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.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
,
p
p
.
60
1
~
613
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
41
.
i
2
.
pp
60
1
-
6
1
3
601
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Fraud de
tectio
n u
sing
Ta
bNet* cla
s
sifier:
a
ma
chine learnin
g
appro
a
ch
G
.
Anis
h M
a
ry
,
S.
Su
dh
a
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
A
p
p
l
i
c
a
t
i
o
n
s,
H
i
n
d
u
s
t
a
n
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
a
n
d
S
c
i
e
n
c
e
,
C
h
e
n
n
a
i
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
11
,
2
0
2
5
R
ev
is
ed
Dec
5
,
202
5
Acc
ep
ted
J
an
11
,
2
0
2
6
De
tec
ti
n
g
fra
u
d
u
len
t
tran
sa
c
ti
o
n
s
is
a
b
ig
c
h
a
ll
e
n
g
e
in
t
h
e
d
i
g
it
a
l
fin
a
n
c
ial
wo
rld
.
Tran
sa
c
ti
o
n
v
o
lu
m
e
s
a
re
g
ro
wi
n
g
q
u
ick
l
y
,
a
n
d
n
e
w
a
tt
a
c
k
m
e
th
o
d
s
o
ften
o
u
tstri
p
trad
it
i
o
n
a
l
d
e
tec
ti
o
n
sy
ste
m
s.
Cu
rre
n
t
fra
u
d
-
d
e
tec
ti
o
n
m
o
d
e
ls
u
su
a
ll
y
lac
k
c
larit
y
a
n
d
d
o
n
o
t
p
e
rfo
rm
re
li
a
b
l
y
o
n
u
n
b
a
lan
c
e
d
re
a
l
-
wo
rld
d
a
tas
e
ts.
Th
is
h
i
g
h
li
g
h
ts
th
e
u
rg
e
n
t
n
e
e
d
f
o
r
c
lea
r
a
n
d
e
x
p
lain
a
b
le
d
e
e
p
-
lea
rn
in
g
m
e
th
o
d
s
f
o
r
tab
u
lar
fin
a
n
c
ial
d
a
ta.
T
h
is
p
a
p
e
r
p
re
se
n
ts
a
n
in
terp
re
tab
le
d
e
e
p
lea
rn
i
n
g
fra
m
e
wo
rk
b
u
il
t
o
n
t
h
e
Tab
Ne
t
c
las
sifier.
I
t
u
se
s
a
tt
e
n
ti
o
n
-
d
ri
v
e
n
fe
a
tu
re
se
lec
ti
o
n
,
s
p
a
rse
re
p
re
se
n
tatio
n
lea
rn
in
g
,
a
n
d
se
q
u
e
n
ti
a
l
d
e
c
isio
n
re
a
so
n
i
n
g
to
m
o
d
e
l
c
o
m
p
lex
i
n
tera
c
ti
o
n
s
a
m
o
n
g
tran
sa
c
ti
o
n
a
l,
d
e
m
o
g
ra
p
h
ic,
a
n
d
g
e
o
g
ra
p
h
ica
l
fa
c
to
rs.
T
h
e
m
o
d
e
l
wa
s
tes
ted
o
n
a
re
a
l
-
wo
r
ld
c
re
d
i
t
c
a
rd
tra
n
sa
c
t
io
n
d
a
tas
e
t
with
2
3
fe
a
tu
re
s.
I
t
a
c
h
iev
e
d
9
9
.
6
9
%
a
c
c
u
ra
c
y
,
a
0
.
9
7
5
F
1
-
sc
o
re
,
a
n
d
a
0
.
9
5
6
ROC
-
A
UC.
Th
is
p
e
rfo
rm
a
n
c
e
o
u
tp
e
rfo
rm
s
b
e
n
c
h
m
a
rk
m
o
d
e
ls
su
c
h
a
s
ra
n
d
o
m
fo
re
st
,
XG
Bo
o
st,
Li
g
h
tG
BM
,
a
n
d
lo
g
is
ti
c
re
g
re
ss
io
n
.
In
a
d
d
it
i
o
n
to
o
u
tstan
d
i
n
g
p
re
d
ictiv
e
re
su
lt
s
.
F
u
rt
h
e
rm
o
re
,
in
terp
re
tab
il
it
y
is
e
n
h
a
n
c
e
d
b
y
Tab
Ne
t'
s
a
tt
e
n
ti
o
n
-
b
a
se
d
fe
a
tu
re
a
tt
rib
u
ti
o
n
.
Th
is
fa
c
il
it
a
tes
th
e
c
lea
r
u
n
d
e
rsta
n
d
in
g
o
f
m
o
d
e
l
d
e
c
isio
n
s,
su
p
p
o
rti
n
g
it
s
u
se
i
n
re
g
u
late
d
fin
a
n
c
ial
e
n
v
iro
n
m
e
n
ts
wh
e
re
p
re
c
isio
n
a
n
d
re
sp
o
n
si
b
il
it
y
a
re
c
ru
c
ial.
K
ey
w
o
r
d
s
:
Dee
p
lear
n
in
g
Featu
r
e
en
g
in
ee
r
i
n
g
Fra
u
d
d
etec
tio
n
I
m
b
alan
ce
d
d
ataset
Ma
ch
in
e
lear
n
in
g
T
ab
Net*
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
:
S.
Su
d
h
a
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
A
p
p
licatio
n
s
,
Hin
d
u
s
tan
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
an
d
Scien
ce
C
h
en
n
ai,
I
n
d
ia
E
m
ail:
s
u
d
h
as@
h
in
d
u
s
tan
u
n
iv
.
ac
.
in
1.
I
NT
RO
D
UCT
I
O
N
I
n
to
d
a
y
’
s
d
ig
ital
ec
o
n
o
m
y
w
h
er
e
as
o
n
lin
e
tr
a
n
s
ac
tio
n
s
th
r
o
u
g
h
m
o
b
ile
b
an
k
i
n
g
an
d
e
-
co
m
m
er
ce
p
latf
o
r
m
s
ar
e
ex
p
a
n
d
in
g
r
ap
id
ly
to
d
etec
tin
g
th
e
f
in
a
n
cial
f
r
au
d
h
as
b
ec
o
m
e
a
c
r
itical
ch
allen
g
e.
Fra
u
d
u
le
n
t
ac
tiv
ities
ar
e
co
n
s
tan
tly
b
ec
o
m
in
g
m
o
r
e
s
o
p
h
is
ticated
to
f
r
eq
u
e
n
t
an
d
o
f
ten
h
i
d
d
en
with
in
n
o
r
m
al
tr
an
s
ac
tio
n
al
b
eh
av
io
r
m
ak
in
g
ac
cu
r
ate
d
etec
tio
n
b
o
th
tech
n
ically
d
if
f
icu
lt
an
d
o
p
er
a
tio
n
ally
ess
en
tial.
T
r
ad
itio
n
al
r
u
le
-
b
ased
f
r
au
d
d
etec
tio
n
s
y
s
tem
s
wh
ich
d
ep
e
n
d
o
n
s
tatic
th
r
esh
o
l
d
s
an
d
m
an
u
ally
p
r
ed
ef
i
n
ed
ru
les h
as b
ec
o
m
e
in
c
r
ea
s
in
g
ly
in
ef
f
icien
t
in
id
e
n
tify
in
g
n
ew
an
d
co
m
p
lex
f
r
au
d
s
ch
em
es.
Ma
ch
in
e
lear
n
in
g
(
ML
)
an
d
d
ee
p
lear
n
in
g
(
DL
)
tech
n
i
q
u
es
h
av
e
g
ai
n
ed
atten
tio
n
f
o
r
t
h
eir
ab
ilit
y
to
en
h
an
ce
f
r
au
d
d
etec
tio
n
th
r
o
u
g
h
d
ata
-
d
r
iv
e
n
p
atter
n
r
ec
o
g
n
itio
n
.
C
lass
i
ca
l
ML
m
o
d
els
s
u
ch
as
lo
g
is
tic
r
eg
r
ess
io
n
,
d
ec
is
io
n
tr
ee
s
an
d
SVM
ar
e
wid
ely
u
s
ed
d
u
e
to
th
e
ea
s
e
w
ith
wh
ich
th
ey
ar
e
im
p
lem
en
ted
an
d
th
eir
s
im
p
licity
an
d
in
ter
p
r
et
ab
ilit
y
.
Ho
wev
er
,
t
h
ese
m
o
d
els
o
f
ten
s
tr
u
g
g
le
to
ca
p
tu
r
e
n
o
n
-
lin
ea
r
f
ea
tu
r
e
in
ter
ac
tio
n
s
an
d
h
ig
h
ly
im
b
al
an
ce
d
d
atasets
b
o
th
o
f
wh
ic
h
ar
e
ty
p
ical
c
h
ar
ac
ter
is
tics
o
f
r
ea
l
-
wo
r
ld
f
i
n
an
cial
f
r
au
d
d
ata.
E
n
s
em
b
le
m
eth
o
d
s
lik
e
as
r
an
d
o
m
f
o
r
est
,
XGBo
o
s
t,
an
d
L
ig
h
tGB
M
p
r
o
v
id
e
i
m
p
r
o
v
e
d
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
u
s
in
g
b
ag
g
i
n
g
an
d
b
o
o
s
tin
g
s
tr
ateg
ies
b
u
t
o
f
ten
r
em
ain
as
b
lac
k
-
b
o
x
m
o
d
els
with
lim
ited
in
ter
p
r
etab
ilit
y
to
a
n
im
p
o
r
tan
t lim
itatio
n
in
r
eg
u
lato
r
y
an
d
c
o
m
p
lian
ce
-
d
r
iv
en
f
in
an
cial
s
ettin
g
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
60
1
-
6
1
3
602
Dee
p
n
eu
r
al
n
etwo
r
k
s
o
f
ten
a
ch
iev
e
s
tr
o
n
g
p
r
e
d
ictiv
e
p
er
f
o
r
m
an
ce
b
u
t
s
im
ilar
ly
lack
tr
a
n
s
p
ar
en
cy
wh
ich
r
ed
u
ce
s
th
eir
p
r
ac
tical
ap
p
licab
ilit
y
wh
er
e
ex
p
lain
a
b
ilit
y
is
r
eq
u
ir
ed
.
T
h
ese
lim
itatio
n
s
h
ig
h
lig
h
t
th
e
n
ee
d
f
o
r
f
r
au
d
d
etec
tio
n
m
o
d
els
th
at
co
m
b
in
e
h
ig
h
p
r
ed
i
ctiv
e
ac
cu
r
ac
y
with
in
ter
p
r
et
ab
ilit
y
o
f
en
ab
lin
g
f
in
an
cial
in
s
titu
tio
n
s
to
d
et
ec
t
u
n
d
er
s
tan
d
an
d
r
esp
o
n
d
to
f
r
a
u
d
u
len
t
ac
tiv
ities
in
a
tr
an
s
p
ar
en
t
an
d
ac
co
u
n
tab
le
m
a
n
n
er
.
T
o
ad
d
r
ess
th
is
s
tu
d
y
p
r
o
p
o
s
es
a
f
r
a
u
d
d
etec
tio
n
f
r
a
m
ewo
r
k
b
ased
o
n
T
ab
Net*
an
in
te
r
p
r
et
ab
le
d
ee
p
lear
n
in
g
a
r
ch
itectu
r
e
d
esig
n
ed
s
p
ec
if
ically
f
o
r
tab
u
lar
d
ata.
Deta
ils
o
f
T
ab
Net’
s
s
tr
u
ctu
r
e
an
d
m
ec
h
an
is
m
ar
e
p
r
esen
ted
in
th
e
m
eth
o
d
s
s
ec
tio
n
h
o
wev
er
o
f
its
ab
ilit
y
to
p
r
o
v
i
d
e
in
s
tan
ce
lev
el
in
ter
p
r
etab
ilit
y
wh
ile
m
ain
tain
in
g
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
m
ak
es it a
s
u
itab
le
ca
n
d
id
at
e
f
o
r
a
n
aly
zin
g
f
in
an
cial
tr
an
s
ac
tio
n
f
r
au
d
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
Ov
er
th
e
p
ast
2
0
y
ea
r
s
,
m
u
ch
r
esear
ch
h
as
b
ee
n
f
o
c
u
s
ed
o
n
ad
d
r
ess
in
g
th
e
p
r
o
b
le
m
o
f
f
r
au
d
d
etec
tio
n
.
T
r
a
d
itio
n
al
class
if
ier
s
s
u
ch
as
d
ec
is
io
n
tr
ee
s
an
d
l
o
g
is
tic
r
eg
r
ess
io
n
.
B
h
attac
h
ar
y
y
a
et
a
l.
[
1
]
wh
er
e
th
e
f
ir
s
t
m
eth
o
d
is
to
p
r
o
v
id
e
f
r
au
d
d
etec
tio
n
s
o
lu
tio
n
s
an
d
u
s
e
th
e
f
ea
tu
r
e
th
r
esh
o
ld
s
to
id
en
tify
t
h
e
s
u
s
p
icio
u
s
tr
an
s
ac
tio
n
s
.
Un
f
o
r
tu
n
ately
,
th
ese
in
itial
m
o
d
els
d
id
n
o
t
ac
co
u
n
t
f
o
r
th
e
co
m
p
lex
an
d
n
o
n
-
lin
ea
r
p
atter
n
s
th
at
ar
e
ty
p
ical
o
f
f
r
au
d
.
T
h
e
a
d
v
en
t
o
f
en
s
em
b
l
e
m
eth
o
d
s
s
u
ch
as
r
an
d
o
m
f
o
r
est
an
d
g
r
a
d
ien
t
boos
tin
g
m
ac
h
in
es
co
n
s
titu
t
ed
a
m
ajo
r
ad
v
a
n
ce
m
en
t.
B
aish
o
lan
et
a
l.
[
2
]
p
r
o
p
o
s
ed
Fra
u
d
X
AI
a
n
in
ter
p
r
etab
le
m
ac
h
in
e
lear
n
i
n
g
f
r
am
ewo
r
k
f
o
r
cr
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
th
at
ef
f
ec
tiv
ely
h
an
d
les
h
ig
h
ly
im
b
alan
ce
d
d
atasets
.
T
h
e
r
esu
lt
h
ig
h
lig
h
t
h
o
w
cr
u
cial
it
i
s
f
o
r
m
o
d
el
to
b
e
in
ter
p
r
eta
b
ilit
y
f
o
r
p
r
ac
tical
d
ep
lo
y
m
e
n
t
in
r
ea
l
-
wo
r
l
d
f
in
a
n
cial
s
y
s
tem
s
.
