I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
,
p
p
.
4
7
4
9
~
4
7
6
2
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
6
.
p
p
4
7
4
9
-
4
7
6
2
4749
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Enha
ncing
cre
di
t
ca
rd f
ra
ud
dete
cti
o
n wit
h
sy
n
thet
i
c
mino
rity
o
v
er
-
sa
mpling
t
ec
hnique
-
in
tegra
ted
extreme
learning
ma
chine
I
m
a
n K
a
dh
im
Aj
la
n
1
,
M
o
ha
m
m
ed
I
bra
him
M
a
hd
i
2
,
H
a
y
der
M
ura
d
1
,
F
a
ha
d T
a
ha
A
L
-
Dhief
3
,
Nurhiza
m
Sa
f
ie
4
,
Ya
s
ir
H
us
s
ein Sha
k
ir
5
,
Ali H
a
s
him
Abb
a
s
6
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
C
o
l
l
e
g
e
o
f
E
d
u
c
a
t
i
o
n
f
o
r
P
u
r
e
S
c
i
e
n
c
e
,
W
a
si
t
U
n
i
v
e
r
si
t
y
,
K
u
t
,
I
r
a
q
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
C
o
l
l
e
g
e
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
mat
i
o
n
T
e
c
h
n
o
l
o
g
y
,
W
a
si
t
U
n
i
v
e
r
s
i
t
y
,
K
u
t
,
I
r
a
q
3
F
a
c
u
l
t
y
o
f
I
n
f
o
r
m
a
t
i
o
n
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
U
n
i
v
e
r
si
t
i
K
e
b
a
n
g
s
a
a
n
M
a
l
a
y
si
a
,
B
a
n
g
i
,
M
a
l
a
y
si
a
4
C
e
n
t
e
r
f
o
r
S
o
f
t
w
a
r
e
T
e
c
h
n
o
l
o
g
y
a
n
d
M
a
n
a
g
e
m
e
n
t
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
ma
t
i
o
n
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
U
n
i
v
e
r
s
i
t
i
K
e
b
a
n
g
sa
a
n
M
a
l
a
y
s
i
a
,
B
a
n
g
i
,
S
e
l
a
n
g
o
r
5
C
o
l
l
e
g
e
o
f
G
r
a
d
u
a
t
e
S
t
u
d
i
e
s,
U
n
i
v
e
r
s
i
t
i
Te
n
a
g
a
N
a
s
i
o
n
a
l
,
K
a
j
a
n
g
,
M
a
l
a
y
s
i
a
6
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
Te
c
h
n
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
C
o
l
l
e
g
e
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
I
mam
Ja’
a
f
a
r
A
l
-
S
a
d
i
q
U
n
i
v
e
r
si
t
y
,
Al
-
M
u
t
h
a
n
n
a
,
I
r
a
q
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
1
3
,
2
0
2
5
R
ev
is
ed
Sep
2
5
,
2
0
2
5
Acc
ep
ted
Oct
1
6
,
2
0
2
5
M
a
n
y
wo
r
k
s
in
c
y
b
e
rse
c
u
rit
y
d
e
tec
ti
o
n
su
ffe
r
fro
m
lo
w
a
c
c
u
r
a
c
y
ra
tes
,
p
a
rti
c
u
larly
i
n
re
a
l
-
wo
rld
a
p
p
li
c
a
ti
o
n
s,
wh
e
re
imb
a
lan
c
e
d
d
a
t
a
se
ts
a
n
d
e
v
o
lv
in
g
fra
u
d
stra
teg
ies
p
o
se
si
g
n
ifi
c
a
n
t
h
u
rd
les
.
Th
is
stu
d
y
in
tr
o
d
u
c
e
s
a
n
o
p
ti
m
ize
d
e
x
trem
e
lea
rn
in
g
m
a
c
h
in
e
(EL
M
)
a
lg
o
rit
h
m
to
a
d
d
r
e
ss
th
e
se
c
h
a
ll
e
n
g
e
s
b
y
d
y
n
a
m
ica
ll
y
a
d
j
u
s
ti
n
g
h
i
d
d
e
n
n
o
d
e
s
ra
n
g
i
n
g
fr
o
m
1
0
t
o
1
0
0
with
a
n
in
c
re
m
e
n
t
ste
p
o
f
1
0
a
n
d
in
teg
ra
ti
n
g
two
a
c
ti
v
a
ti
o
n
fu
n
c
ti
o
n
s.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
u
ti
l
ize
s
th
e
s
y
n
th
e
ti
c
m
in
o
r
it
y
o
v
e
r
-
sa
m
p
li
n
g
tec
h
n
iq
u
e
(S
M
OTE)
to
h
a
n
d
le
c
las
s
im
b
a
lan
c
e
e
ffe
c
ti
v
e
ly
a
n
d
i
n
c
o
r
p
o
ra
tes
a
c
o
m
p
re
h
e
n
siv
e
e
v
a
lu
a
ti
o
n
u
sin
g
d
e
sc
rip
ti
v
e
sta
ti
stics
,
v
isu
a
li
z
a
ti
o
n
,
a
n
d
sig
n
ifi
c
a
n
c
e
tes
ti
n
g
.
