
s
u
p
e
r
v
i
s
ed
lear
n
in
g
,
clu
s
ter
in
g
is
u
n
s
u
p
er
v
is
ed
,
m
ea
n
in
g
th
er
e
ar
e
n
o
p
r
ed
ef
in
e
d
lab
els
to
g
u
id
e
t
h
e
g
r
o
u
p
in
g
p
r
o
ce
s
s
; in
s
tead
,
it a
im
s
to
u
n
co
v
er
in
h
er
en
t stru
ctu
r
es with
in
th
e
d
a
ta.
Am
o
n
g
cl
u
s
ter
in
g
alg
o
r
ith
m
s
,
K
-
m
ea
n
s
is
wid
ely
r
ec
o
g
n
ize
d
an
d
u
tili
ze
d
.
I
t
o
p
er
ates
iter
a
tiv
ely
to
p
ar
titi
o
n
a
d
ataset
in
to
K
-
d
is
tin
ct
clu
s
ter
s
,
wh
er
e
ea
ch
d
ata
p
o
i
n
t
b
elo
n
g
s
ex
clu
s
iv
ely
to
o
n
e
clu
s
ter
[2
0
]
.
T
h
e
f
u
n
d
am
en
tal
o
b
jectiv
e
o
f
K
-
m
ea
n
s
is
to
m
in
im
ize
in
tr
a
-
clu
s
ter
v
ar
iab
ilit
y
wh
ile
m
ax
i
m
izin
g
in
ter
-
clu
s
ter
d
if
f
er
en
ce
s
[2
1
]
.
E
s
s
en
tially
,
t
h
is
m
ea
n
s
m
ak
in
g
d
ata
p
o
in
ts
with
in
ea
ch
clu
s
ter
as
s
im
ilar
as
p
o
s
s
ib
le
wh
ile
en
s
u
r
in
g
d
is
s
im
ilar
ity
b
etwe
en
clu
s
ter
s
.
T
h
e
K
-
m
ea
n
s
alg
o
r
ith
m
f
o
llo
ws th
ese
s
tep
s
in
u
n
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
i
n
g
:
−
Sp
ec
if
y
th
e
n
u
m
b
er
o
f
clu
s
ter
s
KKK.
−
I
n
itialize
ce
n
tr
o
id
s
:
i
n
itially
,
s
h
u
f
f
le
t
h
e
d
ataset
an
d
r
an
d
o
m
ly
s
elec
t
K
-
d
ata
p
o
in
ts
as
c
en
tr
o
id
s
with
o
u
t
r
ep
lace
m
en
t.
−
I
ter
ate
u
n
til
co
n
v
er
g
e
n
ce
:
c
o
n
tin
u
e
iter
atin
g
u
n
til
ce
n
tr
o
id
s
n
o
lo
n
g
er
ch
a
n
g
e,
in
d
icatin
g
s
tab
ilizatio
n
in
clu
s
ter
ass
ig
n
m
en
ts
.
−
Ass
ig
n
d
ata
p
o
in
ts
to
clu
s
ter
s
:
c
alcu
late
th
e
d
i
s
tan
ce
b
etwe
en
ea
ch
d
ata
p
o
i
n
t
an
d
all
ce
n
tr
o
id
s
,
ass
ig
n
in
g
ea
ch
d
ata
p
o
in
t to
th
e
cl
o
s
est ce
n
tr
o
id
.
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:
2
5
0
2
-
4
7
52
E
n
h
a
n
ci
n
g
fa
ke
p
r
o
file d
etec
tio
n
th
r
o
u
g
h
s
u
p
ervis
ed
a
n
d
h
y
b
r
id
…
(
I
s
ma
il B
en
s
a
s
s
i)
261
−
Up
d
ate
ce
n
tr
o
id
s
:
r
ec
o
m
p
u
te
ce
n
tr
o
id
s
b
y
a
v
er
ag
i
n
g
th
e
p
o
s
itio
n
s
o
f
all
d
ata
p
o
in
ts
ass
ig
n
ed
to
ea
ch
clu
s
ter
.
K
-
m
ea
n
s
is
f
av
o
r
ed
f
o
r
its
ab
ilit
y
to
h
an
d
le
lar
g
e
d
atasets
ef
f
icien
tly
,
o
f
f
e
r
in
g
r
elativ
ely
f
ast
co
m
p
u
tatio
n
tim
es
[2
2
]
.
I
t
is
co
m
m
o
n
l
y
ap
p
lied
in
v
ar
io
u
s
f
ield
s
f
o
r
task
s
s
u
ch
as
cu
s
t
o
m
er
s
eg
m
e
n
tatio
n
,
im
ag
e
co
m
p
r
ess
io
n
,
an
d
an
o
m
aly
d
etec
tio
n
,
wh
er
e
u
n
co
v
er
in
g
n
atu
r
al
g
r
o
u
p
in
g
s
in
d
ata
is
es
s
en
tial
f
o
r
an
aly
s
is
an
d
d
ec
is
io
n
-
m
a
k
in
g
.
B
.
T
he
K
-
m
edo
ids
a
lg
o
rit
hm
T
h
e
p
ar
titi
o
n
in
g
a
r
o
u
n
d
m
ed
o
id
s
(
K
-
m
ed
o
id
s
)
alg
o
r
ith
m
is
a
v
ar
ian
t
o
f
th
e
K
-
m
ea
n
s
alg
o
r
ith
m
th
at
d
if
f
er
s
in
its
u
s
e
o
f
m
ed
o
id
s
r
ath
er
th
an
m
ea
n
s
as
ce
n
tr
o
id
s
.
A
m
ed
o
id
with
in
a
clu
s
ter
is
an
o
b
ject
th
at
m
in
im
izes
th
e
av
er
ag
e
d
is
s
im
ilar
ity
to
all
o
th
er
o
b
jects
in
th
at
clu
s
ter
.
T
h
e
K
-
m
ed
o
id
s
alg
o
r
ith
m
s
tar
ts
b
y
co
m
p
u
tin
g
K
-
m
e
d
o
id
s
a
n
d
th
en
ass
ig
n
s
ea
ch
o
b
ject
in
th
e
d
ataset
to
th
e
n
ea
r
est
m
ed
o
i
d
b
ased
o
n
a
c
h
o
s
en
d
is
tan
ce
m
etr
ic.
Su
b
s
eq
u
en
tly
,
it e
v
alu
ates th
e
co
s
t o
f
s
wap
p
in
g
a
d
ata
p
o
in
t PiP_
iPi
with
a
m
ed
o
id
M
iM_
iMi
to
p
o
ten
tially
im
p
r
o
v
e
th
e
clu
s
ter
in
g
ar
r
a
n
g
em
en
t
[2
3
]
.
T
o
o
u
tlin
e
th
e
s
tep
s
o
f
th
e
K
-
m
ed
o
id
s
alg
o
r
ith
m
:
−
Step
1
:
s
elec
t th
e
in
itial K
p
o
in
ts
as m
ed
o
id
s
.
−
Step
2
:
e
v
alu
ate
th
e
c
o
s
t a
s
s
o
ciate
d
with
s
wap
p
in
g
ea
ch
s
ele
cted
m
ed
o
id
with
ea
ch
u
n
s
elec
ted
d
ata
p
o
i
n
t.
−
Step
3
:
c
h
o
o
s
e
th
e
s
wap
th
at
r
esu
lts
in
th
e
lo
west c
o
s
t.
−
Step
4
:
r
ep
ea
t
u
n
til
n
o
f
u
r
th
e
r
im
p
r
o
v
em
en
ts
in
co
s
t
ca
n
b
e
ac
h
iev
ed
,
wh
er
e
ea
ch
u
n
s
elec
ted
p
o
in
t
is
co
m
p
ar
ed
to
its
clo
s
est s
elec
te
d
m
ed
o
i
d
[2
4
]
.