T
h
is
f
in
d
in
g
is
co
n
s
is
ten
t
with
th
e
p
r
esen
t
s
tu
d
y
wh
er
e
T
ab
Net*
th
at
h
as
y
ield
ed
b
o
th
h
i
g
h
ac
c
u
r
ac
y
an
d
ex
p
lain
ab
le
p
r
ed
ictio
n
s
f
o
r
f
r
au
d
d
etec
tio
n
.
Su
ch
co
n
s
is
ten
cy
o
f
th
e
ap
p
r
o
ac
h
es
s
u
g
g
ested
b
y
Fra
u
d
X
AI
.
J
u
r
g
o
v
s
k
y
et
a
l.
[
3
]
al
s
o
ex
p
lo
r
e
d
R
NNs
an
d
n
o
tice
d
th
at
s
u
c
h
m
o
d
el
h
an
d
le
s
eq
u
en
tial
tr
an
s
ac
tio
n
d
ata.
H
o
wev
er
,
at
th
e
m
ain
d
o
wn
s
id
e
to
th
ese
m
o
d
els
a
t
th
e
tim
e
was
th
e
in
ter
p
r
etab
ilit
y
is
s
u
e,
m
ea
n
in
g
th
ese
m
o
d
els
h
av
e
h
ad
a
h
ar
d
er
tim
e
in
th
e
h
ea
v
ily
r
eg
u
lated
to
f
in
an
cial
en
v
ir
o
n
m
en
ts
.
Mo
s
t
r
ec
en
tly
,
T
ab
Net*
h
as
em
er
g
ed
as
a
n
ex
citin
g
n
ew
m
o
d
el
th
at
is
u
s
ed
f
o
r
atten
tio
n
m
ec
h
an
is
m
s
to
id
e
n
tify
r
elev
an
t
f
ea
tu
r
es
at
ea
ch
d
ec
is
io
n
p
o
in
t
wh
ile
r
etain
i
n
g
i
n
ter
p
r
etab
il
ity
.
Ar
ik
a
n
d
Pfister
[
4
]
s
h
o
wed
th
at
T
ab
N
et*
h
ad
co
m
p
ar
ab
le
ac
c
u
r
ac
y
to
g
r
ad
ien
t
b
o
o
s
tin
g
b
u
t
co
u
ld
also
p
r
o
v
i
d
e
th
e
tr
an
s
p
ar
en
cy
.
Ou
r
r
esear
ch
f
o
cu
s
f
u
r
th
e
r
s
th
is
d
ev
el
o
p
m
en
t
b
y
in
v
esti
g
atin
g
a
u
s
e
ca
s
e
f
o
r
T
ab
Net*
in
f
r
au
d
d
etec
tio
n
in
f
in
a
n
cial
tr
an
s
ac
tio
n
s
.
Nie
et
a
l.
[
5
]
p
r
o
p
o
s
ed
a
m
u
ltimo
d
al
f
r
a
u
d
d
etec
tio
n
f
r
a
m
ewo
r
k
co
m
b
in
in
g
te
x
tu
al
L
L
M
em
b
ed
d
in
g
s
with
s
tr
u
ctu
r
ed
f
in
an
cial
an
d
g
o
v
er
n
an
ce
d
ata.
Usi
n
g
g
r
ad
ien
t
-
b
o
o
s
ted
tr
ee
s
an
d
SHAP
in
ter
p
r
etab
ilit
y
,
th
e
m
o
d
el
h
i
g
h
lig
h
ted
k
e
y
f
in
an
cial
an
d
l
in
g
u
is
tic
in
d
icato
r
s
,
ac
h
iev
in
g
s
tr
o
n
g
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
(
AUC
>
0
.
8
5
)
.
T
h
e
s
tu
d
y
d
em
o
n
s
tr
ates
th
e
ef
f
ec
tiv
en
ess
o
f
in
ter
p
r
e
tab
le,
m
u
ltimo
d
al
ap
p
r
o
ac
h
es
f
o
r
f
in
a
n
cial
f
r
au
d
d
etec
tio
n
.
C
h
en
an
d
Gu
estrin
[
6
]
p
r
o
p
o
s
ed
th
e
XGBo
o
s
t
alg
o
r
ith
m
is
th
e
m
ain
r
ea
s
o
n
wh
y
t
h
is
m
o
d
el
is
ab
le
to
ex
ce
l
is
a
b
alan
cin
g
s
p
ee
d
as
wel
l
as
th
e
p
o
wer
o
f
m
ac
h
in
e
lear
n
in
g
task
s
.
T
h
is
is
b
ec
au
s
e
th
e
m
o
d
el
is
ab
le
to
u
s
e
m
a
n
y
o
p
er
atio
n
s
th
at
ca
n
b
e
d
o
n
e
in
a
p
ar
allel
en
v
ir
o
n
m
en
t
t
h
at
m
ak
es
th
is
m
o
d
el
ab
le
to
p
r
o
ce
s
s
th
e
m
illi
o
n
s
o
f
tr
an
s
a
c
tio
n
al
d
ata
th
at
f
r
a
u
d
s
ter
s
f
o
l
lo
w
in
co
m
m
itti
n
g
f
r
au
d
u
le
n
t a
ctiv
ities
.
Ke
et
a
l.
[
7
]
d
is
co
v
er
e
d
th
at
L
ig
h
tGB
M
a
f
r
am
ewo
r
k
to
b
e
i
n
th
e
tu
r
b
o
c
h
ar
g
es
o
f
lear
n
in
g
p
r
o
ce
s
s
.
B
y
ap
p
ly
in
g
tec
h
n
iq
u
es
s
u
ch
as
g
r
ad
ien
t
-
b
ased
o
n
e
-
s
id
e
s
am
p
lin
g
an
d
ex
clu
s
iv
e
f
ea
tu
r
e
b
u
n
d
lin
g
,
u
s
ed
i
n
th
e
L
ig
h
tGB
M
r
esu
lt
ac
h
iev
es
r
e
m
ar
k
ab
le
c
o
m
p
u
tatio
n
al
ef
f
icien
cy
with
n
o
s
ac
r
if
ice
in
ac
c
u
r
ac
y
.
T
h
is
m
ak
es
it
is
a
p
o
wer
h
o
u
s
e
f
o
r
th
e
lar
g
e
-
s
ca
le
d
atasets
th
at
m
ar
k
th
e
f
i
n
an
cial
in
d
u
s
tr
y
.
T
o
en
a
b
le
r
a
p
id
m
o
d
el
r
etr
ain
i
n
g
an
d
d
ep
lo
y
m
en
t
in
d
y
n
am
ic
e
n
v
ir
o
n
m
en
ts
s
o
m
eth
in
g
th
at
th
e
SC
A
R
FF
m
o
d
el
d
o
es
q
u
ick
ly
.
Fio
r
e
et
a
l.
[
8
]
s
u
g
g
ested
th
is
cr
ea
tiv
e
m
et
h
o
d
is
u
s
in
g
g
e
n
er
ativ
e
a
d
v
er
s
ar
ial
n
etwo
r
k
s
(
GANs)
.
B
u
t
in
s
tead
o
f
f
o
c
u
s
in
g
o
n
th
e
ex
is
tin
g
d
ata
an
d
th
eir
a
p
p
r
o
ac
h
ar
tific
ially
g
en
er
ates
r
ea
lis
tic,
s
y
n
th
etic
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
.
T
h
i
s
“d
ata
au
g
m
en
tatio
n
"
g
iv
es
th
e
m
o
d
el
a
m
u
ch
r
ich
er
an
d
m
o
r
e
v
ar
ied
u
n
d
er
s
tan
d
in
g
o
f
wh
at
f
r
au
d
ca
n
lo
o
k
lik
e,
s
ig
n
if
ican
tly
s
h
ar
p
e
n
in
g
i
ts
ab
ilit
y
to
r
ec
o
g
n
ize
t
h
e
n
ew
ly
em
er
g
in
g
f
r
a
u
d
.
C
o
r
r
ea
B
ah
n
s
en
et
a
l
.
[
9
]
d
e
m
o
n
s
tr
ated
th
e
r
elev
an
ce
o
f
th
e
ar
t
o
f
f
ea
tu
r
e
en
g
in
ee
r
in
g
.
B
ased
o
n
t
h
at
to
ca
p
tu
r
e
tem
p
o
r
al
an
d
b
eh
a
v
io
r
asp
ec
ts
s
u
ch
as
th
e
c
o
m
p
ar
is
o
n
f
r
e
q
u
en
c
y
an
d
tim
e
b
et
wee
n
p
u
r
ch
ases
as
cr
u
cial
as
th
e
alg
o
r
ith
m
its
elf
.
T
h
eir
wo
r
k
s
h
o
u
ld
r
em
in
d
u
s
th
at
with
o
u
t
th
ese
in
s
ig
h
tf
u
l
f
ea
tu
r
es,
ev
en
th
e
m
o
s
t
s
o
p
h
is
ticated
m
o
d
el
is
o
p
er
atin
g
with
b
lin
d
er
s
o
n
.
Ma
r
y
et
a
l
.
[
1
0
]
h
as
an
aly
ze
d
a
s
y
s
tem
f
o
r
d
etec
tin
g
a
o
n
lin
e
tr
a
n
s
ac
tio
n
f
r
au
d
th
at
u
s
ag
e
o
f
r
u
le
-
b
ased
s
y
s
tem
with
ea
r
l
y
Ma
ch
i
n
e
L
ea
r
n
in
g
al
g
o
r
ith
m
th
e
im
p
o
r
tan
ce
o
f
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVMs
)
in
class
if
icat
io
n
task
s
an
d
th
en
u
s
es
d
ec
is
io
n
th
r
esh
o
l
d
f
o
r
th
e
an
o
m
aly
class
if
icatio
n
.
I
t
is
l
im
ited
in
th
e
ter
m
s
o
f
s
ca
lab
ilit
y
an
d
ad
ap
tab
ilit
y
to
ev
o
l
v
in
g
f
r
au
d
p
atter
n
s
b
ec
au
s
e
d
esp
ite
it i
s
ef
f
ec
tiv
en
ess
w
ith
in
th
e
s
m
al
l d
ataset.
Fu
tu
r
e
r
esear
ch
m
ay
f
o
cu
s
o
n
ex
p
lo
r
in
g
ad
v
an
ce
d
o
p
tim
izatio
n
tech
n
iq
u
es
b
u
t
th
e
h
y
b
r
i
d
ap
p
r
o
ac
h
es
to
f
u
r
th
er
im
p
r
o
v
e
t
h
e
p
e
r
f
o
r
m
an
ce
o
f
SVMs
in
class
if
icatio
n
task
s
in
v
ar
io
u
s
d
o
m
ain
s
.
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
F
r
a
u
d
d
etec
tio
n
u
s
in
g
Ta
b
N
et
*
cla
s
s
ifier
a
ma
ch
in
e
lea
r
n
in
g
a
p
p
r
o
a
ch
(
G.
A
n
is
h
Ma
r
y
)
603
Van
in
i
et
a
l
.
[
1
1
]
ex
am
in
e
d
to
th
e
f
in
an
cial
f
r
au
d
o
cc
u
r
s
o
v
e
r
a
v
ar
iety
o
f
ch
an
n
els,
in
clu
d
in
g
cr
ed
it
ca
r
d
s
,
in
ter
n
et
b
a
n
k
in
g
,
p
h
o
n
e
b
an
k
in
g
,
ch
eq
u
es,
an
d
e
-
co
m
m
er
ce
,
u
s
ed
a
r
ea
l
d
ataset
f
r
o
m
a
p
r
iv
ate
b
a
n
k
to
ev
alu
atio
n
o
f
th
e
f
r
a
u
d
d
etec
tio
n
m
eth
o
d
o
l
o
g
ies.
Dev
elo
p
in
g
th
is
r
esear
ch
f
r
au
d
p
r
ev
en
tio
n
is
as
a
p
ar
t
o
f
r
is
k
m
an
ag
em
en
t
f
r
am
ew
o
r
k
.
Mo
r
eo
v
er
,
th
eir
r
esear
ch
is
f
o
c
u
s
f
r
o
m
f
r
a
u
d
d
etec
tio
n
id
en
ti
f
icatio
n
is
d
ec
is
io
n
m
ak
in
g
f
o
r
b
o
th
co
m
p
lian
ce
an
d
u
s
er
tr
u
s
t.
E
ac
h
b
a
n
k
in
g
s
ess
io
n
aim
s
to
en
co
d
e
d
e
v
i
atio
n
s
f
r
o
m
ty
p
ical
cu
s
to
m
er
b
eh
a
v
io
u
r
.
Ku
m
a
r
et
a
l
.
[
1
2
]
Alth
o
u
g
h
ea
ch
o
f
th
e
s
e
ap
p
r
o
ac
h
es h
av
e
u
s
e
d
if
f
er
e
n
t
k
in
d
o
f
m
eth
o
d
s
s
u
ch
as
m
ac
h
i
n
e
lear
n
in
g
lo
g
is
tic
r
eg
r
ess
io
n
,
r
an
d
o
m
f
o
r
est,
SVM
alg
o
r
ith
m
o
n
th
e
d
at
aset
an
d
c
o
m
p
ar
e
d
th
eir
p
e
r
f
o
r
m
an
ce
t
o
k
n
o
w
wh
ich
o
n
e
is
b
etter
am
o
n
g
t
h
ese
th
r
ee
.
C
o
m
p
ar
in
g
t
h
e
r
e
s
u
lts
o
f
th
ese
t
h
r
ee
alg
o
r
ith
m
s
,
th
e
f
o
r
est alg
o
r
ith
m
g
iv
es th
e
b
est r
esu
lt.
Kad
am
et
a
l
.
[
1
3
]
h
av
e
a
p
p
l
i
ed
th
e
m
o
d
el
th
at
p
r
o
d
u
ce
d
b
etter
r
esu
lts
wer
e
as
r
an
d
o
m
f
o
r
est,
d
ec
is
io
n
tr
ee
,
an
d
l
o
g
is
tic
r
eg
r
ess
io
n
.
Priy
a
an
d
Sar
ad
h
a
[
1
4
]
m
o
s
t
d
ig
ital
f
r
a
u
d
h
as
em
e
r
g
ed
as
a
p
er
v
asiv
e
th
r
ea
t
ac
r
o
s
s
all
s
ec
to
r
s
,
r
eq
u
ir
in
g
d
ed
icate
d
ef
f
o
r
ts
b
y
o
r
g
an
izatio
n
s
to
im
p
r
o
v
e
s
ec
u
r
ity
m
ea
s
u
r
es.