T
h
e
p
ro
p
o
se
d
EL
M
-
S
M
OTE
m
e
th
o
d
a
c
h
iev
e
s
th
e
h
ig
h
e
st
re
su
lt
s
i
n
c
lu
d
in
g
a
n
a
c
c
u
ra
c
y
o
f
9
9
.
7
1
0
%
,
re
c
a
ll
o
f
8
5
.
8
1
1
%
,
sp
e
c
ifi
c
it
y
o
f
9
9
.
7
4
3
%
,
a
n
d
G
-
m
e
a
n
o
f
9
2
.
0
6
8
%
.
T
h
e
se
o
u
tc
o
m
e
s
re
flec
t
th
e
ro
b
u
stn
e
ss
a
n
d
a
d
a
p
ta
b
il
it
y
o
f
t
h
e
p
ro
p
o
se
d
EL
M
a
lg
o
r
it
h
m
in
d
e
tec
ti
n
g
fra
u
d
u
le
n
t
tran
sa
c
ti
o
n
s.
Th
is
st
u
d
y
e
m
p
h
a
siz
e
s
th
e
imp
o
rta
n
c
e
o
f
a
h
o
li
stic
p
e
rfo
rm
a
n
c
e
a
n
a
ly
sis,
a
d
d
re
ss
in
g
g
a
p
s
in
e
x
isti
n
g
m
e
th
o
d
s
a
n
d
p
r
o
v
i
d
in
g
a
sc
a
lab
le fram
e
wo
rk
fo
r
re
a
l
-
wo
rl
d
fra
u
d
d
e
tec
ti
o
n
a
p
p
li
c
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
C
las
s
im
b
alan
ce
C
r
ed
it c
ar
d
f
r
au
d
d
etec
tio
n
C
y
b
er
s
ec
u
r
ity
E
x
tr
em
e
lear
n
in
g
m
ac
h
in
e
SMOT
E
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
:
Fah
ad
T
ah
a
AL
-
Dh
ie
f
Facu
lty
o
f
I
n
f
o
r
m
atio
n
Scien
c
e
an
d
T
ec
h
n
o
lo
g
y
,
Un
i
v
er
s
iti Ke
b
an
g
s
aa
n
Ma
lay
s
ia
B
an
g
i 4
3
6
0
0
,
Ma
lay
s
ia
E
m
ail: f
ah
ad
@
u
k
m
.
ed
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
I
n
r
ec
e
n
t
y
ea
r
s
,
th
e
e
x
p
o
n
e
n
tial
g
r
o
wth
o
f
d
ig
ital
tr
a
n
s
ac
tio
n
s
h
as
b
r
o
u
g
h
t
co
n
v
en
i
en
ce
an
d
ef
f
icien
cy
to
g
lo
b
al
co
m
m
er
c
e
[
1
]
.
Ho
wev
er
,
th
is
r
is
e
in
d
ig
ital
p
ay
m
en
ts
h
as
also
g
iv
en
r
is
e
to
an
alar
m
in
g
in
cr
ea
s
e
in
f
r
a
u
d
u
le
n
t
ac
tiv
ities
,
p
ar
ticu
lar
ly
in
cr
ed
it
ca
r
d
tr
an
s
ac
tio
n
s
[
2
]
.
C
r
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
b
ec
o
m
es
a
cr
itical
ch
allen
g
e
in
th
e
f
ield
o
f
c
y
b
er
s
ec
u
r
it
y
,
r
eq
u
ir
in
g
in
n
o
v
ativ
e
an
d
ef
f
ec
tiv
e
s
o
lu
tio
n
s
to
p
r
o
tect
f
in
an
cial
s
y
s
tem
s
an
d
co
n
s
u
m
er
t
r
u
s
t
[
3
]
.