T
h
e
K
-
m
ed
o
id
s
alg
o
r
ith
m
is
ef
f
ec
tiv
e
in
s
ce
n
ar
io
s
wh
er
e
th
e
ch
o
ice
o
f
a
r
ep
r
esen
tativ
e
o
b
ject
(
m
ed
o
id
)
is
cr
u
cial,
s
u
ch
as
in
b
io
lo
g
ical
clu
s
ter
in
g
o
r
m
ar
k
et
s
eg
m
en
tatio
n
,
wh
er
e
id
en
tify
i
n
g
t
h
e
m
o
s
t
ty
p
ical
o
r
r
ep
r
esen
tativ
e
m
em
b
er
o
f
a
g
r
o
u
p
is
ess
en
tial f
o
r
ac
cu
r
ate
a
n
aly
s
is
an
d
d
ec
is
io
n
-
m
a
k
in
g
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Su
perv
is
ed
a
lg
o
rit
hm
a
pp
lica
t
io
n r
esu
lt
s
3
.
1
.
1
.
B
uil
din
g
t
he
cla
s
s
if
ica
t
io
n m
o
del
T
r
ain
in
g
:
t
h
is
co
n
s
is
ts
o
f
t
wo
m
o
d
u
les,
tr
ain
in
g
an
d
test
in
g
,
ea
ch
u
s
in
g
a
p
ar
t
o
f
th
e
f
ea
tu
r
e
b
ase
s
u
b
d
iv
id
ed
in
t
o
two
p
ar
ts
,
th
e
tr
ain
in
g
b
ase
an
d
th
e
test
b
as
e.
T
h
e
tr
ain
in
g
m
o
d
u
le
u
s
es
th
e
tr
ain
in
g
b
ase
to
p
r
o
v
id
e
a
d
ec
is
io
n
m
o
d
el,
wh
ile
th
e
test
m
o
d
u
l
e
u
s
es
th
e
test
b
ase
to
m
ea
s
u
r
e
t
h
e
p
e
r
f
o
r
m
an
ce
o
f
t
h
e
m
o
d
el
p
r
o
v
id
e
d
.
Use:
t
h
is
is
th
e
last
an
d
m
o
s
t
i
m
p
o
r
tan
t
p
h
ase
i
n
o
u
r
s
y
s
tem
.
Af
ter
h
a
v
in
g
a
b
etter
p
e
r
f
o
r
m
an
ce
r
ate,
o
r
af
ter
h
a
v
in
g
b
u
ilt
th
e
b
est
m
o
d
el
in
th
e
p
r
ev
io
u
s
p
h
ase,
we
m
u
s
t
u
s
e
it
o
n
n
ew
u
n
p
r
o
ce
s
s
ed
in
f
o
r
m
atio
n
,
an
d
th
e
m
o
d
el
th
at
allo
ws
u
s
to
p
r
ed
ict
th
e
class
o
f
th
e
n
ew
s
itu
atio
n
if
it
i
s
f
alse
o
r
tr
u
e
with
a
d
eg
r
ee
o
f
co
n
f
id
en
ce
as sh
o
wn
in
Fig
u
r
e
2
.
Fig
u
r
e
2
.
Usag
e
p
r
o
ce
s
s
I
n
th
is
p
h
ase,
we
ap
p
ly
f
iv
e
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
i
n
g
alg
o
r
ith
m
s
to
th
e
p
r
o
f
iles
m
en
tio
n
ed
ea
r
lier
:
lo
g
is
tic
r
eg
r
ess
io
n
,
l
i
n
ea
r
SVC
,
KNN,
SVM,
an
d
B
er
n
o
u
lli
NB
.
T
h
ese
ex
p
er
im
en
ts
ar
e
co
n
d
u
cted
u
s
in
g
Scik
it
-
lear
n
[2
5
]
,
[
2
6
]
,
a
Py
th
o
n
-
b
ased
m
ac
h
in
e
lear
n
in
g
lib
r
ar
y
th
at
o
f
f
er
s
a
wid
e
ar
r
ay
o
f
t
o
o
ls
an
d
m
o
d
els f
o
r
s
tr
ea
m
lin
e
d
im
p
le
m
en
tatio
n
o
f
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
.
Fo
llo
win
g
d
ata
clea
n
in
g
to
r
e
m
o
v
e
o
u
tlier
s
an
d
e
n
h
an
ce
d
a
ta
q
u
ality
,
we
ca
r
e
f
u
lly
s
elec
t
ed
a
s
u
b
s
et
o
f
f
ea
tu
r
es
d
ee
m
e
d
p
er
tin
e
n
t
f
o
r
th
e
m
o
d
el
d
ev
el
o
p
m
en
t
p
r
o
ce
s
s
.
T
h
ese
s
elec
ted
f
ea
tu
r
es
ar
e
ess
en
tial
f
o
r
tr
ain
in
g
a
n
d
e
v
alu
atin
g
th
e
p
e
r
f
o
r
m
a
n
ce
o
f
ea
ch
alg
o
r
ith
m
i
n
o
u
r
an
al
y
s
is
.
T
h
e
T
a
b
le
1
is
an
ex
tr
ac
t
f
r
o
m
o
u
r
d
ata
s
et
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
1
,
J
an
u
ar
y
20
25
:
257
-
2
6
8
262
T
ab
le
1
.
E
x
tr
ac
t f
r
o
m
o
u
r
d
ata
b
ase
id
st
a
t
u
s
e
s
_
c
o
u
n
t
f
o
l
l
o
w
e
r
s
_
c
o
u
n
t
f
r
e
i
n
d
s
_
c
o
u
n
t
f
a
v
o
u
r
i
t
e
s
_
c
o
u
n
t
l
i
st
e
d
_
c
o
u
n
t
p
r
o
f
i
l
e
_
u
se
_
b
a
c
k
g
r
o
u
n
d
_
i
ma
g
e
is
F
a
k
e
0
3
6
1
0
5
1
1
2
0
3
7
0
5
4
7
0
2
3
8
5
1
4
5
52
1
.
0
0
1
5
6
5
6
1
6
2
3
1
3
1
5
0
6
3
8
1
9
40
1
.0
0
1
5
6
8
2
7
0
2
4
0
2
4
2
6
4
87
3
2
3
16
1
.0
0
3
6
0
6
7
2
9
2
4
0
5
8
6
6
4
0
6
2
2
1
1
1
8
32
1
.
0
0
4
6
0
1
5
1
2
2
2
0
1
6
62
64
13
0
1
.0
0
T
h
e
p
u
r
p
o
s
e
o
f
m
o
d
el
ev
alu
at
io
n
is
to
ass
es
s
h
o
w
well
a
m
o
d
el
will
p
er
f
o
r
m
o
n
n
ew,
u
n
s
ee
n
d
ata,
k
n
o
wn
as
o
u
t
-
of
-
s
am
p
le
d
ata.
I
t
is
cr
u
cial
n
o
t
to
ev
alu
ate
a
m
o
d
el
o
n
th
e
s
am
e
d
ata
u
s
ed
f
o
r
tr
ain
in
g
b
ec
a
u
s
e
th
is
ca
n
lead
to
o
v
er
f
itti
n
g
[2
7
]
,
wh
er
e
th
e
m
o
d
el
m
e
m
o
r
iz
es
th
e
tr
ain
in
g
d
ata
r
ath
e
r
th
a
n
lear
n
in
g
g
e
n
er
al
p
atter
n
s
,
th
u
s
h
in
d
er
in
g
its
ab
ilit
y
to
g
en
er
alize
.
Af
ter
t
r
ain
in
g
o
n
t
h
e
tr
ain
in
g
s
et,
th
e
p
e
r
f
o
r
m
a
n
ce
o
f
ea
ch
m
o
d
el
is
ev
alu
ated
u
s
in
g
th
e
t
est s
et,
f
o
cu
s
in
g
o
n
th
r
ee
k
e
y
m
etr
ics:
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
a
n
d
r
ec
all.
−
L
o
g
is
tic
r
eg
r
ess
io
n
:
th
e
lo
g
is
t
ic
r
eg
r
ess
io
n
alg
o
r
ith
m
ac
h
ie
v
ed
an
ac
c
u
r
ac
y
o
f
4
8
%,
in
d
i
ca
tin
g
m
o
d
e
r
ate
p
er
f
o
r
m
an
ce
.