T
h
e
ad
v
en
t
o
f
d
ig
itizatio
n
h
as
s
tr
ea
m
lin
ed
d
aily
tr
a
n
s
ac
tio
n
s
b
u
t
h
as
also
ex
p
o
s
ed
v
u
ln
er
ab
ili
ties
th
at
m
alicio
u
s
ac
to
r
s
ca
n
ex
p
lo
it.
Fra
u
d
u
len
t
ac
to
r
s
ar
e
k
n
o
wn
to
ca
r
r
y
o
u
t
tr
an
s
ac
tio
n
s
,
wh
ile
d
is
g
u
is
in
g
th
em
s
elv
es
as
g
en
u
in
e
cu
s
to
m
er
s
ca
u
s
in
g
s
ig
n
if
ican
t
i
n
f
i
n
an
cial
lo
s
s
es
an
d
tar
n
is
h
in
g
b
r
a
n
d
r
ep
u
tatio
n
.
I
n
th
e
Or
g
an
izatio
n
s
f
ac
e
th
r
ea
ts
f
r
o
m
ad
v
an
ce
d
d
ig
ital
f
r
au
d
s
t
er
s
ar
e
in
cr
ea
s
in
g
ly
in
ab
le
to
m
a
n
ip
u
late
t
h
e
wea
k
n
ess
in
d
i
g
ital
ap
p
licatio
n
s
.
T
h
e
ad
d
r
ess
ar
e
c
h
allen
g
e
s
f
o
r
a
ce
n
t
r
alize
d
f
r
au
d
m
a
n
a
g
em
en
t
p
latf
o
r
m
in
ar
ticu
lates
to
a
f
o
r
war
d
-
th
in
k
in
g
ap
p
r
o
ac
h
to
co
u
n
ter
in
g
d
ig
ital
f
r
au
d
.
B
y
th
e
f
o
s
ter
in
g
co
llab
o
r
atio
n
an
d
in
f
o
r
m
atio
n
to
s
h
ar
in
g
am
o
n
g
in
o
r
g
a
n
izatio
n
s
ar
o
u
n
d
t
h
e
wo
r
ld
in
it
is
aim
to
b
u
ild
in
a
r
esil
ien
t
d
ef
en
s
e
ag
ai
n
s
t e
m
er
g
in
g
th
r
ea
ts
.
W
h
ile
d
ev
elo
p
i
n
g
th
e
c
o
m
m
u
n
ity
-
b
ased
f
r
am
ewo
r
k
f
o
r
f
r
au
d
p
r
ev
en
tio
n
.
Vin
ay
a
et
a
l
.
[
1
5
]
n
o
t
ed
i
n
f
i
n
an
cia
l
s
ec
to
r
t
h
r
o
u
g
h
th
e
i
n
te
g
r
at
io
n
o
f
i
n
f
o
r
m
a
ti
o
n
t
ec
h
n
o
l
o
g
y
(
I
T
)
h
as
s
ig
n
i
f
ic
a
n
tl
y
al
te
r
n
at
e
p
ay
m
e
n
t
m
et
h
o
d
s
o
f
p
e
o
p
le
f
r
o
m
tr
a
d
it
io
n
a
l
ca
s
h
t
r
a
n
s
ac
ti
o
n
s
t
o
e
lect
r
o
n
ic
p
a
y
m
e
n
ts
s
u
c
h
as
cr
ed
it
c
ar
d
s
,
m
o
b
il
e
U
PI
b
ase
d
t
r
a
n
s
a
cti
o
n
s
.
I
n
t
h
is
ev
o
l
u
ti
o
n
h
as
i
n
c
r
e
ase
d
t
h
e
s
u
s
ce
p
ti
b
il
it
y
o
f
t
h
es
e
s
y
s
te
m
s
i
n
ill
eg
al
a
cti
v
i
ties
.
T
h
e
y
a
r
e
c
o
m
b
at
t
h
es
e
f
in
an
ci
a
l
i
n
s
t
it
u
ti
o
n
s
to
u
s
e
t
h
e
f
r
a
u
d
d
et
ec
t
io
n
s
y
s
te
m
s
(
FDS
)
t
o
p
r
o
tec
t
t
h
e
c
o
n
s
u
m
e
r
s
ag
ai
n
s
t
f
r
a
u
d
u
le
n
t
t
r
an
s
ac
t
io
n
s
.
ML
a
n
d
d
ee
p
lea
r
n
i
n
g
al
g
o
r
it
h
m
s
h
a
v
e
s
h
o
w
n
q
u
ite
p
r
o
m
is
e
i
n
ef
f
i
cie
n
t
ly
cl
ass
if
y
i
n
g
t
h
e
tr
a
n
s
ac
ti
o
n
s
i
n
g
iv
e
n
d
a
tas
ets.
I
n
t
h
e
i
n
te
g
r
at
e
d
m
ac
h
i
n
e
le
ar
n
i
n
g
an
d
ele
ct
r
o
n
i
c
p
a
y
m
e
n
t
r
e
co
r
d
a
n
al
y
s
is
h
as
th
e
p
o
te
n
ti
al
an
d
s
i
g
n
i
f
ic
a
n
tl
y
t
o
im
p
r
o
v
e
th
e
f
r
a
u
d
d
et
ec
ti
o
n
s
y
s
te
m
s
.
T
h
e
y
ar
e
test
i
n
g
wit
h
d
i
f
f
er
e
n
t
d
atas
ets
is
r
e
c
o
m
m
e
n
d
e
d
t
o
v
a
li
d
at
e
an
d
i
m
p
r
o
v
e
t
h
e
m
et
h
o
d
s
.
M
o
t
i
e
a
n
d
R
aa
h
e
m
i
[
1
6
]
d
i
s
c
u
s
s
e
d
t
h
e
m
t
o
u
s
e
i
n
g
a
t
e
d
n
e
u
r
a
l
n
e
t
w
o
r
k
s
(
GN
N
)
f
o
r
f
r
a
u
d
i
n
f
i
n
a
n
c
e
.
T
h
e
y
a
r
e
h
i
g
h
l
i
g
h
t
e
d
i
n
t
h
e
i
r
s
t
r
e
n
g
t
h
s
i
n
c
u
r
r
e
n
t
a
p
p
l
i
c
at
i
o
n
s
a
n
d
e
x
i
s
ti
n
g
g
a
p
s
.
As
t
h
e
f
r
a
u
d
s
t
e
r
s
g
et
m
o
r
e
a
d
v
a
n
c
e
d
i
n
t
h
e
i
r
t
a
c
t
i
cs
,
t
h
e
k
e
y
t
o
b
u
i
l
d
i
n
g
s
t
r
o
n
g
f
r
a
u
d
d
e
t
e
c
t
i
o
n
s
y
s
t
e
m
s
is
g
o
i
n
g
t
o
b
e
i
m
p
r
o
v
e
d
i
n
G
N
N
,
so
t
h
e
y
c
a
n
h
a
n
d
l
e
w
i
t
h
r
e
a
l
l
y
l
a
r
g
e
d
a
t
a
s
et
s
.
T
h
e
y
a
r
e
f
o
c
u
s
ed
i
n
p
l
u
g
g
i
n
g
t
h
o
s
e
g
a
p
s
t
o
g
i
v
e
f
i
n
a
n
c
i
a
l
s
y
s
t
e
m
s
a
n
d
t
h
e
b
e
s
t
p
o
s
s
i
b
l
e
p
r
o
t
e
c
tio
n
a
g
a
i
n
s
t
f
r
a
u
d
.
S
h
a
r
m
a
e
t
a
l
.
[
1
7
]
s
t
a
t
e
d
t
h
a
t
d
e
te
c
t
i
n
g
f
r
a
u
d
i
n
f
i
n
a
n
c
i
a
l
t
r
a
n
s
a
c
ti
o
n
s
is
a
n
e
s
s
e
n
ti
a
l
a
s
p
e
c
t
o
f
e
n
s
u
r
i
n
g
t
h
e
s
e
c
u
r
i
t
y
a
n
d
t
r
u
s
t
w
o
r
t
h
i
n
es
s
o
f
b
a
n
k
i
n
g
a
n
d
p
r
i
v
a
t
e
f
i
n
a
n
c
i
a
l
s
y
s
te
m
s
.
I
n
t
h
e
d
i
g
i
t
al
t
r
a
n
s
ac
t
io
n
s
o
n
r
i
s
e
a
n
d
c
y
b
e
r
t
h
r
e
a
t
s
g
e
t
t
i
n
g
e
v
e
r
m
o
r
e
c
o
m
p
l
e
x
M
L
t
e
c
h
n
i
q
u
e
s
p
l
a
y
a
n
i
m
p
o
r
t
a
n
t
r
o
l
e
i
n
d
e
t
ec
t
i
n
g
t
h
e
s
u
s
p
i
ci
o
u
s
t
r
a
n
s
a
c
t
i
o
n
s
a
n
d
m
i
ti
g
a
t
i
n
g
f
r
a
u
d
a
c
t
i
v
it
i
es
i
n
t
h
e
b
a
n
k
i
n
g
s
e
c
t
o
r
.
Sn
e
h
a
et
a
l
.
[
1
8
]
n
o
t
ed
th
at
m
o
d
er
n
m
ac
h
i
n
e
le
ar
n
i
n
g
m
et
h
o
d
s
l
ik
e
a
n
e
n
s
em
b
l
e
le
ar
n
i
n
g
a
n
d
d
ee
p
lea
r
n
i
n
g
al
o
n
g
w
it
h
h
y
p
e
r
p
a
r
a
m
et
er
tu
n
i
n
g
h
a
v
e
g
r
ea
tl
y
im
p
r
o
v
ed
th
e
p
er
f
o
r
m
an
ce
o
f
f
r
a
u
d
d
et
ec
ti
o
n
s
y
s
te
m
s
in
t
h
e
b
an
k
i
n
g
i
n
d
u
s
tr
y
.
T
h
es
e
m
o
d
els
t
h
r
o
u
g
h
cl
ass
we
ig
h
t
t
u
n
i
n
g
an
d
o
p
ti
m
a
l
h
y
p
e
r
p
a
r
a
m
ete
r
s
,
th
ese
m
o
d
els
ca
n
b
ett
e
r
a
d
d
r
ess
t
h
e
c
h
all
e
n
g
es
p
o
s
e
d
b
y
im
b
a
la
n
c
e
d
d
ata
,
i
m
p
r
o
v
i
n
g
t
h
e
a
b
ilit
y
t
o
d
e
tec
t
f
r
au
d
u
l
en
t
ac
ti
v
i
ties
.
T
h
e
c
o
n
ti
n
u
o
u
s
r
e
s
ea
r
c
h
an
d
d
ev
el
o
p
m
en
t
o
f
ad
ap
ti
v
e
,
r
o
b
u
s
t
m
o
d
e
ls
a
r
e
ess
e
n
ti
al
to
s
e
cu
r
e
f
i
n
a
n
c
ial
t
r
a
n
s
ac
t
io
n
s
.
K
u
m
ar
et
a
l.
[
1
9
]
to
h
ig
h
l
ig
h
t
th
e
d
if
f
e
r
e
n
t
M
L
t
ec
h
n
iq
u
es
li
k
e
as
lo
g
is
t
ic
r
e
g
r
ess
i
o
n
,
d
e
cisi
o
n
t
r
e
es,
a
n
d
g
r
ad
ie
n
t
b
o
o
s
ti
n
g
we
r
e
p
r
ese
n
t
ed
t
h
ei
r
u
s
a
g
e
i
n
p
r
e
d
ict
in
g
l
o
a
n
d
e
f
a
u
lts
b
y
m
o
d
eli
n
g
t
h
e
co
m
p
le
x
r
el
ati
o
n
s
h
i
p
s
b
etw
ee
n
b
o
r
r
o
we
r
c
h
a
r
ac
t
er
is
ti
cs
a
n
d
t
h
e
li
k
el
ih
o
o
d
o
f
l
o
an
r
ep
a
y
m
e
n
t.
T
o
g
i
v
e
an
ex
am
p
l
e,
l
o
g
is
t
ic
r
e
g
r
ess
i
o
n
is
u
s
e
d
f
o
r
its
i
n
te
r
p
r
et
ati
o
n
in
b
i
n
a
r
y
cl
ass
if
ic
ati
o
n
p
r
o
b
le
m
s
,
wh
er
ea
s
d
e
cisi
o
n
tr
e
es
ar
e
u
t
ili
ze
d
f
o
r
s
t
r
a
ig
h
t
f
o
r
war
d
d
ec
is
i
o
n
m
a
k
i
n
g
t
h
r
o
u
g
h
h
ie
r
a
r
c
h
i
ca
l
d
at
a
p
ar
t
iti
o
n
i
n
g
.
Ag
u
s
tin
o
et
a
l.
[
2
0
]
f
o
c
u
s
es
o
n
th
e
ev
alu
atio
n
o
f
th
e
m
o
s
t
u
s
ef
u
l
m
o
d
els
f
o
r
f
r
au
d
d
et
ec
tio
n
ar
e
f
o
cu
s
o
f
th
e
s
tu
d
y
.
T
h
e
p
ap
er
in
d
icate
s
th
at
n
o
s
in
g
le
alg
o
r
ith
m
g
lo
b
ally
o
u
tp
er
f
o
r
m
s
o
th
e
r
s
in
all
s
ce
n
ar
io
s
,
th
u
s
h
ig
h
lig
h
tin
g
th
e
im
p
o
r
ta
n
ce
o
f
ev
alu
atin
g
m
u
ltip
le
m
o
d
els.
Fo
r
ex
am
p
le,
lo
g
is
tic
r
eg
r
ess
io
n
an
d
lin
ea
r
d
is
cr
im
in
an
t
an
aly
s
is
(
L
DA)
ar
e
f
r
eq
u
en
tly
r
ec
o
g
n
ized
f
o
r
th
eir
a
b
ilit
y
to
h
an
d
le
b
i
n
ar
y
class
if
icatio
n
p
r
o
b
lem
s
an
d
p
r
o
v
id
e
p
r
o
b
ab
ilis
tic
o
u
tp
u
ts
,
wh
ich
ar
e
u
s
ef
u
l
f
o
r
f
r
au
d
d
etec
tio
n
s
y
s
tem
s
.