Sy
s
te
m
s
o
f
cy
b
er
s
ec
u
r
ity
d
etec
tio
n
h
av
e
d
e
v
elo
p
e
d
s
ig
n
if
ican
tly
o
v
er
r
ec
en
t
y
ea
r
s
,
lev
er
ag
in
g
a
d
v
an
ce
m
en
ts
in
ar
tific
ial
in
tellig
en
ce
(
AI
)
,
m
ac
h
in
e
lear
n
in
g
(
ML
)
,
a
n
d
d
ee
p
lear
n
in
g
(
DL
)
[
4
]
.
Am
o
n
g
th
ese,
e
x
tr
em
e
le
ar
n
in
g
m
ac
h
in
es
(
E
L
M)
h
a
v
e
g
ar
n
er
e
d
atten
tio
n
f
o
r
t
h
eir
r
a
p
id
lear
n
in
g
ca
p
ab
ilit
ies
an
d
m
in
im
al
co
m
p
u
tatio
n
al
r
eq
u
ir
em
en
ts
[
5
]
.
E
L
M,
a
s
in
g
le
-
la
y
er
f
ee
d
f
o
r
war
d
n
e
u
r
al
n
etwo
r
k
(
SLFN
)
,
o
f
f
er
s
an
ad
v
an
tag
e
in
h
an
d
lin
g
lar
g
e
d
atasets
,
m
ak
in
g
it
a
p
r
o
m
is
in
g
ca
n
d
id
ate
f
o
r
r
ea
l
-
tim
e
f
r
au
d
d
etec
tio
n
[
6
]
.
Desp
ite
its
p
o
ten
tial,
th
e
ap
p
licatio
n
o
f
E
L
M
alg
o
r
ith
m
in
c
r
ed
it
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
7
4
9
-
4
7
6
2
4750
ca
r
d
f
r
au
d
d
etec
tio
n
r
em
ain
s
u
n
d
er
ex
p
lo
r
ed
,
esp
ec
ially
in
ad
d
r
ess
in
g
class
im
b
alan
ce
s
an
d
in
co
r
p
o
r
atin
g
r
o
b
u
s
t statis
tical
ev
alu
atio
n
s
[
7
]
.
I
n
g
e
n
er
al,
s
ec
u
r
ity
is
a
cr
itica
l
co
m
p
o
n
en
t
o
f
ev
er
y
s
y
s
tem
an
d
o
r
g
an
izatio
n
[
8
]
–
[
1
0
]
.
I
n
p
ar
ticu
lar
,
cr
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
p
o
s
es
u
n
iq
u
e
c
h
allen
g
es
b
ec
a
u
s
e
o
f
th
e
c
o
m
p
lex
ity
an
d
ev
o
lv
i
n
g
tactics
in
v
o
lv
ed
in
f
r
au
d
u
len
t
ac
tiv
ities
[
1
1
]
.
I
n
th
e
s
y
s
tem
s
o
f
cr
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
,
t
h
e
d
atasets
a
r
e
ty
p
ically
h
ig
h
ly
im
b
alan
ce
d
,
with
leg
itima
te
t
r
an
s
ac
tio
n
s
v
astl
y
o
u
tn
u
m
b
er
in
g
f
r
a
u
d
u
len
t
o
n
es
[
1
2
]
.
T
h
i
s
im
b
alan
ce
o
f
ten
s
k
ews
th
e
p
er
f
o
r
m
an
ce
o
f
m
o
d
els,
wh
er
e
it
m
ay
o
b
tain
h
ig
h
o
v
er
all
ac
cu
r
ac
y
b
u
t
p
o
o
r
s
en
s
itiv
ity
an
d
s
p
ec
if
icity
f
o
r
th
e
m
in
o
r
ity
class
(
i.e
.
,
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
)
.
As
f
r
au
d
u
len
t
ac
tiv
ities
co
n
tin
u
e
to
g
r
o
w
i
n
s
o
p
h
is
ticatio
n
,
co
n
v
en
tio
n
al
d
etec
tio
n
s
y
s
tem
s
o
f
ten
f
in
d
it d
if
f
icu
lt
to
k
ee
p
p
ac
e,
h
ig
h
lig
h
tin
g
th
e
im
p
o
r
tan
c
e
o
f
ad
o
p
tin
g
m
o
r
e
a
d
ap
tiv
e
a
n
d
v
er
s
atile
ap
p
r
o
ac
h
es
[
1
3
]
.