−
L
in
ea
r
SVC
:
th
e
lin
ea
r
SVC
a
lg
o
r
ith
m
also
ac
h
ie
v
ed
an
ac
c
u
r
ac
y
o
f
4
8
%,
s
u
g
g
esti
n
g
s
i
m
ilar
p
er
f
o
r
m
an
ce
to
lo
g
is
tic
r
eg
r
ess
io
n
.
−
SVM:
t
h
e
SVM
alg
o
r
ith
m
ac
h
iev
ed
a
h
ig
h
ac
cu
r
ac
y
o
f
9
4
%,
in
d
icatin
g
s
tr
o
n
g
p
r
e
d
ictiv
e
c
ap
ab
ilit
y
.
−
KNN:
t
h
e
KNN
alg
o
r
ith
m
ac
h
iev
ed
a
v
er
y
h
ig
h
ac
cu
r
ac
y
o
f
9
7
%,
d
em
o
n
s
tr
atin
g
ex
ce
llen
t
p
er
f
o
r
m
an
ce
.
−
B
er
n
o
u
lli
NB
:
t
h
e
B
er
n
o
u
ll
i
NB
alg
o
r
ith
m
ac
h
ie
v
ed
a
h
ig
h
ac
cu
r
ac
y
o
f
9
6
%,
in
d
icatin
g
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
in
class
if
icatio
n
t
ask
s
.
T
h
ese
ac
cu
r
ac
y
s
co
r
es
r
e
f
lec
t
h
o
w
well
ea
ch
alg
o
r
ith
m
g
en
er
alize
d
to
th
e
test
d
ata
a
f
ter
b
ein
g
tr
ain
ed
o
n
th
e
tr
ain
in
g
s
et.
Hig
h
er
ac
cu
r
ac
y
s
co
r
es
g
en
er
all
y
in
d
icate
b
etter
p
er
f
o
r
m
an
ce
,
with
SVM,
K
NN,
an
d
B
er
n
o
u
lli
NB
s
h
o
win
g
p
a
r
ticu
lar
ly
s
tr
o
n
g
r
esu
lts
in
th
is
ev
alu
atio
n
.
T
h
ese
m
etr
ics
p
r
o
v
id
e
in
s
ig
h
ts
in
to
th
e
ef
f
ec
tiv
en
ess
o
f
ea
c
h
alg
o
r
ith
m
in
m
ak
in
g
ac
cu
r
ate
p
r
ed
i
ctio
n
s
o
n
n
ew,
u
n
s
ee
n
d
ata
.
3
.
1
.
2
.
Su
mm
a
ry
t
a
ble
T
ab
le
2
s
u
m
m
ar
izes
th
e
p
r
ed
i
ctio
n
ac
cu
r
ac
y
o
f
th
e
al
g
o
r
ith
m
s
em
p
lo
y
ed
in
th
e
s
tu
d
y
.
A
m
o
n
g
th
e
m
,
KNN,
SVM,
an
d
B
er
n
o
u
lli
NB
d
em
o
n
s
tr
ate
n
o
tab
ly
s
u
p
er
i
o
r
p
er
f
o
r
m
a
n
ce
co
m
p
a
r
ed
to
l
in
ea
r
SVC
,
SVM
,
an
d
lo
g
is
tic
r
eg
r
ess
io
n
.
Sp
ec
if
ically
,
KNN
ac
h
iev
es
th
e
h
ig
h
est
ac
cu
r
ac
y
an
d
p
r
ec
is
io
n
,
ac
h
iev
in
g
ap
p
r
o
x
im
ately
9
7
%.
T
h
is
in
d
i
ca
tes
th
at
KNN
ef
f
ec
tiv
el
y
p
r
ed
icts
o
u
tco
m
es
with
a
h
ig
h
d
eg
r
ee
o
f
ac
cu
r
ac
y
co
m
p
ar
ed
to
th
e
o
th
er
al
g
o
r
it
h
m
s
test
ed
.
Ov
er
all,
th
e
T
a
b
le
2
u
n
d
e
r
s
co
r
es
th
e
e
f
f
ec
tiv
en
ess
o
f
KNN,
SVM,
an
d
B
er
n
o
u
lli
NB
in
ac
h
iev
in
g
r
o
b
u
s
t
p
r
ed
ictio
n
ac
c
u
r
ac
y
,
h
ig
h
lig
h
tin
g
th
eir
s
u
itab
ili
ty
f
o
r
th
e
s
p
ec
if
ic
class
if
icatio
n
task
s
ev
alu
ated
in
th
is
s
tu
d
y
.
T
ab
le
2
.
Pre
d
ictio
n
ac
cu
r
ac
y
o
f
o
u
r
al
g
o
r
ith
m
s
C
l
a
s
si
f
i
c
a
t
i
o
n
a
l
g
o
r
i
t
h
m
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
4
8
%
5
2
%
Li
n
e
a
r
S
V
C
4
8
%
5
2
%
S
V
M
9
4
%
9
0
%
B
e
r
n
o
u
l
l
i
N
B
9
6
%
9
3
%
K
N
N
9
7
%
9
7
%
I
n
s
u
m
m
a
r
y
,
o
u
r
s
tu
d
y
ev
alu
a
tes
s
ev
er
al
m
ac
h
in
e
lea
r
n
in
g
alg
o
r
ith
m
s
u
s
in
g
th
e
Scik
it
-
le
ar
n
lib
r
a
r
y
in
Py
th
o
n
.
I
n
Fig
u
r
e
3
r
esu
lts
h
ig
h
lig
h
t
v
a
r
y
in
g
lev
els
o
f
ac
cu
r
ac
y
an
d
p
r
ec
is
io
n
am
o
n
g
th
ese
alg
o
r
ith
m
s
,
as
d
ep
icted
in
Fig
u
r
e
s
3
(
a)
a
n
d
3
(
b
)
wh
ich
is
m
ad
e
u
p
o
f
two
g
r
ap
h
s
,
th
e
g
r
ey
g
r
ap
h
s
h
o
w
s
th
e
ac
cu
r
ac
y
a
n
d
th
e
b
lu
e
g
r
a
p
h
s
h
o
ws
th
e
p
r
ec
is
io
n
r
esu
ltin
g
f
r
o
m
th
e
a
p
p
lic
atio
n
o
f
d
if
f
er
en
t
m
ac
h
in
e
lea
r
n
in
g
alg
o
r
ith
m
s
to
o
u
r
test
b
ase.
Sp
ec
if
ically
,
K
NN,
B
er
n
o
u
lli
NB
,
an
d
SVM
d
em
o
n
s
tr
ate
s
ig
n
if
ican
tly
h
ig
h
er
ac
c
u
r
ac
y
r
ates,
ac
h
iev
in
g
a
p
p
r
o
x
im
ately
9
7
%
,
9
6
%,
a
n
d
9
4
%
r
esp
ec
tiv
ely
.
I
n
c
o
n
tr
ast,
lo
g
is
tic
r
eg
r
ess
io
n
an
d
lin
ea
r
SVC
ex
h
ib
it
co
m
p
ar
ativ
ely
lo
wer
a
cc
u
r
ac
y
in
o
u
r
ev
alu
atio
n
.
T
h
e
s
e
f
in
d
in
g
s
u
n
d
er
s
co
r
e
t
h
e
ef
f
ec
tiv
en
ess
o
f
KNN,
B
er
n
o
u
lli
NB
,
an
d
SVM
in
ac
c
u
r
ately
p
r
ed
ictin
g
o
u
t
co
m
es
f
o
r
th
e
task
s
co
n
s
id
er
ed
in
o
u
r
s
tu
d
y
.