L
ei
et
a
l.
[
2
1
]
AI
in
s
u
p
p
ly
ch
ain
m
a
n
ag
em
e
n
t
:
AI
,
esp
ec
ially
ML
alg
o
r
ith
m
s
,
is
p
lay
in
g
a
k
ey
r
o
le
in
m
o
d
er
n
izin
g
s
u
p
p
ly
c
h
ain
s
b
y
im
p
r
o
v
in
g
d
ec
is
io
n
m
ak
in
g
th
r
o
u
g
h
ad
v
an
ce
d
d
ata
an
aly
s
is
,
h
elp
in
g
co
m
p
an
ies
m
a
k
e
s
cien
tific
d
ec
is
io
n
s
u
s
in
g
f
in
an
cial
in
d
ex
d
ata.
R
is
k
Ma
n
ag
e
m
en
t
Am
id
Glo
b
al
Un
ce
r
tain
ties
ex
p
lo
r
es
t
h
e
n
e
ed
f
o
r
AI
-
d
r
iv
e
n
to
o
ls
to
m
a
n
ag
e
in
cr
ea
s
ed
r
is
k
s
f
r
o
m
g
lo
b
al
u
n
ce
r
tain
ties
lik
e
C
o
v
id
-
19.
E
n
jo
lr
as
an
d
Ma
d
iès
[
2
2
]
Usi
n
g
b
o
th
q
u
an
titativ
e
d
ata
s
u
ch
as
r
is
k
s
co
r
es
an
d
cr
iter
ia
q
u
alitativ
e
d
ata
s
u
ch
as
an
aly
s
t’
s
o
p
in
io
n
s
in
s
u
p
p
ly
ch
ain
m
an
ag
em
en
t
th
is
p
ap
er
ex
am
in
es
th
e
im
p
o
r
tan
t
r
o
le
b
a
n
k
s
p
lay
in
p
r
ed
ictin
g
f
i
n
an
cial
d
is
tr
ess
.
Alth
o
u
g
h
th
er
e
is
a
s
ig
n
i
f
ican
t li
ter
atu
r
e
o
f
p
r
ed
ictin
g
f
in
an
cial
d
is
tr
ess
in
a
v
ar
io
u
s
s
ec
to
r
.
Ad
d
r
ess
in
g
th
e
ag
r
icu
ltu
r
al
s
ec
to
r
is
lar
g
ely
o
v
e
r
lo
o
k
e
d
d
esp
ite
th
e
h
ig
h
f
in
a
n
cial
r
is
k
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
60
1
-
6
1
3
604
ass
o
ciate
d
with
ag
r
icu
ltu
r
e
an
d
th
e
s
ec
to
r
'
s
r
elian
ce
o
n
b
an
k
lo
an
s
.
A
f
ield
n
o
t
u
s
u
ally
i
n
clu
d
ed
in
f
in
an
cial
cr
is
is
r
esear
ch
.
C
o
m
p
ar
ed
to
a
n
aly
s
ts
o
p
in
io
n
s
,
r
is
k
s
co
r
es,
p
ar
ticu
lar
ly
ass
ess
in
g
co
u
n
ter
p
ar
ty
r
is
k
,
ar
e
m
o
r
e
ef
f
ec
tiv
e
p
r
ed
icto
r
s
o
f
f
i
n
an
ci
al
cr
is
is
ev
en
ts
an
d
th
ei
r
d
u
r
at
io
n
s
.
T
h
e
f
in
d
in
g
s
ar
e
a
p
p
lica
b
le
to
o
t
h
er
s
ec
to
r
s
s
u
ch
as
s
m
all
an
d
m
ed
iu
m
e
n
ter
p
r
is
es,
g
u
id
in
g
f
u
tu
r
e
r
esear
ch
an
d
r
is
k
m
an
ag
em
e
n
t
s
tr
ateg
ies
in
b
r
o
ad
er
ec
o
n
o
m
ic
co
n
tex
ts
.
Mu
tem
i
an
d
B
ac
ao
[
2
3
]
h
as
g
ain
e
d
in
th
e
r
ap
id
g
r
o
wth
o
f
th
e
e
-
co
m
m
er
ce
s
ec
to
r
,
f
u
r
th
er
ac
ce
ler
ated
b
y
th
e
C
o
v
id
-
1
9
p
an
d
em
ic,
h
as
led
to
a
s
ig
n
if
ican
t
in
cr
ea
s
e
in
d
ig
ital
f
r
au
d
an
d
ass
o
ciate
d
f
in
an
cial
lo
s
s
es.
T
h
e
r
is
e
in
o
n
lin
e
f
r
a
u
d
h
i
g
h
lig
h
ts
th
e
u
r
g
en
t
n
ee
d
f
o
r
s
tr
o
n
g
cy
b
e
r
s
ec
u
r
ity
an
d
an
ti
-
f
r
au
d
m
ea
s
u
r
es
to
m
ain
tain
a
s
ec
u
r
e
e
-
co
m
m
e
r
ce
e
n
v
ir
o
n
m
en
t.
Ho
wev
er
,
r
esear
c
h
in
f
r
a
u
d
d
etec
tio
n
co
n
tin
u
es
to
ch
alle
n
g
es,
m
ain
ly
d
u
e
to
a
la
ck
o
f
r
ea
l
-
wo
r
l
d
d
atasets
,
b
ec
au
s
e
o
f
th
is
,
it
lim
it
s
th
e
d
ev
elo
p
m
en
t
an
d
test
in
g
o
f
ef
f
ec
tiv
e
s
o
l
u
tio
n
s
.
Hu
an
g
[
2
4
]
p
r
esen
ts
an
o
p
ti
m
ized
L
ig
h
tGB
M
m
o
d
el
f
o
r
o
n
lin
e
cr
e
d
it
ca
r
d
f
r
au
d
d
ete
ctio
n
.
T
h
is
m
o
d
el
ad
d
r
ess
th
e
g
r
o
win
g
n
ee
d
f
o
r
ef
f
ec
tiv
e
s
o
lu
tio
n
s
b
e
ca
u
s
e
to
th
e
g
r
o
w
in
e
-
co
m
m
er
ce
an
d
ass
o
ciate
d
f
r
au
d
r
is
k
s
.
T
h
e
s
tu
d
y
u
s
es
t
h
e
I
E
E
E
-
C
I
S
Fra
u
d
Dete
ctio
n
d
a
taset
with
m
o
r
e
th
an
o
n
e
m
illi
o
n
s
am
p
le
o
f
ev
alu
ate
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
m
o
d
el.
C
o
m
p
ar
e
d
to
tr
ad
it
io
n
al
m
o
d
els
lik
e
SVM
,
XG
B
o
o
s
t
an
d
R
an
d
o
m
Fo
r
est,
L
ig
h
tGB
M
-
b
ased
ap
p
r
o
ac
h
s
h
o
ws b
etter
r
esu
lts
co
m
p
ar
ed
to
tr
ad
itio
n
al
m
o
d
els.
I
n
ad
d
itio
n
,
th
e
p
ap
er
in
tr
o
d
u
ce
d
u
s
ef
u
l
f
ea
tu
r
e
en
g
in
ee
r
in
g
tech
n
iq
u
es
a
n
d
u
s
es
B
ay
esian
o
p
tim
izati
o
n
f
o
r
au
to
m
atic
h
y
p
er
p
ar
am
eter
tu
n
in
g
i
n
wh
ich
in
cr
ea
s
e
th
e
m
o
d
el
ac
cu
r
ac
y
an
d
p
e
r
f
o
r
m
an
ce
in
f
r
au
d
d
etec
tio
n
.
Pan
[
2
5
]
t
h
is
p
ap
e
r
is
s
tr
u
ct
u
r
ed
th
e
ap
p
licatio
n
o
f
m
ac
h
in
e
lear
n
in
g
in
f
in
an
cial
tr
an
s
ac
tio
n
.
T
h
e
p
ap
er
s
h
o
w
th
at
f
r
au
d
d
etec
tio
n
a
n
d
p
r
e
v
en
tio
n
,
h
ig
h
lig
h
tin
g
its
ad
v
an
tag
es
o
v
er
tr
ad
itio
n
al
m
eth
o
d
s
in
d
ea
lin
g
with
co
m
p
lex
f
r
au
d
p
atter
n
s
.
W
h
ile
ad
d
r
ess
in
g
ch
allen
g
es
s
u
ch
as
d
ata
q
u
ality
,
m
o
d
el
in
ter
p
r
etatio
n
an
d
in
teg
r
atio
n
with
ex
is
tin
g
s
y
s
te
m
s
.
3.
DATAS
E
T
DE
SI
G
N
I
n
th
is
s
tu
d
y
th
e
d
ataset
i
s
a
w
ell
-
o
r
g
an
ized
d
e
n
o
r
m
alize
d
tr
an
s
ac
tio
n
al
tab
le
cr
ea
ted
s
p
ec
if
ically
f
o
r
an
aly
zin
g
b
an
k
f
r
au
d
d
etec
tio
n
.
I
t in
clu
d
es 2
3
q
u
alities
th
at
ar
e
o
r
g
a
n
ized
in
to
f
o
u
r
m
ain
g
r
o
u
p
s
:
a.
Dem
o
g
r
ap
h
ics o
f
th
e
ca
r
d
h
o
ld
er
s
b.
I
n
f
o
r
m
atio
n
o
n
th
e
r
eg
io
n
an
d
th
e
ec
o
lo
g
y
c.
T
h
e
tr
an
s
ac
tio
n
f
o
r
id
e
n
tific
atio
n
d.
Me
tad
ata
f
o
r
class
if
icati
o
n
o
f
f
r
au
d
.
A
tim
estam
p
in
f
o
r
m
atio
n
is
to
id
en
tif
y
(
tr
a
n
s
_
d
ate_
tim
e)
u
n
iq
u
ely
f
o
r
ea
c
h
tr
an
s
ac
tio
n
r
ec
o
r
d
.
I
t
also
in
clu
d
es
th
e
in
f
o
r
m
atio
n
o
f
th
e
d
ate
an
d
tim
e
o
f
th
e
tr
an
s
ac
tio
n
,
th
e
cr
ed
it
ca
r
d
n
u
m
b
er
(
cc
_
n
u
m
)
th
e
n
th
e
n
am
e
o
f
th
e
m
er
ch
a
n
t,
th
e
s
p
en
d
in
g
ca
teg
o
r
y
(
ca
te
g
o
r
y
)
,
an
d
th
e
am
o
u
n
t
o
f
th
e
tr
a
n
s
ac
tio
n
(
am
t)
.
E
ac
h
tr
an
s
ac
tio
n
also
h
as
a
u
n
i
q
u
e
h
ash
r
ef
er
e
n
ce
(
t
r
an
s
_
n
u
m
)
a
n
d
a
Un
ix
tim
estam
p
(
u
n
ix
_
tim
e)
,
wh
ich
m
ak
es
it
p
o
s
s
ib
le
to
d
o
ac
c
u
r
ate
tim
e
-
s
er
ies an
d
b
eh
a
v
io
r
al
an
al
y
tics
.
T
h
is
d
ataset
in
clu
d
es
a
v
ar
iety
o
f
p
er
s
o
n
al
an
d
d
e
m
o
g
r
a
p
h
ic
in
f
o
r
m
atio
n
lik
e
f
ir
s
t
an
d
last
n
am
es,
g
en
d
er
,
s
tr
ee
t a
d
d
r
ess
,
jo
b
titl
e,
d
ate
o
f
b
ir
th
(
d
o
b
)
,
an
d
city
p
o
p
u
latio
n
(
city
_
p
o
p
)
,
to
lin
k
tr
an
s
ac
tio
n
s
to
th
eir
ca
r
d
h
o
ld
e
r
s
.
Geo
g
r
a
p
h
ic
co
o
r
d
in
ate
is
a
n
am
ely
ca
r
d
h
o
ld
e
r
latitu
d
e
an
d
lo
n
g
itu
d
e
(
lat,
lo
n
g
)
a
n
d
th
e
n
th
e
m
er
ch
an
t
co
o
r
d
in
ates
(
m
er
c
h
_
lat,
m
er
ch
_
lo
n
g
)
.
T
h
is
allo
ws
th
e
d
ev
elo
p
m
e
n
t
o
f
d
is
tan
ce
-
b
ased
an
d
lo
ca
tio
n
-
awa
r
e
f
ea
tu
r
es,
wh
ich
a
r
e
b
en
ef
icial
f
o
r
s
p
atio
-
tem
p
o
r
al
f
r
a
u
d
m
o
d
elin
g
.
T
h
e
d
ataset
u
s
ed
in
th
is
s
tu
d
y
was
a
b
in
ar
y
class
if
icatio
n
task
.
T
h
en
wh
er
e
th
e
tar
g
et
v
ar
iab
le
is
f
r
au
d
is
1
wh
en
th
e
tr
an
s
ac
tio
n
is
f
r
au
d
u
len
t
an
d
0
o
t
h
er
wis
e.
Su
ch
a
lab
elin
g
s
ch
em
a
m
ak
es
it
ea
s
ier
f
o
r
s
u
p
er
v
is
ed
m
ac
h
in
e
lea
r
n
in
g
tech
n
iq
u
es
to
d
is
tin
g
u
is
h
tr
u
e
a
ctio
n
s
f
r
o
m
th
e
f
ak
e
o
n
es.
T
h
e
d
ata
s
ch
em
a,
p
r
e
-
p
r
o
ce
s
s
in
g
m
eth
o
d
o
lo
g
y
an
d
e
n
g
in
ee
r
in
g
o
f
f
ea
tu
r
es
p
ip
elin
es
ar
e
well
d
o
cu
m
en
ted
to
en
s
u
r
e
r
ep
r
o
d
u
cib
ilit
y
.
An
o
n
y
m
ize
d
s
am
p
le
d
ata
an
d
co
d
e
ar
e
p
r
o
v
id
ed
as su
p
p
le
m
en
tal
in
f
o
r
m
atio
n
.
4.
DATA P
R
E
P
RO
CE
SS
I
NG
I
t
is
im
p
o
r
tan
t
t
o
tr
an
s
f
o
r
m
r
aw
tr
an
s
ac
tio
n
al
d
ata
in
t
o
a
m
ea
n
in
g
f
u
l
an
d
an
aly
za
b
le
f
o
r
m
at.