Pre
v
io
u
s
r
esear
ch
h
as
e
x
p
lo
r
e
d
a
wid
e
r
a
n
g
e
o
f
ML
tech
n
iq
u
es
f
o
r
cr
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
,
in
clu
d
in
g
l
o
g
is
tic
r
eg
r
ess
i
o
n
(
L
R
)
[
1
4
]
,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
(
SVMs)
[
1
5
]
,
r
an
d
o
m
f
o
r
ests
(
R
Fs
)
[
1
6
]
,
an
d
n
eu
r
al
n
etwo
r
k
s
(
NNs
)
[
1
7
]
.
Alth
o
u
g
h
t
h
e
s
e
alg
o
r
ith
m
s
h
av
e
p
r
o
d
u
ce
d
en
c
o
u
r
a
g
in
g
r
esu
lts
,
m
an
y
s
till
d
ep
e
n
d
o
n
ar
c
h
itectu
r
es
with
a
f
ix
ed
n
u
m
b
er
o
f
n
eu
r
o
n
s
,
wh
ic
h
r
estricts
th
eir
ca
p
ac
ity
to
h
an
d
le
d
iv
er
s
e
an
d
co
m
p
lex
d
ata
p
atter
n
s
.
I
n
ad
d
itio
n
,
s
e
v
er
al
cy
b
er
s
ec
u
r
ity
d
etec
tio
n
s
tu
d
ies
h
av
e
r
ep
o
r
t
ed
s
u
b
o
p
tim
al
ac
cu
r
ac
y
,
p
a
r
ticu
lar
ly
wh
en
ap
p
lied
to
r
ea
l
-
wo
r
ld
d
atasets
th
at
co
n
tain
n
o
is
e
an
d
ir
r
e
g
u
lar
itie
s
.
An
o
th
er
s
ig
n
if
ica
n
t
g
ap
id
en
t
if
ied
in
th
e
e
x
is
tin
g
liter
atu
r
e
is
th
e
ab
s
en
ce
o
f
th
o
r
o
u
g
h
s
tatis
tica
l
ev
alu
atio
n
s
.
Ma
n
y
s
tu
d
ies ass
ess
m
o
d
el
p
er
f
o
r
m
an
ce
u
s
in
g
o
n
ly
a
lim
ited
r
a
n
g
e
o
f
m
etr
ic
s
,
ty
p
ically
ac
cu
r
ac
y
o
r
p
r
ec
is
io
n
,
wh
ile
o
v
er
l
o
o
k
i
n
g
cr
u
cial
m
ea
s
u
r
es
s
u
ch
as
s
p
ec
if
icity
,
G
-
m
ea
n
an
d
F
-
m
ea
s
u
r
e.
T
h
is
n
ar
r
o
w
ev
alu
atio
n
a
p
p
r
o
a
ch
h
am
p
er
s
a
co
m
p
r
eh
en
s
iv
e
u
n
d
er
s
tan
d
in
g
o
f
m
o
d
el
ef
f
ec
tiv
en
ess
,
esp
ec
ially
in
th
e
co
n
tex
t
o
f
im
b
alan
ce
d
d
atasets
wh
er
e
ac
cu
r
ac
y
alo
n
e
f
ails
to
ca
p
tu
r
e
tr
u
e
p
er
f
o
r
m
a
n
ce
.
M
o
r
eo
v
e
r
,
th
e
lim
ited
u
s
e
o
f
d
escr
ip
tiv
e
s
tatis
tics
,
v
is
u
aliza
tio
n
,
an
d
s
ig
n
i
f
ican
c
e
test
in
g
f
u
r
th
er
r
ed
u
ce
s
th
e
in
ter
p
r
etab
ilit
y
an
d
r
eliab
ilit
y
o
f
th
e
r
ep
o
r
ted
r
esu
lts
.
I
n
co
n
tr
ast,
th
ese
tech
n
iq
u
e
s
ar
e
f
u
n
d
am
e
n
tal
to
v
alid
atin
g
th
e
r
o
b
u
s
tn
ess
o
f
ML
m
o
d
els.
Descr
ip
tiv
e
s
tati
s
tics
o
f
f
er
v
alu
ab
le
in
s
ig
h
ts
i
n
to
d
ata
d
is
tr
ib
u
tio
n
an
d
v
ar
i
ab
ilit
y
,
v
is
u
aliza
tio
n
h
elp
s
u
n
co
v
er
u
n
d
er
ly
in
g
tr
en
d
s
an
d
an
o
m
alies,
an
d
s
i
g
n
if
ican
ce
test
in
g
d
eter
m
in
e
s
wh
eth
er
o
b
s
er
v
e
d
d
if
f
er
en
ce
s
in
p
er
f
o
r
m
a
n
ce
ar
e
s
tatis
tically
m
ea
n
in
g
f
u
l
r
at
h
er
th
an
d
u
e
to
r
an
d
o
m
v
a
r
iatio
n
.