T
h
ey
em
p
h
asize
th
e
im
p
o
r
ta
n
ce
o
f
s
elec
tin
g
ap
p
r
o
p
r
iate
m
ac
h
in
e
lear
n
i
n
g
al
g
o
r
ith
m
s
b
a
s
ed
o
n
t
h
e
s
p
ec
if
ic
ch
ar
ac
ter
is
tics
an
d
r
eq
u
ir
e
m
en
ts
o
f
th
e
d
ataset
an
d
th
e
p
r
o
b
le
m
d
o
m
ain
.
3
.
2
.
E
f
f
ec
t
o
f
un
s
up
er
v
is
ed
a
lg
o
rit
hm
s
I
n
itially
,
we
r
em
o
v
e
th
e
s
eg
m
en
tatio
n
o
f
o
u
r
d
ataset
b
y
ex
cl
u
d
in
g
t
h
e
“
is
_
f
ak
e
”
f
ield
.
T
o
a
ch
iev
e
th
e
h
ig
h
est
ac
cu
r
ac
y
f
o
r
o
u
r
m
o
d
el,
it
’
s
ess
en
tial
to
d
eter
m
in
e
th
e
o
p
tim
al
n
u
m
b
er
o
f
cl
u
s
ter
s
,
d
en
o
ted
as
k
,
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:
2
5
0
2
-
4
7
52
E
n
h
a
n
ci
n
g
fa
ke
p
r
o
file d
etec
tio
n
th
r
o
u
g
h
s
u
p
ervis
ed
a
n
d
h
y
b
r
id
…
(
I
s
ma
il B
en
s
a
s
s
i)
263
f
o
r
o
u
r
d
atas
et.
T
h
e
E
lb
o
w
m
e
th
o
d
[
2
8
]
-
[
3
0
]
is
wid
ely
u
s
ed
t
o
id
en
tify
t
h
is
o
p
tim
al
k
v
al
u
e.
Af
ter
tr
ain
in
g
o
u
r
m
o
d
el
with
v
ar
i
o
u
s
v
alu
es
o
f
k
u
s
in
g
b
o
th
th
e
tr
ai
n
in
g
a
n
d
test
in
g
d
atasets
,
we
d
eter
m
in
e
th
at
th
e
o
p
tim
al
k
v
alu
e
f
o
r
o
u
r
m
o
d
el
is
3
,
as
illu
s
tr
ated
in
Fig
u
r
e
4
.
T
h
is
ap
p
r
o
ac
h
h
elp
s
u
s
estab
lis
h
th
e
n
u
m
b
er
o
f
clu
s
ter
s
th
at
b
est o
r
g
an
izes o
u
r
d
ata
w
h
ile
m
ax
im
izin
g
th
e
ac
cu
r
ac
y
o
f
o
u
r
clu
s
ter
i
n
g
m
o
d
el.
Fig
u
r
e
5
d
is
p
lay
s
th
e
clu
s
ter
in
g
o
u
tco
m
e
o
f
o
u
r
d
ata,
wh
e
r
e
ca
s
es
ar
e
g
r
o
u
p
ed
i
n
to
th
r
ee
d
is
tin
ct
clu
s
ter
s
r
ep
r
esen
ted
b
y
c
o
lo
r
s
:
y
ello
w,
p
u
r
p
le,
a
n
d
g
r
ee
n
.
T
h
ese
clu
s
ter
s
ar
e
d
eter
m
in
e
d
b
ased
o
n
th
e
s
ix
p
ar
am
eter
s
d
ep
icted
in
T
a
b
le
1
.
T
h
is
v
is
u
aliza
tio
n
is
g
en
er
a
ted
u
s
in
g
th
e
m
atp
lo
tlib
lib
r
ar
y
in
Py
th
o
n
[3
1
]
,
[3
2
]
,
allo
win
g
u
s
to
v
is
u
ally
u
n
d
e
r
s
tan
d
h
o
w
o
u
r
d
ata
p
o
in
ts
ar
e
d
is
tr
ib
u
ted
ac
r
o
s
s
d
if
f
er
en
t
clu
s
ter
s
.
I
t
p
r
o
v
id
es
a
clea
r
r
e
p
r
esen
tati
o
n
o
f
h
o
w
th
e
cl
u
s
ter
in
g
alg
o
r
ith
m
h
as
o
r
g
a
n
ized
o
u
r
d
a
taset
b
ased
o
n
th
e
s
elec
ted
p
a
r
am
eter
s
.
(
a)
(
b
)
Fig
u
r
e
3
.
T
h
e
o
v
er
all
p
r
e
d
ictio
n
(
a)
ac
c
u
r
ac
y
a
n
d
(
b
)
p
r
ec
is
io
n
o
f
s
u
p
e
r
v
is
ed
alg
o
r
ith
m
s
Fig
u
r
e
4
.
E
r
r
o
r
r
ate
v
s
K
-
v
alu
e
Fig
u
r
e
5
.
Par
titi
o
n
o
f
ca
s
es in
to
th
r
ee
clu
s
ter
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
1
,
J
an
u
ar
y
20
25
:
257
-
2
6
8
264
3
.
2
.
1
.
Co
m
bin
a
t
io
n o
f
t
he
K
-
m
ea
ns
a
nd
t
he
s
up
er
v
is
e
d m
a
chine le
a
rning
a
lg
o
rit
hm
s
T
h
e
T
a
b
le
3
p
r
o
v
id
es
a
s
u
m
m
ar
y
o
f
th
e
im
p
ac
t
o
f
th
e
K
-
m
ea
n
s
alg
o
r
ith
m
o
n
th
e
p
er
f
o
r
m
an
ce
o
f
v
ar
io
u
s
alg
o
r
ith
m
s
p
r
e
v
io
u
s
l
y
ev
alu
ated
(
KNN,
SVM,
l
i
n
ea
r
SVC
,
B
er
n
o
u
lli
NB
,
l
o
g
is
tic
r
eg
r
ess
io
n
)
.
No
tab
ly
,
KNN,
B
er
n
o
u
lli
NB
,
an
d
SVM
d
em
o
n
s
tr
ate
s
ig
n
if
i
ca
n
tly
h
ig
h
e
r
ac
cu
r
ac
y
r
ates
a
n
d
p
r
ec
is
io
n
lev
els
(
9
7
%,
9
3
%,
a
n
d
9
0
%,
an
d
8
7
%,
8
6
%,
8
6
% r
esp
ec
tiv
ely
)
co
m
p
ar
ed
to
lo
g
is
tic
r
eg
r
ess
io
n
a
n
d
lin
ea
r
SVC
.
T
ab
le
3
.
Per
f
o
r
m
an
ce
an
d
p
r
e
cisi
o
n
c
o
m
p
ar
is
o
n
b
ef
o
r
e
an
d
af
ter
i
n
clu
s
io
n
o
f
K
-
m
ea
n
s
clu
s
ter
in
g
C
l
a
s
si
f
i
c
a
t
i
o
n
a
l
g
o
r
i
t
h
m
A
c
c
u
r
a
y
b
e
f
o
r
e
K
-
m
e
a
n
s
A
c
c
u
r
a
c
y
a
f
t
e
r
K
-
m
e
a
n
s
P
r
e
c
i
s
i
o
n
b
e
f
o
r
e
K
-
m
e
a
n
s
P
r
e
c
i
s
i
o
n
a
f
t
e
r
K
-
m
e
a
n
s
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
4
8
%
8
7
%
5
2
%
8
7
%
Li
n
e
a
r
S
V
C
4
8
%
8
7
%
5
2
%
8
7
%
S
V
M
9
4
%
8
6
%
9
0
%
8
6
%
B
e
r
n
o
u
l
l
i
NB
9
6
%
8
6
%
9
3
%
8
6
%
K
N
N
9
7
%
8
7
%
9
7
%
8
7
%
T
h
ese
r
esu
lts
u
n
d
er
s
co
r
e
th
e
ef
f
ec
tiv
en
ess
o
f
K
-
m
ea
n
s
clu
s
ter
in
g
in
en
h
an
cin
g
th
e
p
er
f
o
r
m
an
ce
o
f
class
if
icatio
n
alg
o
r
ith
m
s
,
p
ar
ticu
lar
ly
f
o
r
task
s
wh
er
e
g
r
o
u
p
in
g
d
ata
in
to
clu
s
ter
s
b
ased
o
n
s
im
ilar
ity
im
p
r
o
v
es
p
r
ed
ictiv
e
ac
c
u
r
ac
y
an
d
p
r
ec
is
io
n
.