T
h
e
p
r
o
ce
d
u
r
e,
h
o
wev
er
,
d
em
an
d
s
ap
p
r
o
p
r
iate
p
r
ep
r
o
ce
s
s
in
g
an
d
ex
p
l
o
r
ato
r
y
d
ata
an
aly
s
is
(
E
DA)
.
T
h
er
e
ar
e
m
an
y
o
b
s
tacle
s
in
h
er
en
t
to
th
e
f
in
an
cial
tr
an
s
ac
tio
n
d
ataset.
T
h
ese
ar
e
class
im
b
ala
n
ce
,
s
k
ewe
d
d
is
tr
ib
u
tio
n
an
d
m
ix
tu
r
e
o
f
c
o
n
tin
u
o
u
s
an
d
ca
teg
o
r
ical
f
ea
tu
r
es.
T
h
ese
p
r
o
b
lem
s
in
th
e
d
ata
ar
e
n
o
t
tack
led
,
an
d
wh
ic
h
ca
n
lead
to
p
o
o
r
m
o
d
el
p
er
f
o
r
m
an
ce
.
T
h
is
d
em
o
n
s
tr
ates
th
e
p
o
te
n
tial
b
en
ef
its
o
f
c
o
m
p
r
eh
e
n
s
iv
e
d
ata
p
r
ep
r
o
ce
s
s
in
g
f
o
r
en
h
an
cin
g
p
r
e
d
ictiv
e
ac
cu
r
ac
y
as
well
as
f
o
r
o
b
tain
in
g
v
alu
a
b
le
in
s
ig
h
ts
.
Patter
n
s
ass
o
ciate
d
with
f
r
au
d
an
d
n
o
n
-
f
r
au
d
tr
an
s
ac
tio
n
s
wer
e
in
v
esti
g
ated
in
th
is
s
tu
d
y
.
W
e
h
av
e
lear
n
t
f
r
o
m
th
is
s
tu
d
y
th
e
s
ig
n
if
ican
ce
o
f
tim
e
p
er
io
d
s
o
f
r
elate
d
tr
an
s
ac
tio
n
s
,
th
e
b
e
h
av
io
r
o
f
cu
s
to
m
er
s
,
m
er
ch
an
t
r
is
k
p
r
o
f
iles
(
e
x
ter
n
al
f
r
au
d
)
,
th
e
i
n
d
u
s
tr
y
'
s
s
u
s
ce
p
tib
ilit
y
to
f
r
a
u
d
an
d
ag
e
g
r
o
u
p
o
f
d
em
o
g
r
ap
h
y
(
i
n
t
er
n
al
f
r
au
d
)
.
I
n
th
e
s
ec
tio
n
,
we
r
ep
o
r
t r
esu
lts
f
r
o
m
ex
p
lo
r
ato
r
y
v
is
u
aliza
tio
n
an
d
s
tatis
tical
an
aly
s
is
in
s
i
g
h
ts
.
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
F
r
a
u
d
d
etec
tio
n
u
s
in
g
Ta
b
N
et
*
cla
s
s
ifier
a
ma
ch
in
e
lea
r
n
in
g
a
p
p
r
o
a
ch
(
G.
A
n
is
h
Ma
r
y
)
605
T
ab
Net
co
n
s
is
ts
o
f
a
s
eq
u
en
ce
o
f
d
ec
is
io
n
s
tep
s
,
as
illu
s
tr
ated
in
Fig
u
r
e
1
.
An
atten
tio
n
m
ask
is
p
r
o
d
u
ce
d
at
ea
ch
s
tep
to
atten
d
to
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
e
s
.
Su
ch
a
co
n
f
ig
u
r
atio
n
f
a
cilitates
m
o
d
elin
g
h
ig
h
-
o
r
d
er
in
ter
ac
tio
n
s
am
o
n
g
tr
an
s
ac
tio
n
al,
d
em
o
g
r
ap
h
ic,
g
e
o
g
r
a
p
h
ical,
an
d
tem
p
o
r
al
f
ea
tu
r
es,
wh
ile
m
ain
tain
in
g
in
ter
p
r
etab
ilit
y
f
o
r
f
r
a
u
d
d
etec
tio
n
.
Fra
u
d
is
m
o
r
e
lik
ely
to
h
ap
p
en
in
m
u
c
h
s
m
aller
tim
e
in
ter
v
als,
e.
g
.
r
i
g
h
t
af
ter
a
p
r
ev
io
u
s
tr
an
s
ac
tio
n
.
T
h
e
r
ap
i
d
-
f
ir
e
n
atu
r
e
o
f
th
e
tr
a
n
s
ac
tio
n
s
s
u
g
g
ests
th
at
th
e
cr
i
m
in
als
ar
e
tr
y
i
n
g
t
o
r
u
n
a
s
er
ies
o
f
ch
ar
g
es
o
n
a
ca
r
d
b
ef
o
r
e
th
e
ac
co
u
n
t
is
b
lo
ck
ed
o
r
f
lag
g
ed
.
On
th
e
o
th
er
h
a
n
d
,
leg
itima
te
u
s
er
s
th
at
m
ay
tak
e
lo
n
g
er
a
n
d
h
av
e
v
ar
ied
tim
es
b
etwe
en
th
eir
s
p
en
d
in
g
,
w
h
ich
ar
e
m
o
r
e
in
lin
e
with
s
p
en
d
in
g
n
o
r
m
s
.
T
h
e
d
en
s
ity
p
l
o
t
r
ev
ea
led
a
s
ig
n
if
ican
t
s
p
ik
e
o
f
f
r
au
d
u
len
t
b
e
h
av
io
r
in
th
e
0
-
3
0
0
0
s
ec
o
n
d
s
;
th
u
s
,
th
is
tim
e
v
ar
iab
le
m
ig
h
t b
e
a
g
o
o
d
ca
n
d
id
a
te
f
o
r
p
r
e
d
ictiv
e
m
o
d
el
lin
g
.
I
n
Fi
g
u
r
e
2
il
lu
s
tr
ates
t
h
e
d
i
s
tr
i
b
u
ti
o
n
o
f
to
tal
t
r
a
n
s
a
cti
o
n
s
p
er
c
u
s
t
o
m
e
r
,
d
is
t
in
g
u
is
h
i
n
g
b
etwe
e
n
leg
iti
m
a
te
a
n
d
f
r
a
u
d
u
le
n
t
a
cc
o
u
n
ts
.
T
h
e
x
-
a
x
is
r
e
p
r
es
en
ts
th
e
“N
u
m
b
e
r
o
f
T
r
an
s
ac
ti
o
n
s
p
e
r
C
u
s
t
o
m
e
r
,
”
w
h
ile
th
e
y
-
a
x
is
i
n
d
i
ca
t
es
“D
e
n
s
it
y
,
”
r
e
f
le
cti
n
g
h
o
w
c
o
m
m
o
n
e
ac
h
t
r
a
n
s
a
cti
o
n
c
o
u
n
t
is
a
f
te
r
n
o
r
m
ali
zi
n
g
t
h
e
d
is
t
r
i
b
u
ti
o
n
s
.
T
h
e
b
l
u
e
cu
r
v
e
r
e
p
r
ese
n
ts
l
e
g
iti
m
a
te
c
u
s
t
o
m
e
r
s
,
a
n
d
t
h
e
r
ed
cu
r
v
e
r
e
p
r
ese
n
ts
f
r
au
d
u
l
en
t
cu
s
t
o
m
e
r
s
.
At
a
n
y
g
iv
e
n
p
o
in
t a
lo
n
g
t
h
e
x
-
ax
is
,
a
h
i
g
h
e
r
cu
r
v
e
i
n
d
ic
ate
s
th
at
ty
p
e
o
f
c
u
s
t
o
m
er
is
m
o
r
e
f
r
e
q
u
e
n
t
at
th
at
t
r
a
n
s
a
cti
o
n
c
o
u
n
t
.
T
h
e
l
ar
g
e
o
v
e
r
la
p
p
i
n
g
p
e
ak
o
n
th
e
l
ef
t
s
h
o
ws th
at
m
o
s
t
c
u
s
t
o
m
e
r
s
,
w
h
et
h
er
l
eg
iti
m
at
e
o
r
f
r
a
u
d
u
le
n
t
,
c
o
n
d
u
ct
r
el
ati
v
ely
f
ew
t
r
a
n
s
a
cti
o
n
s
.
Sm
all
er
p
e
ak
s
f
ar
th
er
t
o
t
h
e
r
i
g
h
t
c
o
r
r
esp
o
n
d
t
o
h
i
g
h
ly
ac
ti
v
e
c
u
s
t
o
m
er
s
w
it
h
t
h
o
u
s
a
n
d
s
o
f
t
r
a
n
s
a
cti
o
n
s
.
T
h
ese
p
e
a
k
s
a
p
p
e
ar
i
n
b
o
t
h
c
u
r
v
es
b
u
t
wi
th
s
li
g
h
tl
y
d
i
f
f
er
e
n
t
h
ei
g
h
ts
,
s
u
g
g
est
in
g
t
h
at
c
er
tai
n
h
ig
h
-
a
cti
v
it
y
r
a
n
g
es
m
a
y
b
e
m
o
r
e
o
r
less
ass
o
ci
at
ed
wi
th
f
r
a
u
d
u
le
n
t
b
e
h
a
v
i
o
r
.
Fig
u
r
e
1
.
T
im
e
s
in
ce
las
t tr
an
s
ac
tio
n
d
is
tr
ib
u
tio
n
Fig
u
r
e
2
.
C
u
s
to
m
er
tr
a
n
s
ac
tio
n
co
u
n
t
d
is
tr
ib
u
tio
n
T
h
e
s
tu
d
y
aim
s
to
h
ig
h
lig
h
t
in
th
e
m
er
ch
an
t
r
is
k
s
co
r
es
in
r
elatio
n
to
f
r
au
d
lab
els
an
d
also
p
r
o
v
id
e
d
f
u
r
th
er
ju
s
tific
atio
n
f
o
r
th
e
u
s
e
o
f
m
er
ch
a
n
t
lev
el
ch
ar
ac
ter
is
tics
.
T
h
e
r
esu
lts
Fig
u
r
e
3
s
h
o
ws
also
in
d
icate
d
th
at
tr
an
s
ac
tio
n
s
w
h
er
e
f
r
a
u
d
o
cc
u
r
r
e
d
alwa
y
s
h
ad
h
ig
h
e
r
m
er
c
h
an
t
r
is
k
s
co
r
es
th
an
tr
an
s
ac
tio
n
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
60
1
-
6
1
3
606
wh
er
e
f
r
au
d
d
id
n
o
t
o
cc
u
r
.
Fu
r
th
er
m
o
r
e
,
m
er
c
h
an
ts
wh
er
e
f
r
au
d
o
cc
u
r
r
e
d
h
ad
a
wid
er
r
a
n
g
e
as
well
as
s
o
m
e
ex
tr
em
e
o
u
tlie
r
s
,
in
d
icatin
g
th
at
f
r
au
d
d
o
es
n
o
t
o
cc
u
r
alo
n
g
a
co
n
tin
u
u
m
wh
er
e
th
e
r
e
ar
e
m
in
o
r
d
i
f
f
er
en
ce
s
,
r
ath
er
,
th
er
e
ar
e
v
a
r
io
u
s
lev
e
ls
o
f
f
r
au
d
in
ten
s
ity
.
T
h
e
r
a
n
g
e
o
f
v
ar
iab
ilit
y
m
a
d
e
clea
r
th
at
m
o
d
els
m
u
s
t
in
clu
d
e
m
er
ch
a
n
t
lev
el
an
o
m
alies,
as
th
e
p
o
in
t
o
f
s
ale
i
s
g
en
er
ally
o
n
e
o
f
th
e
p
r
im
ar
y
m
ec
h
an
is
m
s
th
r
o
u
g
h
wh
ich
f
r
au
d
m
an
if
ests
.
T
h
e
f
r
au
d
r
ate
b
y
m
er
c
h
an
t
ca
teg
o
r
izatio
n
g
a
v
e
in
s
ig
h
ts
in
to
th
e
s
ec
to
r
-
b
ased
o
u
tco
m
e
o
f
t
h
e
f
r
au
d
.
Fig
u
r
e
4
s
h
o
ws
th
e
m
er
ch
an
t
ca
teg
o
r
ies
s
u
ch
as
s
h
o
p
p
in
g
_
n
et,
m
is
c_
n
et,
an
d
g
r
o
ce
r
y
_
p
o
s
ca
teg
o
r
ies
all
s
h
o
wed
ex
ce
s
s
iv
e
f
r
au
d
r
ates
,
wh
ich
im
p
lies
th
at
g
r
o
ce
r
y
p
u
r
ch
ases
m
ay
b
e
m
o
r
e
lik
ely
to
b
e
ex
p
l
o
ited
wh
eth
er
o
n
lin
e
o
r
at
th
e
POS.
T
h
e
lar
g
e
m
a
r
k
u
p
,
esp
ec
ially
in
g
r
o
ce
r
y
r
etail
m
ak
es
th
is
ty
p
e
o
f
p
u
r
ch
as
e
attr
ac
tiv
e
to
s
o
m
e
f
r
au
d
s
ter
s
.
C
o
n
v
er
s
ely
,
ca
teg
o
r
ies
s
u
ch
as
p
er
s
o
n
al
ca
r
e
an
d
en
ter
tain
m
en
t
h
ad
v
er
y
litt
le
ass
o
ciatio
n
with
f
r
au
d
.
T
h
ese
f
in
d
in
g
s
s
u
g
g
est
th
at
f
r
a
u
d
s
ter
s
ar
e
n
o
t
tar
g
etin
g
s
ec
to
r
s
th
at
ar
e
n
o
t
f
r
e
q
u
en
tly
ac
ce
s
s
ed
o
r
p
u
r
c
h
ased
.
Acc
o
r
d
in
g
ly
,
it
is
f
u
n
d
am
e
n
tally
n
ec
ess
ar
y
to
ap
p
r
o
p
r
iately
e
n
co
d
e
m
er
ch
an
t
t
y
p
es
b
y
ca
teg
o
r
ical
lev
els
as
th
ese
en
c
o
d
in
g
s
ac
t
as
p
r
io
r
k
n
o
wled
g
e
o
f
f
r
a
u
d
s
ter
s
lev
er
a
g
ed
in
p
r
e
p
r
o
ce
s
s
in
g
.