Desp
ite
th
eir
im
p
o
r
tan
ce
,
s
u
ch
m
eth
o
d
s
a
r
e
f
r
eq
u
en
tly
n
eg
lecte
d
in
cr
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
r
esear
ch
.
Ov
er
all
,
ex
is
tin
g
s
tu
d
ies
in
th
is
d
o
m
ain
co
n
tin
u
e
t
o
f
ac
e
s
ev
er
al
lim
itatio
n
s
,
s
u
m
m
ar
ized
as
i)
m
a
n
y
m
o
d
els
r
ely
o
n
a
f
ix
ed
n
u
m
b
er
o
f
n
e
u
r
o
n
n
o
d
e
s
,
r
ed
u
cin
g
a
d
ap
tab
ilit
y
to
d
i
v
er
s
e
d
atasets
;
ii)
cy
b
er
s
ec
u
r
ity
d
etec
tio
n
s
y
s
tem
s
o
f
ten
ex
h
i
b
it
lo
w
ac
cu
r
ac
y
,
p
ar
ticu
lar
ly
wh
e
n
test
ed
o
n
r
ea
l
-
wo
r
l
d
d
ata
;
iii)
d
e
s
cr
ip
tiv
e
s
tatis
t
ics,
v
is
u
aliza
tio
n
,
an
d
s
ig
n
i
f
ican
c
e
test
in
g
ar
e
co
m
m
o
n
ly
e
x
cl
u
d
ed
f
r
o
m
e
v
alu
atio
n
f
r
am
e
wo
r
k
s
;
an
d
iv
)
m
o
s
t
cy
b
er
s
ec
u
r
ity
d
etec
tio
n
s
y
s
tem
s
ar
e
ass
e
s
s
ed
u
s
in
g
a
l
im
it
ed
s
et
o
f
ev
alu
atio
n
m
etr
ics,
ig
n
o
r
in
g
a
h
o
lis
tic
p
er
f
o
r
m
an
ce
an
al
y
s
is
.
T
h
e
cu
r
r
en
t
s
tu
d
y
aim
s
to
ad
d
r
ess
th
ese
lim
itatio
n
s
b
y
e
x
p
lo
r
in
g
t
h
e
ap
p
licatio
n
o
f
E
L
M
f
o
r
cr
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
u
s
in
g
a
d
y
n
am
ic
ap
p
r
o
ac
h
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
in
c
o
r
p
o
r
ates
v
ar
y
in
g
n
u
m
b
er
s
o
f
h
id
d
en
n
o
d
es
an
d
e
v
alu
ates
t
wo
ac
tiv
atio
n
f
u
n
ctio
n
s
t
o
id
e
n
tify
th
e
o
p
tim
al
co
n
f
ig
u
r
atio
n
.
Ad
d
itio
n
ally
,
th
is
s
tu
d
y
em
p
h
asizes
a
co
m
p
r
e
h
e
n
s
iv
e
ev
alu
atio
n
f
r
a
m
ewo
r
k
,
i
n
clu
d
in
g
ac
c
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
s
p
ec
if
icity
,
F
-
m
ea
s
u
r
e,
G
-
m
ea
n
,
d
escr
ip
ti
v
e
s
tatis
t
ics,
v
is
u
aliza
t
io
n
,
an
d
s
ig
n
if
ican
ce
test
in
g
.
I
n
ad
d
itio
n
,
to
o
v
er
c
o
m
e
th
e
p
r
ev
io
u
s
ly
m
en
tio
n
ed
lim
itatio
n
s
,
th
is
p
ap
er
s
u
m
m
ar
izes
th
e
f
o
llo
win
g
k
ey
co
n
tr
i
b
u
tio
n
s
:
i)
d
y
n
am
ic
h
id
d
en
n
o
d
e
co
n
f
ig
u
r
atio
n
s
: u
n
lik
e
tr
ad
itio
n
al
ap
p
r
o
ac
h
es,
th
is
s
tu
d
y
ev
alu
ates E
L
M
with
v
ar
y
in
g
n
u
m
b
er
s
o
f
h
id
d
en
n
o
d
es,
r
a
n
g
in
g
f
r
o
m
1
0
to
1
0
0
,
t
o
id
en
tif
y
th
e
o
p
tim
al
co
n
f
ig
u
r
atio
n
f
o
r
c
r
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
;
ii)
co
m
p
ar
is
o
n
o
f
ac
tiv
atio
n
f
u
n
ctio
n
s
:
two
wid
ely
u
s
ed
ac
tiv
atio
n
f
u
n
ctio
n
s
(
i.e
.