T
h
e
y
h
ig
h
lig
h
t
th
e
im
p
o
r
tan
ce
o
f
in
t
eg
r
atin
g
cl
u
s
ter
in
g
tech
n
iq
u
es
lik
e
K
-
m
ea
n
s
to
o
p
tim
ize
th
e
o
u
tco
m
es
o
f
m
ac
h
in
e
lear
n
in
g
m
o
d
els
in
v
ar
io
u
s
ap
p
licatio
n
d
o
m
ain
s
as
in
d
icate
d
in
Fig
u
r
e
6
wh
ich
is
m
ad
e
u
p
o
f
two
g
r
ap
h
s
,
in
Fig
u
r
e
6
(
a)
th
e
g
r
ap
h
in
g
r
ay
s
ch
em
atize
s
th
e
ac
cu
r
ac
y
a
f
ter
ap
p
ly
i
n
g
b
o
th
th
e
K
m
ea
n
s
al
g
o
r
i
th
m
o
f
clu
s
ter
in
g
with
t
h
e
v
ar
io
u
s
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
ab
o
v
e
-
m
en
tio
n
e
d
,
to
o
u
r
te
s
t
b
ase
an
d
th
e
g
r
a
p
h
in
b
lu
e
s
ch
em
atize
s
th
e
p
r
ec
is
io
n
in
Fig
u
r
e
6
(
b
)
.
(
a)
(
b
)
Fig
u
r
e
6
.
T
h
e
o
v
er
all
p
r
e
d
ictio
n
(
a)
ac
c
u
r
ac
y
a
n
d
(
b
)
p
r
ec
is
io
n
o
f
s
u
p
e
r
v
is
ed
alg
o
r
ith
m
s
T
h
e
im
p
ac
t
o
f
t
h
e
K
-
m
ea
n
s
alg
o
r
ith
m
o
n
o
u
r
d
ata
is
s
tr
ik
in
g
,
p
ar
ticu
lar
ly
in
e
n
h
an
cin
g
b
o
th
p
r
ec
is
io
n
a
n
d
a
cc
u
r
ac
y
ac
r
o
s
s
s
ev
er
al
alg
o
r
ith
m
s
th
at
in
itially
y
ield
ed
lo
wer
r
esu
lts
.
T
h
is
im
p
r
o
v
em
en
t
is
esp
ec
ially
n
o
tab
le
f
o
r
alg
o
r
ith
m
s
th
at
p
r
ev
io
u
s
ly
s
h
o
wed
lo
wer
p
er
f
o
r
m
a
n
ce
m
etr
ics.
K
-
m
ea
n
s
clu
s
ter
in
g
h
as
ef
f
ec
tiv
ely
co
n
tr
i
b
u
ted
to
r
e
f
in
in
g
th
e
p
r
e
d
ictio
n
s
an
d
class
if
icatio
n
s
m
ad
e
b
y
th
ese
alg
o
r
ith
m
s
,
h
ig
h
lig
h
tin
g
its
r
o
le
in
o
p
tim
izin
g
m
o
d
el
o
u
tco
m
es th
r
o
u
g
h
b
etter
d
ata
o
r
g
an
izatio
n
an
d
clu
s
ter
-
b
ased
a
n
aly
s
is
.
3
.
2
.
2
.
Co
m
bin
a
t
io
n o
f
t
he
K
-
m
edo
ids
a
nd
t
he
s
up
er
v
i
s
ed
m
a
chine le
a
rning
a
lg
o
rit
hm
s
T
ab
le
4
p
r
o
v
id
es
a
s
u
m
m
ar
y
o
f
h
o
w
th
e
K
-
m
e
d
o
id
s
alg
o
r
ith
m
in
f
lu
e
n
ce
s
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
alg
o
r
ith
m
s
an
aly
ze
d
i
n
th
e
p
r
e
v
io
u
s
s
ec
tio
n
(
KNN,
SVM,
l
in
ea
r
SVC
,
B
er
n
o
u
lli
NB
,
an
d
l
o
g
is
tic
r
eg
r
ess
io
n
).
I
n
d
ee
d
,
KNN,
B
er
n
o
u
lli
NB
,
an
d
SVM
d
em
o
n
s
tr
ate
s
ig
n
if
i
ca
n
tly
h
ig
h
er
ac
cu
r
ac
y
a
n
d
p
r
ec
is
io
n
r
ates
(
9
7
%,
9
3
%,
an
d
9
0
%,
an
d
9
9
%,
9
5
%
,
an
d
9
0
% r
esp
ec
tiv
ely
)
c
o
m
p
ar
ed
to
lo
g
is
tic
r
eg
r
ess
io
n
an
d
lin
ea
r
SVC
.
T
ab
le
4
.
Per
f
o
r
m
an
ce
an
d
p
r
e
cisi
o
n
c
o
m
p
ar
is
o
n
b
ef
o
r
e
an
d
af
ter
i
n
clu
s
io
n
o
f
K
-
m
ed
o
id
s
c
lu
s
ter
in
g
C
l
a
s
si
f
i
c
a
t
i
o
n
a
l
g
o
r
i
t
h
m
A
c
c
u
r
a
y
b
e
f
o
r
e
K
-
m
e
d
o
i
d
s
A
c
c
u
r
a
c
y
a
f
t
e
r
K
-
m
e
d
o
i
d
s
P
r
e
c
i
s
i
o
n
b
e
f
o
r
e
K
-
m
e
d
o
i
d
s
P
r
e
c
i
s
i
o
n
a
f
t
e
r
K
-
m
e
d
o
i
d
s
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
4
8
%
9
1
%
5
2
%
9
1
%
Li
n
e
a
r
S
V
C
4
8
%
9
1
%
5
2
%
9
1
%
S
V
M
9
4
%
9
0
%
9
0
%
9
0
%
B
e
r
n
o
u
l
l
i
NB
9
6
%
9
5
%
9
3
%
9
5
%
K
N
N
9
7
%
9
9
%
9
7
%
9
9
%
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:
2
5
0
2
-
4
7
52
E
n
h
a
n
ci
n
g
fa
ke
p
r
o
file d
etec
tio
n
th
r
o
u
g
h
s
u
p
ervis
ed
a
n
d
h
y
b
r
id
…
(
I
s
ma
il B
en
s
a
s
s
i)
265
T
h
ese
r
esu
lts
u
n
d
er
s
co
r
e
th
e
i
m
p
ac
tf
u
l
r
o
le
o
f
th
e
K
-
m
ed
o
i
d
s
alg
o
r
ith
m
in
en
h
an
cin
g
t
h
e
p
r
ed
ictiv
e
ca
p
ab
ilit
ies
o
f
th
ese
alg
o
r
ith
m
s
.
B
y
f
o
cu
s
in
g
o
n
m
ed
o
id
s
r
at
h
er
th
an
ce
n
tr
o
id
s
,
K
-
m
e
d
o
id
s
ef
f
ec
tiv
ely
r
e
f
in
es
clu
s
ter
in
g
d
ec
is
io
n
s
,
t
h
er
eb
y
o
p
tim
izin
g
t
h
e
p
e
r
f
o
r
m
an
ce
o
f
c
lass
if
icatio
n
alg
o
r
ith
m
s
in
v
ar
io
u
s
d
ata
-
d
r
iv
e
n
task
s
wh
ich
ex
p
lain
s
th
e
v
alu
es
s
h
o
wn
in
Fig
u
r
e
7
.