C
o
d
i
n
g
d
o
m
ain
an
d
m
er
ch
a
n
t
ca
teg
o
r
y
v
u
ln
e
r
ab
ilit
ies
ca
n
allo
w
th
e
p
r
ed
ictiv
e
s
y
s
tem
to
o
b
je
ctiv
ely
d
if
f
er
en
tiate
f
r
au
d
m
o
n
it
o
r
in
g
p
r
o
ce
s
s
es
an
d
ac
tiv
ities
in
h
ig
h
v
o
l
u
m
e,
p
r
ed
ictab
le,
an
d
u
n
s
tr
u
ctu
r
e
d
s
ec
to
r
s
to
elicit
g
r
ea
ter
ef
f
icien
cies in
f
r
au
d
,
g
iv
en
th
e
o
f
ten
-
lar
g
e
p
er
ce
n
tag
e
o
f
lo
s
s
.
Fig
u
r
e
3
.
Me
r
c
h
an
t r
is
k
s
co
r
e
v
s
f
r
au
d
Fig
u
r
e
4
.
Fra
u
d
r
ate
b
y
m
e
r
ch
an
t c
ateg
o
r
y
An
ex
am
in
atio
n
o
f
th
e
f
r
au
d
d
is
tr
ib
u
tio
n
ac
r
o
s
s
ag
e
g
r
o
u
p
s
Fig
u
r
e
5
s
h
o
ws
th
at
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
wer
e
m
o
s
t
co
n
ce
n
tr
ated
in
th
e
3
1
–
5
0
ag
e
r
an
g
e,
wh
ich
also
co
r
r
esp
o
n
d
s
to
th
e
d
em
o
g
r
a
p
h
ic
with
th
e
h
ig
h
est
tr
an
s
ac
tio
n
a
ctiv
ity
.
T
h
is
co
r
r
elatio
n
im
p
lies
th
at
f
r
au
d
p
r
e
v
alen
ce
is
p
ar
tly
in
f
lu
en
ce
d
b
y
v
o
lu
m
e
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
F
r
a
u
d
d
etec
tio
n
u
s
in
g
Ta
b
N
et
*
cla
s
s
ifier
a
ma
ch
in
e
lea
r
n
in
g
a
p
p
r
o
a
ch
(
G.
A
n
is
h
Ma
r
y
)
607
o
f
u
s
ag
e
.
Yo
u
n
g
er
u
s
er
s
(
b
elo
w
2
0
y
ea
r
s
)
an
d
o
ld
er
u
s
er
s
(
ab
o
v
e
7
0
y
ea
r
s
)
h
ad
less
f
r
au
d
u
len
t
ca
s
es,
lik
el
y
d
u
e
to
lo
wer
f
r
eq
u
en
cy
o
f
u
s
e
in
d
ig
ital
f
in
an
cial
s
er
v
ices.
Ag
e
alo
n
e
m
ay
n
o
t
m
atter
;
h
o
wev
er
,
it
ca
n
b
e
v
alu
ab
le
co
n
tex
t in
co
n
ju
n
ctio
n
with
o
th
er
f
ea
tu
r
es.
I
n
Fig
u
r
e
6
th
e
p
r
ep
r
o
ce
s
s
in
g
lo
g
ics
s
h
o
w
th
at
f
r
au
d
d
et
ec
tio
n
ca
n
n
o
t
o
n
ly
r
ely
o
n
t
r
an
s
ac
tio
n
am
o
u
n
ts
an
d
b
asic
f
r
a
u
d
lab
e
ls
.
R
ath
er
,
a
b
ig
g
er
p
ictu
r
e
o
f
f
r
au
d
is
o
b
s
er
v
ed
b
y
f
ac
to
r
in
g
in
tim
e
p
atter
n
s
,
cu
s
to
m
er
h
ab
its
,
ch
a
r
ac
ter
is
tics
o
f
m
er
ch
a
n
ts
,
an
d
d
em
o
g
r
ap
h
ic
b
ac
k
g
r
o
u
n
d
.
E
ac
h
f
ea
t
u
r
e
h
as
b
ee
n
p
r
o
p
e
r
ly
s
ca
led
,
f
u
lly
en
co
d
ed
an
d
k
e
p
t
th
r
o
u
g
h
m
o
d
ellin
g
in
o
r
d
e
r
to
d
ev
elo
p
m
ac
h
in
e
lear
n
in
g
m
o
d
els
with
g
o
o
d
p
r
ed
ictio
n
p
r
o
p
e
r
ties
.
T
h
e
p
o
ten
tial
f
o
r
s
tr
o
n
g
a
n
d
g
e
n
er
al
izab
le
f
r
au
d
d
etec
tio
n
is
m
ad
e
p
o
s
s
ib
le
th
r
o
u
g
h
p
r
e
p
r
o
ce
s
s
in
g
b
ec
au
s
e
p
r
o
ce
s
s
in
g
im
b
alan
ce
s
r
em
o
v
in
g
n
o
is
e
an
d
esti
m
atin
g
v
alu
ab
le
r
elat
io
n
s
h
ip
s
is
cr
u
cial.
Fig
u
r
e
5
.
Fra
u
d
d
is
tr
ib
u
tio
n
ac
r
o
s
s
ag
e
g
r
o
u
p
s
Fig
u
r
e
6
.
Fra
u
d
v
s
L
eg
it
tr
a
n
s
ac
tio
n
s
b
y
h
o
u
r
o
f
d
a
y
5.
F
RUIT
F
U
L
M
E
T
H
O
DO
L
O
G
Y
W
e
p
r
o
p
o
s
ed
ar
c
h
itectu
r
e
s
h
o
wn
in
Fig
u
r
e
7
is
b
u
ilt
o
n
T
ab
Net*
a
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
s
p
ec
if
ically
d
esig
n
ed
f
o
r
tab
u
lar
d
ata
th
at
co
m
b
in
es
h
ig
h
p
er
f
o
r
m
a
n
ce
with
in
h
er
en
t
i
n
ter
p
r
etab
ilit
y
.
T
h
e
m
o
d
el
o
p
er
ates
th
r
o
u
g
h
a
s
eq
u
en
tial,
m
u
lti
-
s
tep
r
ea
s
o
n
in
g
p
r
o
ce
s
s
.
B
ased
o
n
ea
ch
s
tep
i
n
th
e
i
n
p
u
t
f
ea
tu
r
es
f
ir
s
t
p
ass
th
r
o
u
g
h
a
s
h
ar
ed
Fea
tu
r
e
T
r
an
s
f
o
r
m
er
n
etwo
r
k
th
r
o
u
g
h
a
s
er
ies
o
f
f
u
ll
y
c
o
n
n
ec
te
d
lay
er
s
with
g
ated
lin
ea
r
u
n
it
(
GL
U)
ac
tiv
atio
n
s
.
T
o
cr
ea
te
a
p
r
o
ce
s
s
in
th
is
r
ep
r
esen
tatio
n
is
th
en
f
ed
in
to
an
Atten
tiv
e
T
r
an
s
f
o
r
m
e
r
b
u
t
wh
ich
a
cts
as
a
f
ea
tu
r
e
s
elec
tio
n
m
ec
h
an
is
m
.
Usi
n
g
a
p
r
io
r
s
ca
le
o
f
f
ea
t
u
r
e
u
s
ag
e
f
r
o
m
th
e
p
r
ev
io
u
s
s
tep
a
n
d
th
e
So
f
tMa
x
ac
tiv
atio
n
f
u
n
ctio
n
(
as
s
p
ec
if
ied
b
y
m
ask
_
ty
p
e=
'
s
p
ar
s
em
ax
'
)
,
th
e
Atten
tiv
e
T
r
an
s
f
o
r
m
e
r
g
e
n
er
ates
a
s
p
a
r
s
e,
in
s
tan
ce
-
wis
e
m
ask
th
at
s
elec
tiv
ely
f
o
cu
s
es
o
n
o
n
ly
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
es
f
o
r
th
at
s
p
ec
if
ic
d
ec
i
s
io
n
s
tep
.
T
h
en
th
e
s
elec
ted
f
ea
t
u
r
es
ar
e
th
en
p
r
o
ce
s
s
ed
b
y
a
s
tep
-
d
ep
en
d
e
n
t
Featu
r
e
T
r
an
s
f
o
r
m
e
r
,
with
p
a
r
t
o
f
its
o
u
tp
u
t
co
n
tr
ib
u
tin
g
to
th
e
f
in
al
p
r
e
d
ictio
n
an
d
an
o
th
er
p
ar
t
b
ein
g
f
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
60
1
-
6
1
3
608
b
ac
k
to
g
u
id
e
th
e
f
ea
t
u
r
e
s
elec
tio
n
in
th
e
n
e
x
t
s
tep
.
T
h
is
p
r
o
ce
s
s
en
co
u
r
ag
es
th
e
m
o
d
el
t
o
lear
n
a
co
ll
ec
tiv
e
d
ec
is
io
n
f
r
o
m
m
u
ltip
le
r
ea
s
o
n
i
n
g
s
tep
s
.
Fig
u
r
e
7
.
Pro
p
o
s
ed
T
a
b
leNe
t*
ar
ch
itectu
r
e
–
a
h
ig
h
-
lev
el
ap
p
r
o
ac
h
Fin
ally
,
th
e
o
u
tp
u
ts
f
r
o
m
all
s
tep
s
ar
e
ag
g
r
eg
ated
an
d
p
ass
ed
th
r
o
u
g
h
a
f
i
n
al
lin
ea
r
lay
er
an
d
So
f
tMa
x
ac
tiv
atio
n
to
p
r
o
d
u
c
e
th
e
class
if
icatio
n
p
r
o
b
ab
ilit
ies
f
o
r
f
r
au
d
o
r
n
o
n
-
f
r
a
u
d
.
C
r
u
cially
,
th
e
m
ask
s
f
r
o
m
all
s
tep
s
an
d
ac
r
o
s
s
all
in
s
tan
ce
s
ca
n
b
e
ag
g
r
e
g
ated
to
c
alcu
late
th
e
g
lo
b
al
f
ea
tu
r
e
im
p
o
r
tan
ce
,
p
r
o
v
id
i
n
g
a
clea
r
,
m
o
d
el
-
wid
e
ex
p
lan
at
io
n
o
f
wh
ich
f
ac
t
o
r
s
(
lik
e
d
i
s
tan
ce
,
m
er
ch
an
t_
r
is
k
,
o
r
lo
g
_
am
t)
wer
e
m
o
s
t
in
f
lu
en
tial
in
th
e
m
o
d
el'
s
p
r
e
d
ictio
n
s
.
T
h
e
atten
tio
n
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
m
ec
h
an
is
m
i
n
T
ab
Net*
en
ab
les
in
ter
p
r
etab
le
d
ec
is
io
n
lo
g
ic,
al
ig
n
in
g
with
tr
u
s
two
r
th
y
AI
p
r
in
cip
les,
m
ak
in
g
it
s
u
itab
le
f
o
r
f
in
an
cial
s
ec
u
r
ity
an
d
r
eg
u
lato
r
y
co
m
p
lian
ce
.
T
h
is
en
tire
ar
ch
itectu
r
e
is
o
p
tim
ized
u
s
in
g
a
lear
n
in
g
r
ate
o
f
2
e
-
2
with
a
Step
L
R
s
ch
ed
u
ler
an
d
is
tr
ain
ed
in
b
atch
es
o
f
1
0
2
4
with
a
v
ir
tu
al
b
atch
s
ize
o
f
1
2
8
f
o
r
s
tab
le
lear
n
in
g
o
n
th
e
im
b
alan
ce
d
d
ataset
6.
M
O
DE
L
T
R
AIN
I
NG
AND
E
VA
L
UA
T
I
O
N
T
h
e
class
if
ier
T
ab
Net*
was
im
p
lem
en
ted
u
s
in
g
t
h
e
T
en
s
o
r
Flo
w
f
r
am
ewo
r
k
,
wh
ich
wo
r
k
s
b
est
with
tab
u
lar
tr
a
n
s
ac
tio
n
al
d
ata
.
T
h
e
m
o
d
el
co
m
b
in
es
th
e
f
ea
t
u
r
e
t
r
an
s
f
o
r
m
er
an
d
atten
tiv
e
tr
an
s
f
o
r
m
er
m
o
d
u
les
f
o
r
s
eq
u
en
tial,
in
s
tan
ce
-
wis
e
f
ea
tu
r
e
s
elec
tio
n
an
d
r
e
aso
n
in
g
.
At
ea
ch
d
ec
is
io
n
s
tep
,
th
e
in
p
u
t
f
ea
tu
r
es
f
ir
s
t
g
o
th
r
o
u
g
h
a
s
h
ar
ed
T
r
an
s
f
o
r
m
er
.
T
h
is
co
n
s
is
ts
o
f
f
u
lly
co
n
n
ec
ted
lay
er
s
with
g
ated
lin
ea
r
u
n
it
(
GL
U)
ac
tiv
atio
n
s
.
T
h
ese
lay
e
r
s
allo
w
f
o
r
n
o
n
-
lin
ea
r
tr
an
s
f
o
r
m
atio
n
s
an
d
r
e
d
u
ce
d
im
en
s
io
n
alit
y
.
T
h
e
o
u
t
p
u
t
th
en
g
o
es
to
an
Atten
tiv
e
T
r
an
s
f
o
r
m
er
,
wh
ich
u
s
es
a
s
p
ar
s
em
ax
ac
tiv
atio
n
to
cr
ea
te
a
f
ea
tu
r
e
s
elec
tio
n
m
ask
.
T
h
is
m
ask
h
elp
s
th
e
n
etwo
r
k
f
o
cu
s
o
n
th
e
m
o
s
t
r
ele
v
an
t f
ea
tu
r
es
f
o
r
ea
ch
tr
an
s
ac
tio
n
,
im
p
r
o
v
in
g
b
o
th
p
er
f
o
r
m
a
n
ce
an
d
u
n
d
e
r
s
tan
d
in
g
.
At
ea
ch
s
te
p
,
a
s
tep
-
s
p
ec
if
ic
Featu
r
e
T
r
an
s
f
o
r
m
er
tak
es
th
e
s
elec
ted
f
ea
tu
r
es
an
d
m
ak
es
p
ar
tial
p
r
ed
ictio
n
s
.
A
s
eq
u
en
tial
d
ec
is
io
n
p
r
o
ce
s
s
co
m
b
in
es
th
ese
p
ar
tial
r
e
s
u
lts
to
cr
ea
te
th
e
f
in
al
o
u
tp
u
t
lay
er
.