,
s
ig
m
o
id
an
d
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U)
)
ar
e
ass
es
s
ed
to
d
eter
m
in
e
th
eir
im
p
ac
t
o
n
m
o
d
el
p
er
f
o
r
m
an
ce
;
iii)
co
m
p
r
eh
en
s
iv
e
s
tatis
tical
ev
alu
atio
n
:
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
i
n
clu
d
es
a
th
o
r
o
u
g
h
e
v
alu
atio
n
f
r
am
ew
o
r
k
,
s
u
ch
as
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
s
p
ec
if
icity
,
F
-
m
ea
s
u
r
e,
G
-
m
ea
n
,
an
d
c
o
n
f
u
s
io
n
m
atr
ices.
Descr
ip
tiv
e
s
tatis
tics
,
v
is
u
aliza
tio
n
,
an
d
s
ig
n
if
ica
n
ce
test
in
g
ar
e
u
t
ilized
to
e
n
s
u
r
e
a
r
o
b
u
s
t
an
al
y
s
is
;
iv
)
ad
d
r
ess
in
g
class
im
b
alan
ce
:
th
e
s
y
n
th
etic
m
in
o
r
ity
o
v
er
-
s
am
p
li
n
g
tech
n
iq
u
e
(
SMOT
E
)
is
em
p
l
o
y
ed
t
o
b
alan
ce
t
h
e
d
ataset,
en
s
u
r
i
n
g
eq
u
itab
le
lea
r
n
in
g
f
o
r
b
o
t
h
m
ajo
r
ity
an
d
m
in
o
r
it
y
class
es
;
an
d
v
)
h
o
lis
tic
p
er
f
o
r
m
an
ce
in
s
ig
h
ts
:
b
y
in
teg
r
ati
n
g
m
u
ltip
le
m
etr
ics
an
d
s
tatis
tical
tech
n
iq
u
es,
th
is
s
tu
d
y
p
r
o
v
id
es
a
n
u
an
ce
d
u
n
d
er
s
tan
d
in
g
o
f
E
L
M’
s
ef
f
ec
tiv
en
ess
in
cr
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
.
T
h
e
r
est
o
f
t
h
e
c
u
r
r
en
t
p
a
p
er
i
s
s
tr
u
ctu
r
ed
as
f
o
llo
ws:
s
ec
tio
n
2
p
r
esen
ts
t
h
e
r
elate
d
wo
r
k
s
p
r
esen
ted
in
d
etec
tin
g
cr
ed
it
ca
r
d
f
r
au
d
.
Sectio
n
3
d
eliv
er
s
an
d
ex
p
lain
s
th
e
p
r
o
p
o
s
ed
m
eth
o
d
i
n
ter
m
s
o
f
th
e
d
atab
ase
an
d
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
.
Sectio
n
4
g
iv
es
th
e
ex
p
er
im
en
tal
s
etu
p
an
d
r
esu
lts
an
aly
s
is
.
Sectio
n
5
p
r
o
v
id
es th
e
d
is
cu
s
s
io
n
o
f
th
e
ex
p
er
im
en
tal
r
esu
lts
.
Fin
ally
,
s
ec
tio
n
6
p
r
esen
ts
th
e
co
n
clu
s
io
n
o
f
th
is
p
ap
er
.
2.
RE
L
AT
E
D
WO
RK
R
ec
e
n
tl
y
,
r
ese
ar
c
h
e
r
s
a
r
e
h
ig
h
l
y
i
n
t
er
este
d
in
f
r
a
u
d
d
et
ec
t
io
n
d
u
e
t
o
t
h
e
i
n
c
r
e
asi
n
g
p
r
e
v
al
e
n
ce
o
f
f
r
a
u
d
u
l
en
t
ac
ti
v
ities
,
p
ar
tic
u
l
ar
ly
in
t
h
e
f
in
a
n
ci
al
s
ec
t
o
r
[
1
8
]
.