W
h
ich
is
m
ad
e
u
p
o
f
two
g
r
ap
h
s
,
in
t
h
e
Fig
u
r
e
7
(
a)
g
r
ay
s
ch
em
atize
s
th
e
ac
cu
r
ac
y
af
t
er
ap
p
l
y
in
g
b
o
th
th
e
K
-
m
ed
o
id
s
as
u
n
s
u
p
er
v
is
ed
alg
o
r
ith
m
s
with
th
e
v
ar
io
u
s
s
u
p
er
v
is
ed
m
ac
h
i
n
e
lear
n
in
g
a
lg
o
r
ith
m
s
ab
o
v
e
-
m
e
n
tio
n
ed
,
t
o
o
u
r
test
b
ase
an
d
th
e
g
r
ap
h
in
b
lu
e
s
ch
e
m
atize
s
th
e
p
r
ec
is
io
n
in
Fig
u
r
e
7
(
b
)
.
(
a)
(
b
)
Fig
u
r
e
7
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
(
a)
a
cc
u
r
ac
y
an
d
(
b
)
p
r
ec
is
io
n
b
ef
o
r
e
a
n
d
af
te
r
K
-
m
ed
o
id
s
alg
o
r
ith
m
T
h
ese
f
in
d
in
g
s
in
d
icate
th
at
in
co
r
p
o
r
atin
g
clu
s
ter
in
g
m
eth
o
d
s
,
p
ar
ticu
lar
ly
K
-
m
ea
n
s
an
d
K
-
m
ed
o
id
s
,
p
o
s
itiv
ely
in
f
lu
e
n
ce
s
th
e
p
r
e
d
ictiv
e
ca
p
ab
ilit
y
o
f
class
if
icatio
n
alg
o
r
ith
m
s
f
o
r
id
en
tif
y
in
g
f
a
k
e
p
r
o
f
iles
.
T
h
e
s
u
b
s
tan
tial
en
h
a
n
ce
m
en
ts
in
ac
cu
r
ac
y
an
d
p
r
ec
is
io
n
u
n
d
er
s
co
r
e
th
e
ef
f
ec
tiv
en
ess
o
f
cl
u
s
ter
in
g
tech
n
i
q
u
es
in
b
o
ls
ter
in
g
th
e
o
v
er
all
p
er
f
o
r
m
a
n
ce
o
f
m
ac
h
i
n
e
lear
n
in
g
m
o
d
els.
Fig
u
r
e
8
illu
s
tr
ates
th
e
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
o
f
th
is
h
y
b
r
i
d
ap
p
r
o
ac
h
,
h
i
g
h
lig
h
tin
g
its
ef
f
icac
y
in
ac
cu
r
ately
class
i
f
y
in
g
an
d
id
en
tify
in
g
f
ak
e
p
r
o
f
iles
with
in
s
o
cial
n
etwo
r
k
s
wh
er
e
th
e
Fig
u
r
e
8
(
a
)
g
r
ey
s
h
o
ws
th
e
r
esu
ltin
g
u
p
p
er
ac
c
u
r
ac
y
b
y
ap
p
l
y
in
g
th
e
K
-
m
ed
o
i
d
s
alg
o
r
ith
m
wit
h
th
e
KNN
as
s
u
p
e
r
v
is
ed
m
a
ch
in
e
lear
n
i
n
g
alg
o
r
ith
m
,
an
d
th
e
g
r
ap
h
in
b
lu
e
s
h
o
ws in
Fig
u
r
e
8
(
b
)
t
h
e
p
r
ec
i
s
io
n
o
f
th
e
latter
.
(
a)
(
b
)
Fig
u
r
e
8
.
T
h
e
KNN
alg
o
r
ith
m
(
a)
a
cc
u
r
ac
y
an
d
(
b
)
p
r
ec
is
io
n
b
ef
o
r
e
a
n
d
af
te
r
K
-
m
ed
o
id
s
alg
o
r
ith
m
I
n
o
u
r
s
p
ec
if
ic
d
ataset,
th
e
o
p
t
im
al
r
esu
lts
in
ter
m
s
o
f
ac
cu
r
a
cy
an
d
p
r
ec
is
io
n
ar
e
ac
h
iev
ed
th
r
o
u
g
h
a
h
y
b
r
id
ap
p
r
o
ac
h
:
u
s
in
g
K
-
m
ed
o
id
s
as
a
n
u
n
s
u
p
er
v
is
ed
a
lg
o
r
ith
m
to
p
ar
titi
o
n
th
e
d
ataset
in
to
co
h
esiv
e
clu
s
ter
s
,
an
d
t
h
en
e
m
p
lo
y
i
n
g
KNN
as
a
s
u
p
er
v
is
ed
al
g
o
r
it
h
m
f
o
r
d
etec
tin
g
f
ak
e
p
r
o
f
ile
s
.
T
h
is
is
wh
y
o
u
r
ap
p
r
o
ac
h
is
s
o
im
p
o
r
tan
t c
o
m
p
ar
ed
to
ex
is
tin
g
ap
p
r
o
ac
h
es i
n
ter
m
s
o
f
ac
cu
r
ac
y
,
with
a
s
ig
n
if
ican
t ju
m
p
in
th
e
v
alu
e
o
f
t
h
e
latter
,
ap
p
r
o
ac
h
in
g
1
0
0
%
with
a
v
alu
e
o
f
9
9
%.
As
a
r
esu
lt,
th
e
o
b
jectiv
e
o
f
m
ar
r
y
in
g
s
u
p
er
v
is
ed
an
d
u
n
s
u
p
er
v
is
ed
alg
o
r
ith
m
s
i
s
ac
h
iev
ed
,
an
d
o
u
r
s
tu
d
ies g
u
ar
an
tee
g
o
o
d
r
e
s
u
lts
in
p
r
ed
ict
in
g
f
ak
e
p
r
o
f
iles
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
1
,
J
an
u
ar
y
20
25
:
257
-
2
6
8
266
4.
CO
NCLU
SI
O
N
AND
P
E
RS
P
E
CT
I
VE
S
T
h
e
s
tu
d
y
p
r
esen
ted
a
co
m
p
r
e
h
en
s
iv
e
m
eth
o
d
o
l
o
g
y
aim
ed
a
t
d
etec
tin
g
f
ak
e
p
r
o
f
iles
o
n
s
o
cial
m
ed
ia
p
latf
o
r
m
s
b
y
lev
er
a
g
in
g
m
u
ltip
le
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
.
T
h
e
r
esear
ch
in
v
o
lv
e
d
a
n
ex
ten
s
iv
e
ex
p
lo
r
atio
n
o
f
v
ar
i
o
u
s
u
s
er
b
eh
a
v
io
r
s
,
en
g
ag
em
en
t
m
etr
ics,
an
d
co
n
ten
t
attr
i
b
u
tes
to
co
n
s
tr
u
ct
r
o
b
u
s
t
m
o
d
els
ca
p
a
b
le
o
f
ac
cu
r
ately
id
en
tify
in
g
d
ec
ep
tiv
e
p
r
o
f
iles
.
I
n
itially
,
th
e
d
ataset
u
n
d
er
wen
t
clu
s
ter
in
g
u
s
in
g
u
n
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
s
u
ch
as
K
-
m
ea
n
s
in
itially
an
d
s
u
b
s
eq
u
en
tly
K
-
m
ed
o
id
s
,
g
r
o
u
p
in
g
s
im
ilar
p
r
o
f
iles
b
ase
d
o
n
s
h
ar
e
d
c
h
ar
ac
ter
is
tics
.
T
h
is
clu
s
ter
in
g
p
r
o
ce
s
s
f
ac
ilit
ated
th
e
id
en
tific
atio
n
o
f
co
h
esiv
e
g
r
o
u
p
s
with
in
th
e
d
ataset,
en
h
an
cin
g
th
e
ef
f
icie
n
cy
o
f
s
u
b
s
eq
u
e
n
t c
lass
if
icatio
n
task
s
.
Fo
llo
win
g
clu
s
ter
in
g
,
class
if
icatio
n
task
s
wer
e
p
er
f
o
r
m
ed
u
s
in
g
a
r
a
n
g
e
o
f
s
u
p
e
r
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
in
clu
d
i
n
g
KNN,
SVM,
B
er
n
o
u
lli
NB
,
lo
g
is
tic
r
eg
r
ess
io
n
,
an
d
l
in
ea
r
SVC
.