T
h
e
So
f
tMa
x
ac
tiv
atio
n
in
th
e
last
lay
er
tu
r
n
s
th
e
c
o
m
b
in
ed
d
ec
is
io
n
s
co
r
es
i
n
to
clas
s
p
r
o
b
a
b
ilit
ies
th
at
s
h
o
w
h
o
w
lik
ely
it is
th
at
a
tr
a
n
s
ac
tio
n
is
f
ak
e
o
r
r
ea
l
6
.
1
.
T
ra
ini
ng
c
o
nfig
ura
t
io
n
T
h
e
Ad
am
o
p
tim
izer
was
u
s
ed
to
tr
ai
n
th
e
m
o
d
el,
s
tar
tin
g
with
a
lear
n
in
g
r
ate
o
f
2
×
1
0
⁻²
an
d
th
e
n
u
tili
zin
g
a
Step
L
R
s
ch
ed
u
le
r
to
s
lo
wly
lo
wer
th
e
lear
n
in
g
r
a
te
af
ter
ev
er
y
two
e
p
o
ch
s
.
W
e
tr
ain
ed
th
e
m
o
d
el
f
o
r
1
0
ep
o
ch
s
u
s
in
g
a
b
atch
s
ize
o
f
1
0
2
4
an
d
a
v
ir
t
u
al
b
atch
s
ize
o
f
1
2
8
to
k
ee
p
th
e
g
r
a
d
ie
n
t
u
p
d
ates
s
tab
le
o
n
th
e
im
b
alan
ce
d
d
ataset.
T
h
e
l
o
s
s
f
u
n
ctio
n
was
b
in
ar
y
cr
o
s
s
-
en
tr
o
p
y
,
wh
ich
is
g
o
o
d
f
o
r
b
in
ar
y
class
if
icatio
n
ap
p
licatio
n
s
.
C
lass
weig
h
ts
we
r
e
u
s
ed
to
p
u
n
is
h
m
is
class
if
y
in
g
m
in
o
r
ity
(
f
r
au
d
u
len
t)
d
ata.
T
o
m
itig
ate
o
v
e
r
f
itti
n
g
,
s
ev
er
a
l r
eg
u
lar
izatio
n
tec
h
n
iq
u
es we
r
e
ap
p
lied
:
a.
Sp
ar
s
e
r
eg
u
lar
izatio
n
(
λ
ₛ
=
1
e−
4
)
o
n
atten
tio
n
m
ask
s
,
en
s
u
r
in
g
th
at
o
n
ly
th
e
m
o
s
t r
elev
an
t
f
ea
tu
r
es we
r
e
u
tili
ze
d
p
er
in
s
tan
ce
.
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
F
r
a
u
d
d
etec
tio
n
u
s
in
g
Ta
b
N
et
*
cla
s
s
ifier
a
ma
ch
in
e
lea
r
n
in
g
a
p
p
r
o
a
ch
(
G.
A
n
is
h
Ma
r
y
)
609
b.
B
atch
n
o
r
m
aliza
tio
n
with
in
Fe
atu
r
e
T
r
an
s
f
o
r
m
er
b
lo
ck
s
to
s
t
ab
ilize
ac
tiv
atio
n
d
is
tr
ib
u
tio
n
s
.
c.
E
a
r
l
y
s
t
o
p
p
i
n
g
b
a
s
e
d
o
n
v
a
l
i
d
a
t
i
o
n
A
U
C
,
p
r
e
v
e
n
t
i
n
g
u
n
n
e
c
e
s
s
a
r
y
t
r
a
i
n
i
n
g
e
p
o
c
h
s
o
n
c
e
p
e
r
f
o
r
m
a
n
c
e
p
l
a
t
e
a
u
e
d
.
7.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
T
a
b
Net*
is
tr
ain
ed
o
n
th
e
p
r
o
ce
s
s
ed
d
ata
s
et
o
f
th
e
tr
an
s
ac
tio
n
al
d
a
ta
f
o
r
ten
ep
o
ch
s
.
T
a
b
le
1
is
th
e
m
o
d
el
p
er
f
o
r
m
an
ce
co
m
p
ar
is
o
n
atta
in
ed
a
s
tab
le
p
o
in
t
o
f
c
o
n
v
e
r
g
en
ce
in
th
e
eig
h
th
ep
o
ch
its
elf
,
w
h
er
e
th
e
h
ig
h
e
s
t
test
ac
cu
r
ac
y
o
f
9
9
.
6
9
%,
F1
-
s
co
r
e
o
f
0
.
9
7
5
,
an
d
th
e
h
i
g
h
est
v
alu
e
o
f
t
h
e
R
OC
-
AU
C
o
f
0
.
9
5
6
was
attain
ed
.
T
h
is
s
u
d
d
en
in
cr
ea
s
e
in
AUC
v
alu
es
f
r
o
m
ep
o
ch
o
n
e
(
0
.
8
2
)
to
e
p
o
ch
f
o
u
r
(
0
.
9
5
)
em
p
h
asizes th
e
ef
f
icien
cy
o
f
th
e
m
o
d
el
i
n
ex
tr
ac
tin
g
r
elev
an
t f
ea
tu
r
es f
r
o
m
th
e
i
m
b
a
lan
ce
d
d
ata.
Featu
r
e
im
p
o
r
tan
ce
an
aly
s
is
r
ev
ea
led
th
at
th
e
m
o
s
t
in
f
lu
en
t
ial
v
ar
iab
les
wer
e
th
e
lo
g
o
f
t
r
an
s
ac
tio
n
am
o
u
n
t
(
lo
g
_
am
t,
5
3
%),
tr
an
s
ac
tio
n
ty
p
e
(
tr
a
n
s
_
ca
t,
2
3
%),
m
er
ch
an
t
r
is
k
s
co
r
e
(
r
is
k
_
s
co
r
e,
1
0
.
6
%),
a
n
d
city
p
o
p
u
latio
n
(
city
_
p
o
p
,
5
.
9
%)
.
T
im
e
v
ar
iab
les,
d
em
o
g
r
ap
h
ics,
an
d
n
u
m
er
ic
v
ar
iab
les
ar
e
o
f
m
o
d
er
ate
im
p
o
r
tan
ce
,
wh
er
e
v
a
r
iab
les
h
av
in
g
an
y
k
in
d
o
f
id
s
u
ch
as
ca
r
d
n
o
/car
d
I
D
(
ca
r
d
No
/car
d
I
D)
ar
e
o
f
n
eg
lig
ib
le
im
p
o
r
tan
ce
.
I
n
s
u
m
m
a
r
y
,
th
e
f
in
d
in
g
s
s
h
o
w
th
at
T
ab
Net*
o
u
t
p
er
f
o
r
m
s
s
tan
d
ar
d
m
ac
h
in
e
lea
r
n
in
g
b
e
n
ch
m
ar
k
s
n
o
t
o
n
ly
in
ter
m
s
o
f
ac
c
u
r
ac
y
an
d
r
eliab
ilit
y
.
Fu
r
th
er
m
o
r
e,
it
is
less
b
lack
b
o
x
ap
p
ea
l
th
at
d
em
o
n
s
tr
ates
tr
an
s
p
ar
en
cy
an
d
ac
co
u
n
ta
b
le
b
y
th
e
s
tak
eh
o
ld
er
s
.
Ov
er
all,
T
ab
Net*
r
ep
r
esen
ts
a
tr
u
s
tw
o
r
th
y
,
in
ter
p
r
etab
le,
an
d
s
ca
lab
le
ap
p
r
o
ac
h
to
th
e
p
r
o
b
lem
o
f
f
in
a
n
cial
f
r
au
d
d
e
tectio
n
.
T
h
e
im
p
r
o
v
em
e
n
ts
in
p
er
f
o
r
m
an
ce
an
d
in
ter
p
r
etab
le
a
r
e
s
ig
n
if
ican
t
a
d
v
an
ce
s
o
v
er
e
x
is
tin
g
to
o
ls
,
b
y
g
r
ea
tly
r
e
d
u
cin
g
f
alse
n
eg
ativ
es
an
d
f
i
n
an
cial
r
is
k
to
o
r
g
a
n
izatio
n
s
.
T
ab
le
1
.
Mo
d
el
p
er
f
o
r
m
a
n
ce
c
o
m
p
ar
is
o
n
M
o
d
e
l
/
A
p
p
r
o
a
c
h
A
c
c
u
r
a
c
y
(
%)
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
S
c
o
r
e
R
O
C
-
AUC
R
e
mar
k
s
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
9
2
.
1
5
0
.
8
8
0
.
8
1
0
.
8
4
0
.
8
6
Li
n
e
a
r
b
a
s
e
l
i
n
e
;
l
i
mi
t
e
d
n
o
n
-
l
i
n
e
a
r
c
a
p
t
u
r
e
D
e
c
i
s
i
o
n
t
r
e
e
9
3
.
4
0
0
.
9
0
0
.
8
5
0
.
8
7
0
.
8
8
I
n
t
e
r
p
r
e
t
a
b
l
e
b
u
t
o
v
e
r
f
i
t
s
R
a
n
d
o
m f
o
r
e
s
t
9
5
.
8
0
0
.
9
3
0
.
8
9
0
.
9
1
0
.
9
1
R
o
b
u
st
e
n
sem
b
l
e
o
p
a
q
u
e
d
e
c
i
si
o
n
s
S
V
M
9
4
.
2
0
0
.
9
1
0
.
8
7
0
.
8
9
0
.
8
9
Ef
f
e
c
t
i
v
e
a
f
t
e
r
s
c
a
l
i
n
g
;
h
i
g
h
c
o
s
t
X
G
B
o
o
st
9
6
.
5
0
0
.
9
4
0
.
9
1
0
.
9
2
0
.
9
3
G
r
a
d
i
e
n
t
-
b
o
o
st
e
d
t
r
e
e
s;
l
o
w
t
r
a
n
s
p
a
r
e
n
c
y
Li
g
h
t
G
B
M
9
6
.
8
0
0
.
9
5
0
.
9
2
0
.
9
3
0
.
9
4
F
a
st
b
o
o
s
t
i
n
g
;
s
t
i
l
l
a
b
l
a
c
k
b
o
x
D
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
(
D
N
N
)
9
7
.
2
0
0
.
9
6
0
.
9
4
0
.
9
5
0
.
9
4
H
i
g
h
p
e
r
f
o
r
m
a
n
c
e
n
o
n
-
i
n
t
e
r
p
r
e
t
a
b
l
e
En
se
mb
l
e
h
y
b
r
i
d
mo
d
e
l
s
9
7
.
8
0
0
.
9
6
0
.
9
5
0
.
9
5
0
.
9
5
S
t
r
o
n
g
b
u
t
r
e
s
o
u
r
c
e
-
i
n
t
e
n
si
v
e
P
r
o
p
o
se
d
T
a
b
N
e
t
*
9
9
.
6
9
0
.
9
8
0
.
9
7
0
.
9
7
5
0
.
9
5
6
H
i
g
h
e
s
t
a
c
c
u
r
a
c
y
w
i
t
h
i
n
t
e
r
p
r
e
t
a
b
i
l
i
t
y
v
i
a
a
t
t
e
n
t
i
o
n
7
.
1
.
St
a
t
is
t
ica
l
s
ig
nifica
nce
a
na
ly
s
is
T
o
f
in
d
o
u
t
if
th
e
p
er
f
o
r
m
an
ce
im
p
r
o
v
em
en
ts
o
f
th
e
p
r
o
p
o
s
ed
T
ab
Net*
m
o
d
el
o
v
e
r
c
o
n
v
en
ti
o
n
al
m
eth
o
d
s
wer
e
s
tatis
tical
ly
s
i
g
n
if
ican
t,
we
u
s
ed
a
p
air
ed
t
-
test
an
d
a
W
ilco
x
o
n
s
ig
n
ed
-
r
an
k
test
wer
e
co
n
d
u
cte
d
ac
r
o
s
s
f
iv
e
cr
o
s
s
-
v
alid
atio
n
f
o
ld
s
.
T
ab
le
2
:
s
h
o
w
s
a
co
m
p
ar
is
o
n
o
f
T
a
b
Net*
’
s
Acc
u
r
ac
y
,
F1
-
s
co
r
e,
an
d
R
OC
-
AUC wi
th
th
o
s
e
o
f
t
h
e
s
tr
o
n
g
est b
as
elin
es
—
XGB
o
o
s
t a
n
d
L
ig
h
tGB
M
T
h
e
lo
w
p
-
v
alu
es
(
<
0
.
0
5
)
s
h
o
w
th
at
T
ab
Net*
’
s
p
er
f
o
r
m
an
c
e
im
p
r
o
v
em
en
t
is
s
tatis
tically
s
ig
n
if
ican
t
an
d
u
n
lik
ely
d
u
e
to
r
an
d
o
m
v
ar
iatio
n
.
T
h
is
co
n
f
ir
m
s
th
e
m
o
d
el’
s
s
tr
en
g
th
an
d
ab
ilit
y
to
wo
r
k
in
m
o
d
el
r
ea
l
-
wo
r
ld
f
r
au
d
d
etec
tio
n
s
ce
n
ar
io
s
.
T
h
e
s
u
p
er
io
r
ity
o
f
T
ab
Net*
b
ec
au
s
e
it
u
s
es
atten
tio
n
-
d
r
iv
e
n
f
ea
tu
r
e
s
elec
tio
n
an
d
s
p
ar
s
e
r
ep
r
esen
tatio
n
lear
n
in
g
.
T
h
is
allo
ws
th
e
m
o
d
el
t
o
f
o
c
u
s
o
n
th
e
m
o
s
t
r
elev
an
t
attr
ib
u
tes
f
o
r
ea
ch
tr
an
s
ac
tio
n
d
y
n
am
ically
.
Un
li
k
e
tr
ee
-
b
ased
e
n
s
em
b
les
th
at
ag
g
r
e
g
ate
d
ec
is
io
n
s
ac
r
o
s
s
r
an
d
o
m
s
u
b
s
ets
o
f
f
ea
tu
r
es.
T
ab
Net*
em
p
lo
y
s
s
eq
u
en
tial
atten
tio
n
m
ask
s
to
p
er
f
o
r
m
in
s
tan
ce
-
wis
e
r
ea
s
o
n
in
g
.
T
h
is
ap
p
o
r
ac
h
h
elp
s
p
r
ev
e
n
t o
v
er
f
itti
n
g
in
im
b
alan
ce
d
d
ata,
r
ed
u
c
es n
o
i
s
e
an
d
p
r
o
d
u
ce
s
clea
r
f
ea
tu
r
e
im
p
o
r
tan
ce
s
.