As
t
h
e
v
o
l
u
m
e
o
f
f
i
n
a
n
cial
t
r
a
n
s
a
cti
o
n
s
c
o
n
ti
n
u
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
cred
it c
a
r
d
fr
a
u
d
d
etec
tio
n
w
ith
s
yn
th
etic
min
o
r
ity
o
ve
r
…
(
I
ma
n
K
a
d
h
im
A
jla
n
)
4751
to
g
r
o
w,
s
o
d
o
es
t
h
e
c
o
m
p
l
ex
ity
o
f
d
et
ec
ti
n
g
f
r
a
u
d
u
le
n
t
b
e
h
a
v
i
o
r
.
ML
t
ec
h
n
i
q
u
es
h
a
v
e
g
ai
n
ed
c
o
n
s
i
d
e
r
a
b
l
e
atte
n
t
io
n
f
o
r
t
h
ei
r
ab
ilit
y
t
o
i
d
en
t
if
y
p
att
er
n
s
i
n
l
a
r
g
e
d
a
tase
t
s
,
m
a
k
i
n
g
t
h
em
e
f
f
ec
ti
v
e
to
o
ls
f
o
r
f
r
a
u
d
d
et
ec
t
io
n
.
R
esea
r
c
h
e
r
s
h
av
e
e
x
p
lo
r
e
d
a
v
a
r
i
et
y
o
f
a
lg
o
r
it
h
m
s
,
r
a
n
g
in
g
f
r
o
m
t
r
a
d
it
io
n
al
m
e
th
o
d
s
li
k
e
d
ec
is
io
n
t
r
ee
(
DT
)
an
d
LR
to
m
o
r
e
a
d
v
a
n
ce
d
a
p
p
r
o
a
c
h
es
s
u
c
h
as
DL
a
n
d
e
n
s
e
m
b
le
m
e
th
o
d
s
.
T
r
i
v
e
d
i
et
a
l
.
[
1
9
]
p
r
ese
n
te
d
an
ef
f
e
cti
v
e
c
r
e
d
it
ca
r
d
f
r
au
d
d
et
ec
t
io
n
m
e
ch
a
n
is
m
t
h
at
in
c
o
r
p
o
r
a
tes
a
f
ee
d
b
a
ck
s
y
s
t
em
b
as
ed
o
n
M
L
m
et
h
o
d
o
lo
g
i
es.
T
h
e
f
ee
d
b
a
ck
ap
p
r
o
ac
h
e
n
h
a
n
c
es
t
h
e
c
lass
i
f
i
er
'
s
d
ete
cti
o
n
r
at
e
w
h
il
e
e
n
s
u
r
in
g
co
s
t
-
ef
f
e
cti
v
e
n
ess
.
T
h
e
s
t
u
d
y
ev
al
u
at
es
th
e
p
e
r
f
o
r
m
a
n
c
e
o
f
s
ev
er
al
ML
m
o
d
e
ls
,
i
n
cl
u
d
i
n
g
R
F,
DT
,
ar
ti
f
ic
ial
n
e
u
r
al
n
etw
o
r
k
s
(
ANN
)
,
SV
M,
n
aï
v
e
B
ay
es
(
NB
)
,
L
R
,
an
d
g
r
a
d
ie
n
t
b
o
o
s
t
in
g
cl
ass
i
f
ie
r
(
GB
M
)
,
a
p
p
li
ed
t
o
a
s
lig
h
t
ly
i
m
b
al
a
n
ce
d
c
r
e
d
i
t
ca
r
d
f
r
a
u
d
d
atas
et
.
T
h
e
d
at
aset
c
o
n
s
is
ts
o
f
t
r
a
n
s
a
cti
o
n
d
at
a
f
r
o
m
E
u
r
o
p
e
an
ac
c
o
u
n
t
h
o
l
d
e
r
s
,
w
it
h
2
8
4
,
8
0
7
tr
a
n
s
ac
ti
o
n
s
.
B
o
t
h
r
aw
a
n
d
p
r
e
-
p
r
o
ce
s
s
ed
d
a
ta
we
r
e
u
s
ed
in
t
h
e
e
x
p
e
r
im
e
n
ts
.
T
h
e
r
es
u
l
ts
s
h
o
we
d
t
h
a
t
R
F
a
c
h
ie
v
e
d
a
n
a
cc
u
r
ac
y
o
f
9
4
.
9
9
%
,
NB
9
1
.
8
8
%,
L
R
9
0
.
4
4
%,
SVM
9
3
.
9
6
%,
k
-
n
ea
r
es
t
n
ei
g
h
b
o
r
s
(
K
NN
)
9
4
.
9
9
%
,
DT
9
0
.
9
9
%
,
an
d
GB
M
9
4
.
0
0
%
.