T
h
ese
alg
o
r
ith
m
s
u
tili
ze
d
th
e
s
im
ilar
ities
in
p
r
o
f
ile
f
ea
tu
r
es to
e
f
f
e
ctiv
ely
class
if
y
p
r
o
f
iles
as e
ith
er
g
en
u
i
n
e
o
r
f
ak
e.
T
h
e
r
esu
lts
o
f
th
e
s
tu
d
y
d
em
o
n
s
tr
ated
th
e
ef
f
icac
y
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
in
d
etec
tin
g
f
ak
e
p
r
o
f
iles
,
ac
h
ie
v
in
g
p
r
o
m
is
in
g
o
u
tco
m
es
in
te
r
m
s
o
f
ac
cu
r
a
cy
an
d
r
eliab
ilit
y
.
B
y
co
m
b
in
in
g
clu
s
ter
in
g
an
d
class
if
icatio
n
tech
n
iq
u
es,
th
e
m
o
d
el
ef
f
ec
tiv
ely
d
is
tin
g
u
is
h
ed
d
ec
ep
tiv
e
p
r
o
f
iles
f
r
o
m
g
e
n
u
in
e
o
n
es,
o
f
f
er
in
g
v
alu
ab
le
in
s
ig
h
ts
f
o
r
co
m
b
atin
g
f
r
au
d
u
len
t a
ctiv
ities
o
n
s
o
cia
l m
ed
ia
p
latf
o
r
m
s
.
L
o
o
k
in
g
f
o
r
war
d
,
f
u
t
u
r
e
r
esea
r
ch
will
f
o
cu
s
o
n
f
u
r
th
er
v
ali
d
atin
g
th
e
m
o
d
el
u
s
in
g
lar
g
er
an
d
m
o
r
e
d
iv
er
s
e
d
atasets
co
llected
f
r
o
m
v
ar
io
u
s
s
o
cial
m
ed
ia
p
latf
o
r
m
s
.
Ad
d
itio
n
ally
,
alter
n
ativ
e
m
ac
h
in
e
lear
n
i
n
g
an
d
d
ee
p
lear
n
in
g
a
p
p
r
o
ac
h
es
will
b
e
ex
p
lo
r
ed
to
e
n
h
an
ce
d
etec
tio
n
ac
cu
r
ac
y
a
n
d
r
o
b
u
s
t
n
ess
.
T
h
is
in
clu
d
es
in
v
esti
g
atin
g
en
s
em
b
le
lear
n
in
g
m
eth
o
d
s
,
n
e
u
r
al
n
etwo
r
k
ar
ch
itectu
r
es,
an
d
ad
v
a
n
ce
d
f
ea
tu
r
e
en
g
in
ee
r
in
g
tech
n
iq
u
es.
Mo
r
e
o
v
er
,
f
u
tu
r
e
s
tu
d
ies will a
im
to
d
ev
elo
p
a
d
a
p
tiv
e
m
o
d
els ca
p
a
b
le
o
f
d
y
n
a
m
ically
ad
ju
s
tin
g
to
ev
o
lv
in
g
p
atter
n
s
o
f
f
a
k
e
p
r
o
f
i
le
cr
ea
tio
n
an
d
d
etec
tio
n
.
B
y
i
n
teg
r
atin
g
r
ea
l
-
tim
e
d
ata
a
n
aly
s
is
an
d
co
n
tin
u
o
u
s
m
o
d
el
r
ef
in
em
e
n
t,
th
e
o
b
jectiv
e
is
to
cr
ea
te
m
o
r
e
r
esil
ien
t
d
etec
tio
n
s
y
s
tem
s
th
at
ca
n
ef
f
ec
tiv
ely
m
itig
ate
th
e
s
p
r
ea
d
o
f
f
ak
e
p
r
o
f
iles
o
n
s
o
cial
m
ed
ia
p
latf
o
r
m
s
.
I
n
co
n
clu
s
i
o
n
,
th
e
s
tu
d
y
estab
lis
h
es
a
f
o
u
n
d
atio
n
f
o
r
o
n
g
o
in
g
r
esear
c
h
in
th
e
f
ield
o
f
f
ak
e
p
r
o
f
ile
d
etec
tio
n
,
illu
s
tr
atin
g
th
e
p
o
t
en
tial
o
f
m
ac
h
in
e
lear
n
in
g
a
lg
o
r
ith
m
s
to
ad
d
r
ess
th
is
p
er
s
is
ten
t
ch
allen
g
e.
T
h
r
o
u
g
h
co
n
tin
u
o
u
s
ad
v
a
n
ce
m
en
ts
an
d
r
esear
ch
ef
f
o
r
ts
,
th
e
g
o
al
is
to
d
ev
elo
p
s
o
p
h
is
ticated
d
etec
tio
n
s
y
s
tem
s
th
at
u
p
h
o
ld
th
e
in
teg
r
ity
an
d
tr
u
s
two
r
th
in
ess
o
f
s
o
ci
al
m
ed
ia
p
latf
o
r
m
s
in
th
e
f
ac
e
o
f
ev
o
lv
i
n
g
th
r
ea
ts
p
o
s
ed
b
y
d
ec
ep
tiv
e
en
titi
es.
T
h
ese
s
y
s
tem
s
co
u
ld
also
b
e
ap
p
lied
ac
r
o
s
s
v
ar
io
u
s
d
o
m
ain
s
,
in
clu
d
in
g
s
o
ci
al
n
etwo
r
k
s
,
e
-
co
m
m
er
ce
p
latf
o
r
m
s
,
f
o
r
u
m
s
,
a
n
d
b
a
n
k
in
g
web
s
ites
.
I
n
th
e
f
u
tu
r
e,
we
h
o
p
e
t
o
u
s
e
o
th
er
m
ac
h
in
e
an
d
d
e
e
p
lear
n
in
g
tech
n
i
q
u
e
s
an
d
ap
p
ly
o
u
r
ap
p
r
o
ac
h
in
o
th
er
r
ele
v
an
t
ar
ea
s
th
at
af
f
ec
t
p
eo
p
le
’
s
liv
es,
in
o
r
d
e
r
to
m
in
im
ize
attac
k
s
,
s
ca
m
s
an
d
f
r
a
u
d
in
b
ig
d
ata
s
y
s
tem
s
.
RE
F
E
R
E
NC
E
S
[
1
]
D
.
R
a
m
a
l
i
n
g
a
m
a
n
d
V
.
C
h
i
n
n
a
i
a
h
,
“
F
a
k
e
p
r
o
f
i
l
e
d
e
t
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s
i
n
l
a
r
g
e
-
sca
l
e
o
n
l
i
n
e
s
o
c
i
a
l
n
e
t
w
o
r
k
s
:
a
c
o
m
p
r
e
h
e
n
s
i
v
e
r
e
v
i
e
w
,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
6
5
,
p
p
.
1
6
5
–
1
7
7
,
Ja
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
e
l
e
c
e
n
g
.
2
0
1
7
.
0
5
.
0
2
0
.
[
2
]
G
.
H
a
j
d
u
,
Y
.
M
i
n
o
s
o
,
R
.
L
o
p
e
z
,
M
.
A
c
o
st
a
,
a
n
d
A
.
El
l
e
i
t
h
y
,
“
U
se
o
f
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
t
o
i
d
e
n
t
i
f
y
f
a
k
e
p
r
o
f
i
l
e
s
,
”
i
n
2
0
1
9
I
EEE
L
o
n
g
I
s
l
a
n
d
S
y
s
t
e
m
s,
A
p
p
l
i
c
a
t
i
o
n
s
a
n
d
T
e
c
h
n
o
l
o
g
y
C
o
n
f
e
re
n
c
e
,
L
I
S
AT
2
0
1
9
,
M
a
y
2
0
1
9
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
LI
S
A
T.