T
ab
le
2
.
Statis
tical
s
ig
n
if
ican
t c
o
m
p
ar
is
o
n
M
e
t
r
i
c
C
o
m
p
a
r
e
d
mo
d
e
l
s
t
-
st
a
t
i
s
t
i
c
p
-
v
a
l
u
e
R
e
s
u
l
t
A
c
c
u
r
a
c
y
T
a
b
N
e
t
*
v
s
X
G
B
o
o
s
t
5
.
8
4
0
.
0
0
2
1
S
i
g
n
i
f
i
c
a
n
t
(
p
<
0
.
0
0
5
)
F1
-
sc
o
r
e
T
a
b
N
e
t
*
v
s
L
i
g
h
t
G
B
M
6
.
4
7
0
.
0
0
1
5
S
i
g
n
i
f
i
c
a
n
t
(
p
<
0
.
0
0
5
)
R
O
C
-
AUC
T
a
b
N
e
t
*
v
sX
G
B
o
o
s
t
4
.
9
3
0
.
0
0
4
3
S
i
g
n
i
f
i
c
a
n
t
(
p
<
0
.
0
0
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
60
1
-
6
1
3
610
7
.
2
.
Dis
cus
s
io
n
M
o
r
e
o
v
e
r
,
t
h
e
s
p
a
r
s
e
m
a
x
a
c
t
iv
a
t
i
o
n
f
u
n
c
t
i
o
n
e
n
a
b
l
e
s
t
h
e
T
ab
N
e
t
*
m
o
d
el
t
o
i
g
n
o
r
e
a
l
l
i
r
r
e
le
v
a
n
t
i
n
p
u
t
v
a
r
i
a
b
l
e
s
w
h
ic
h
i
m
p
r
o
v
e
s
t
h
e
m
o
d
e
l
’
s
i
n
t
e
r
p
r
e
t
a
b
i
li
t
y
a
n
d
e
f
f
i
c
i
e
n
c
y
.
T
h
e
g
l
o
b
a
l
f
e
a
t
u
r
e
i
m
p
o
r
t
a
n
c
e
m
a
p
s
d
e
m
o
n
s
t
r
a
t
e
t
h
e
d
o
m
i
n
a
n
t
r
o
l
es
o
f
e
c
o
n
o
m
i
c
a
n
d
b
e
h
a
v
i
o
r
a
l
v
a
r
i
a
b
l
e
s
s
u
c
h
a
s
t
r
a
n
s
a
c
ti
o
n
a
m
o
u
n
t
a
n
d
m
e
r
c
h
a
n
t
r
i
s
k
s
c
o
r
es
.
T
h
i
s
s
e
r
v
es
a
s
a
n
e
v
i
d
e
n
c
e
o
f
t
h
e
m
o
d
e
l
s
l
o
g
i
c
a
l
r
e
a
s
o
n
i
n
g
p
r
o
c
e
s
s
.
T
h
es
e
f
i
n
d
i
n
g
s
a
r
e
s
u
p
p
o
r
t
e
d
b
y
t
h
e
s
i
g
n
i
f
i
c
a
n
ce
o
f
t
h
e
r
e
s
u
l
t
s
;
t
h
u
s
,
t
h
e
y
m
a
k
e
a
r
e
a
l
i
m
p
r
o
v
e
m
e
n
t
.
C
o
m
p
a
r
e
d
t
o
t
h
e
r
u
n
n
e
r
s
-
u
p
L
i
g
h
t
G
B
M
a
n
d
X
G
B
o
o
s
t
r
es
u
l
ts
t
h
e
r
ec
a
l
l
a
n
d
o
v
e
r
a
l
l
F
-
s
c
o
r
e
s
o
f
t
h
e
r
e
s
u
l
t
s
c
o
n
t
r
i
b
u
t
e
d
b
y
T
a
b
N
e
t*
a
r
e
m
u
c
h
h
i
g
h
e
r
,
a
s
w
ell
a
s
m
ai
n
t
a
i
n
i
n
g
t
h
e
s
a
m
e
l
e
v
e
l
o
f
i
n
t
e
r
p
r
e
t
a
b
i
l
it
y
.
T
h
e
r
e
f
o
r
e
,
t
h
i
s
m
a
k
es
T
a
b
N
et*
a
r
e
l
i
a
b
le
s
y
s
t
e
m
t
h
a
t
m
e
e
ts
c
u
r
r
e
n
t
r
e
g
u
l
a
t
i
o
n
s
i
n
f
i
n
a
n
c
i
al
f
r
a
u
d
d
e
t
e
c
t
i
o
n
.
T
h
e
T
a
b
N
e
t
*
m
o
d
e
l
w
a
s
t
r
ai
n
e
d
o
v
e
r
1
0
e
p
o
c
h
s
a
n
d
s
h
o
w
e
d
s
t
r
o
n
g
p
e
r
f
o
r
m
a
n
c
e
i
n
f
r
a
u
d
d
e
t
e
c
t
i
o
n
.
I
t
ac
h
i
e
v
e
d
t
es
t
ac
c
u
r
ac
y
o
f
9
9
.
6
9
%
i
n
e
p
o
c
h
8
w
i
t
h
a
t
e
s
t
A
UC
o
f
0
.
9
5
6
,
d
e
m
o
n
s
t
r
a
t
i
n
g
i
ts
a
b
i
l
i
t
y
t
o
d
i
s
t
i
n
g
u
i
s
h
b
e
t
w
e
e
n
f
r
a
u
d
u
l
e
n
t
a
n
d
g
e
n
u
i
n
e
t
r
a
n
s
a
c
t
i
o
n
s
.
A
d
d
i
t
io
n
a
l
l
y
,
t
h
e
i
n
c
r
e
a
s
e
i
n
A
UC
v
a
l
u
e
s
f
r
o
m
a
s
t
a
r
ti
n
g
p
o
i
n
t
o
f
0
.
8
2
i
n
e
p
o
c
h
0
t
o
a
v
a
l
u
e
o
f
0
.
9
5
+
i
n
e
p
o
c
h
4
s
h
o
w
s
t
h
at
t
h
is
m
o
d
e
l
is
c
a
p
a
b
l
e
o
f
l
e
a
r
n
i
n
g
e
f
f
e
c
ti
v
e
f
ea
t
u
r
e
r
e
p
r
e
s
e
n
t
at
i
o
n
s
i
r
r
e
s
p
e
ct
iv
e
o
f
t
h
e
h
i
g
h
-
c
l
a
s
s
i
m
b
a
l
a
n
ce
s
.
Mo
r
e
o
v
e
r
,
t
h
e
t
r
a
in
in
g
o
f
t
h
is
m
o
d
el
is
s
ta
b
le
d
u
e
t
o
th
e
S
te
p
L
R
s
c
h
e
d
u
le
r
u
s
e
d
d
u
r
i
n
g
o
p
tim
i
za
t
i
o
n
.
Ad
d
it
io
n
a
ll
y
,
f
e
at
u
r
e
im
p
o
r
ta
n
c
e
a
n
al
y
s
is
s
h
e
d
So
m
e
L
i
g
h
t
o
n
t
h
e
d
e
cisi
o
n
-
m
a
k
i
n
g
p
r
o
c
ess
o
f
th
e
T
a
b
N
et*
m
o
d
el.
I
t
is
c
le
ar
th
at
lo
g
-
s
c
ale
d
t
r
a
n
s
a
cti
o
n
a
m
o
u
n
t,
lo
g
_
a
m
t
,
is
t
h
e
m
o
s
t i
m
p
o
r
t
an
t
f
ea
tu
r
e
i
n
p
r
e
d
ict
in
g
f
r
a
u
d
,
co
n
t
r
i
b
u
ti
n
g
o
v
er
5
3
%
t
o
t
h
e
i
m
p
o
r
ta
n
ce
s
c
o
r
es
.
T
h
is
is
f
o
ll
o
wed
b
y
t
r
a
n
s
_
ca
t
wit
h
2
3
%
im
p
o
r
t
a
n
ce
,
m
e
r
c
h
a
n
t
r
is
k
s
co
r
es
w
it
h
1
0
.
6
%
i
m
p
o
r
t
an
ce
,
an
d
c
it
y
p
o
p
u
l
ati
o
n
wit
h
5
.
9
%
im
p
o
r
ta
n
c
e
s
c
o
r
es.
T
im
e
-
r
el
at
ed
v
a
r
i
ab
les
,
h
o
u
r
a
n
d
ag
e,
co
n
t
r
i
b
u
te
m
o
d
er
atel
y
,
wh
e
r
ea
s
ca
r
d
_
n
u
m
b
e
r
,
m
e
r
c
h
an
t_
id
,
a
n
d
g
e
o
g
r
a
p
h
i
c
l
o
ca
t
io
n
r
e
p
r
ese
n
t
v
a
r
i
ab
les
o
f
less
i
m
p
o
r
ta
n
t
.
A
tte
n
ti
o
n
m
as
k
v
is
u
ali
za
t
io
n
s
a
n
d
g
l
o
b
al
f
ea
tu
r
e
i
m
p
o
r
t
an
ce
p
l
o
ts
s
h
o
w
w
h
i
ch
b
e
h
a
v
i
o
r
al
,
te
m
p
o
r
a
l,
a
n
d
c
o
n
t
ex
t
u
al
f
e
at
u
r
es
m
o
s
t
af
f
e
ct
T
a
b
Net
’
s
p
r
e
d
ict
io
n
s
,
i
m
p
r
o
v
i
n
g
m
o
d
el
t
r
a
n
s
p
a
r
e
n
cy
.
T
h
is
m
e
an
s
t
h
e
m
o
d
e
l
ca
n
r
e
d
u
c
e
n
o
is
e
a
n
d
em
p
h
as
ize
t
h
e
k
ey
b
e
h
a
v
i
o
r
al
a
n
d
co
n
t
e
x
t
u
al
f
ac
t
o
r
s
o
f
f
r
a
u
d
.
T
h
e
T
ab
Net
class
if
ier
'
s
R
O
C
cu
r
v
e
is
p
r
esen
ted
i
n
F
ig
u
r
e
8
,
wh
ich
s
h
o
ws
th
at
th
e
T
ab
Ne
t
ac
h
iev
ed
th
e
h
ig
h
est
ac
cu
r
ac
y
in
p
r
ed
i
ctin
g
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
.
T
h
e
cu
r
v
e
g
r
o
ws
s
teep
ly
to
war
d
th
e
u
p
p
e
r
-
lef
t
an
g
le
o
f
th
e
p
ag
e
,
s
ig
n
if
y
in
g
a
h
ig
h
tr
u
e
p
o
s
itiv
e
r
ate
(
s
en
s
itiv
ity
)
f
r
o
m
th
e
lo
w
f
alse
p
o
s
itiv
e
r
ate
.
An
AUC
o
f
0
.
9
8
4
9
f
o
r
t
h
e
n
ea
r
id
ea
l
class
if
icatio
n
p
er
f
o
r
m
an
ce
s
h
o
win
g
th
e
p
o
ten
tial
o
f
t
h
e
m
o
d
el
to
ea
s
ily
d
is
cr
im
in
ate
am
o
n
g
two
class
es
in
th
e
p
r
o
b
lem
,
t
h
at
i
s
f
r
a
u
d
u
len
t
a
n
d
n
o
n
-
f
r
au
d
u
len
t
tr
a
n
s
ac
tio
n
.
Ou
r
AUC
r
esu
lt
is
co
n
s
is
ten
t
with
th
e
f
ea
tu
r
e
im
p
o
r
tan
ce
an
aly
s
is
,
wh
ich
r
ev
ea
led
tr
a
n
s
ac
tio
n
am
o
u
n
t
(
lo
g
_
am
t)
,
m
er
ch
an
t
r
is
k
s
co
r
e,
an
d
tr
a
n
s
ac
tio
n
ca
teg
o
r
y
(
t
r
an
s
_
ca
t)
as
th
e
to
p
3
m
o
s
t
im
p
o
r
tan
t
f
ea
tu
r
es.
T
ab
Net
ac
h
iev
es
b
etter
p
r
ed
ictiv
e
p
e
r
f
o
r
m
a
n
ce
th
an
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
lik
e
r
an
d
o
m
f
o
r
ests
(
AUC
~0
.
9
5
)
an
d
lo
g
is
tic
r
eg
r
ess
io
n
(
AUC
~0
.
9
2
)
,
wh
ile
allo
win
g
in
te
r
p
r
etab
ilit
y
th
r
o
u
g
h
atten
tio
n
m
ec
h
an
is
m
s
.
T
h
ese
r
esu
lts
d
em
o
n
s
tr
ate
th
at
T
a
b
Net
is
c
ap
ab
le
o
f
s
o
lv
in
g
th
e
class
im
b
alan
ce
p
r
o
b
lem
,
co
n
v
er
g
in
g
with
ex
ce
llen
t
g
e
n
er
aliza
tio
n
ca
p
ab
ilit
y
,
an
d
a
ch
iev
in
g
a
r
o
b
u
s
t
fin
an
cial
f
r
au
d
d
etec
tio
n
with
h
ig
h
p
r
ec
is
io
n
an
d
r
ec
all,
lead
in
g
to
p
r
o
s
p
ec
ts
f
o
r
a
p
p
licatio
n
in
f
in
an
cial
f
r
au
d
d
etec
tio
n
.
T
h
e
g
lo
b
al
f
ea
tu
r
e
im
p
o
r
tan
ce
s
u
m
m
ar
ies
also
p
r
o
v
id
e
u
s
ef
u
l
in
s
ig
h
ts
f
o
r
in
t
er
est
r
ate
-
o
n
ly
l
o
an
s
.
T
r
a
n
s
ac
tio
n
s
with
s
u
ch
as
h
ig
h
lo
g
_
am
t,
h
ig
h
r
is
k
m
er
c
h
an
t,
ce
r
tain
tr
a
n
s
_
ca
t
ar
e
m
o
r
e
lik
ely
to
b
e
r
e
v
iewe
d
/aler
ted
s
o
th
at
y
o
u
ca
n
g
et
ah
ea
d
o
f
f
r
au
d
p
r
o
b
lem
s
an
d
r
ed
u
ce
y
o
u
r
e
x
p
o
s
u
r
e
to
f
in
a
n
c
ial
r
is
k
.
Fig
u
r
e
8
.
T
h
e
R
OC
cu
r
v
e
f
o
r
t
h
e
T
ab
Net*
clas
s
if
ier
Evaluation Warning : The document was created with Spire.PDF for Python.