2
0
1
9
.
8
8
1
7
3
3
0
.
[
3
]
M
.
M
.
S
w
e
a
n
d
N
.
N
.
M
y
o
,
“
F
a
k
e
a
c
c
o
u
n
t
s
d
e
t
e
c
t
i
o
n
o
n
t
w
i
t
t
e
r
u
si
n
g
b
l
a
c
k
l
i
st
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
1
7
t
h
I
EEE
/
A
C
I
S
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
,
I
C
I
S
2
0
1
8
,
J
u
n
.
2
0
1
8
,
p
p
.
5
6
2
–
5
6
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
S
.
2
0
1
8
.
8
4
6
6
4
9
9
.
[
4
]
P
.
W
a
n
d
a
a
n
d
H
.
J.
Ji
e
,
“
D
e
e
p
P
r
o
f
i
l
e
:
F
i
n
d
i
n
g
f
a
k
e
p
r
o
f
i
l
e
i
n
o
n
l
i
n
e
s
o
c
i
a
l
n
e
t
w
o
r
k
u
s
i
n
g
d
y
n
a
m
i
c
C
N
N
,
”
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
S
e
c
u
r
i
t
y
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
5
2
,
p
.
1
0
2
4
6
5
,
Ju
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
i
s
a
.
2
0
2
0
.
1
0
2
4
6
5
.
[
5
]
S
.
K
o
d
a
t
i
,
K
.
P
.
R
e
d
d
y
,
S
.
M
e
k
a
l
a
,
P
.
L.
S
.
M
u
r
t
h
y
,
a
n
d
P
.
C
.
S
.
R
e
d
d
y
,
“
D
e
t
e
c
t
i
o
n
o
f
f
a
k
e
p
r
o
f
i
l
e
s
o
n
t
w
i
t
t
e
r
u
si
n
g
h
y
b
r
i
d
S
V
M
a
l
g
o
r
i
t
h
m,
”
E
3
S
W
e
b
o
f
C
o
n
f
e
re
n
c
e
s
,
v
o
l
.
3
0
9
,
p
.
0
1
0
4
6
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
5
1
/
e
3
s
c
o
n
f
/
2
0
2
1
3
0
9
0
1
0
4
6
.
[
6
]
E.
P
.
M
e
s
h
r
a
m,
R
.
B
h
a
m
b
u
l
k
a
r
,
P
.
P
o
k
a
l
e
,
K
.
K
h
a
r
b
i
k
a
r
,
a
n
d
A
.
A
w
a
c
h
a
t
,
“
A
u
t
o
ma
t
i
c
d
e
t
e
c
t
i
o
n
o
f
f
a
k
e
p
r
o
f
i
l
e
u
si
n
g
ma
c
h
i
n
e
le
a
r
n
i
n
g
o
n
I
n
st
a
g
r
a
m,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
S
c
i
e
n
t
i
f
i
c
R
e
se
a
rc
h
i
n
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
p
p
.
1
1
7
–
1
2
7
,
M
a
y
2
0
2
1
,
d
o
i
:
1
0
.
3
2
6
2
8
/
i
j
sr
st
2
1
8
3
3
0
.
[
7
]
P
.
C
h
a
k
r
a
b
o
r
t
y
,
C
.
S
.
M
u
z
a
mm
e
l
,
M
.
K
h
a
t
u
n
,
S
.
F
.
I
sl
a
m,
a
n
d
S
.
R
a
h
man
,
“
A
u
t
o
ma
t
i
c
st
u
d
e
n
t
a
t
t
e
n
d
a
n
c
e
sy
st
e
m
u
si
n
g
f
a
c
e
r
e
c
o
g
n
i
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
A
d
v
a
n
c
e
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
9
,
n
o
.
3
,
p
p
.
9
3
–
9
9
,
F
e
b
.
2
0
2
0
,
d
o
i
:
1
0
.
3
5
9
4
0
/
i
j
e
a
t
.
b
4
2
0
7
.
0
2
9
3
2
0
.
[
8
]
M
.
A
h
m
e
d
,
P
.
C
h
a
k
r
a
b
o
r
t
y
,
a
n
d
T
.
C
h
o
u
d
h
u
r
y
,
“
B
a
n
g
l
a
d
o
c
u
m
e
n
t
c
a
t
e
g
o
r
i
z
a
t
i
o
n
u
si
n
g
d
e
e
p
R
N
N
m
o
d
e
l
w
i
t
h
a
t
t
e
n
t
i
o
n
mec
h
a
n
i
s
m
,
”
i
n
L
e
c
t
u
re
N
o
t
e
s
i
n
N
e
t
w
o
r
k
s
a
n
d
S
y
s
t
e
m
s
,
v
o
l
.
2
9
1
,
2
0
2
2
,
p
p
.
1
3
7
–
1
4
7
.
[
9
]
Z.
R
u
st
a
m,
S
.
H
a
r
t
i
n
i
,
I
.
W
i
r
a
sa
t
i
,
a
n
d
J.
E
.
A
u
r
e
l
i
a
,
“
A
n
a
p
p
r
o
a
c
h
f
o
r
C
O
V
I
D
-
1
9
d
e
t
e
c
t
i
o
n
u
si
n
g
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
f
e
a
t
u
r
e
s
o
n
c
h
e
st
X
-
r
a
y
i
ma
g
e
s
,
”
J
o
u
r
n
a
l
o
f
T
h
e
o
ret
i
c
a
l
a
n
d
A
p
p
l
i
e
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
9
9
,
n
o
.
6
,
p
p
.
1
4
5
2
–
1
4
6
0
,
2
0
2
1
.
[1
0
]
A
.
O
u
h
m
i
d
a
,
A
.
R
a
i
h
a
n
i
,
B
.
C
h
e
r
r
a
d
i
,
a
n
d
O
.
T
e
r
r
a
d
a
,
“
A
n
o
v
e
l
a
p
p
r
o
a
c
h
f
o
r
p
a
r
k
i
n
so
n
’
s
d
i
sea
se
d
e
t
e
c
t
i
o
n
b
a
s
e
d
o
n
v
o
i
c
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
a
n
d
f
e
a
t
u
r
e
s
s
e
l
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
O
n
l
i
n
e
a
n
d
B
i
o
m
e
d
i
c
a
l
En
g
i
n
e
e
ri
n
g
(
i
J
O
E)
,
v
o
l
.
1
7
,
n
o
.
1
0
,
p
p
.
1
1
1
–
1
3
0
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
3
9
9
1
/
i
j
o
e
.
v
1
7
i
1
0
.
2
4
4
9
9
.
[1
1
]
L.
W
a
n
g
e
t
a
l
.
,
“
A
n
i
n
t
e
l
l
i
g
e
n
t
g
a
st
r
i
c
c
a
n
c
e
r
s
c
r
e
e
n
i
n
g
me
t
h
o
d
b
a
se
d
o
n
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
a
n
d
su
p
p
o
r
t
v
e
c
t
o
r
mac
h
i
n
e
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
4
3
,
n
o
.
7
,
p
p
.
7
2
0
–
7
2
5
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
0
/
1
2
0
6
2
1
2
X
.
2
0
1
9
.
1
6
4
0
3
4
5
.
[1
2
]
Z.
A
.
K
a
k
a
r
a
s
h
,
H
.
S
.
Ez
a
t
,
S
.
A
.
O
mar,
a
n
d
N
.
F
.
A
h
m
e
d
,
“
T
i
me
ser
i
e
s
f
o
r
e
c
a
s
t
i
n
g
b
a
s
e
d
o
n
s
u
p
p
o
r
t
v
e
c
t
o
r
mac
h
i
n
e
u
s
i
n
g
p
a
r
t
i
c
l
e
sw
a
r
m
o
p
t
i
m
i
z
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
1
,
n
o
.
1
,
p
p
.
7
6
–
8
8
,
M
a
r
.
2
0
2
2
,
d
o
i
:
1
0
.
4
7
8
3
9
/
i
j
c
.
2
1
.
1
.
2
5
2
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.