H
o
w
e
v
e
r
,
th
is
w
o
r
k
h
as
b
ee
n
i
g
n
o
r
e
d
th
e
s
t
atis
t
ic
al
a
n
al
y
s
is
o
f
t
h
e
p
r
ese
n
t
e
d
m
e
th
o
d
s
.
A
n
o
t
h
e
r
w
o
r
k
h
as
b
ee
n
f
o
c
u
s
ed
o
n
u
t
ili
zi
n
g
m
u
lti
p
l
e
ML
a
lg
o
r
i
th
m
s
,
s
u
c
h
as
SVM
alg
o
r
it
h
m
,
KNN
al
g
o
r
it
h
m
,
an
d
A
NN
al
g
o
r
it
h
m
,
t
o
p
r
e
d
ic
t
t
h
e
o
cc
u
r
r
en
ce
o
f
f
r
a
u
d
[
2
0
]
.
T
h
e
s
tu
d
y
co
m
p
a
r
es
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
ese
s
u
p
e
r
v
is
e
d
ML
t
ec
h
n
i
q
u
es
wi
th
DL
ap
p
r
o
ac
h
es
to
d
is
ti
n
g
u
is
h
b
e
tw
ee
n
f
r
a
u
d
u
le
n
t
a
n
d
n
o
n
-
f
r
au
d
u
l
en
t
t
r
a
n
s
ac
t
io
n
s
.
T
h
e
r
es
u
l
ts
s
h
o
w
th
at
th
e
K
NN
a
lg
o
r
it
h
m
ac
h
ie
v
ed
a
n
a
cc
u
r
ac
y
o
f
0
.
9
9
8
2
,
f
o
ll
o
we
d
b
y
t
h
e
AN
N
wit
h
0
.
9
9
9
2
,
a
n
d
t
h
e
SV
M
wit
h
0
.
9
3
4
9
.
Al
th
o
u
g
h
t
h
is
w
o
r
k
ac
h
ie
v
e
d
h
i
g
h
r
esu
lts
i
n
th
e
f
r
a
u
d
d
et
ec
ti
o
n
,
t
h
e
s
u
g
g
este
d
w
o
r
k
h
as
n
o
t
b
ee
n
e
v
al
u
at
ed
i
n
t
er
m
s
o
f
o
t
h
e
r
im
p
o
r
ta
n
t
m
e
as
u
r
es
s
u
c
h
as
s
p
e
ci
f
ici
ty
,
F
-
m
e
as
u
r
e,
a
n
d
G
-
m
ea
n
.
Mo
r
eo
v
e
r
,
th
er
e
was
n
o
s
tatis
t
ic
al
e
v
al
u
a
ti
o
n
.
Alar
f
aj
et
a
l.
[
2
1
]
h
as
b
ee
n
ai
m
ed
to
ad
d
r
ess
ch
allen
g
es
i
n
cr
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
,
in
clu
d
in
g
th
e
av
ailab
ilit
y
o
f
p
u
b
lic
d
ata,
h
i
g
h
class
im
b
alan
ce
,
c
h
an
g
es
i
n
f
r
au
d
p
atter
n
s
,
an
d
h
ig
h
f
al
s
e
alar
m
r
ates.
T
h
e
ex
is
tin
g
liter
atu
r
e
p
r
esen
ts
s
e
v
er
al
ML
-
b
ased
ap
p
r
o
ac
h
es
f
o
r
f
r
au
d
d
etec
tio
n
,
s
u
ch
as
E
L
M,
DT
,
R
F,
SVM,
L
R
,
an
d
ex
tr
em
e
g
r
a
d
ien
t
b
o
o
s
tin
g
(
XGBo
o
s
t
)
.
Ho
wev
er
,
d
u
e
to
lo
w
ac
cu
r
ac
y
,
th
er
e
r
em
ain
s
a
n
ee
d
to
im
p
lem
en
t
s
tate
-
of
-
t
h
e
-
ar
t
DL
alg
o
r
ith
m
s
to
r
ed
u
ce
f
r
a
u
d
lo
s
s
es.
T
h
e
s
tu
d
y
f
o
c
u
s
es
o
n
lev
er
a
g
in
g
r
ec
en
t
d
ev
elo
p
m
e
n
ts
in
DL
to
im
p
r
o
v
e
f
r
au
d
d
etec
tio
n
p
e
r
f
o
r
m
an
ce
.
A
c
o
m
p
ar
ativ
e
an
al
y
s
is
b
e
twee
n
ML
a
n
d
DL
alg
o
r
ith
m
s
was
co
n
d
u
cted
to
id
en
tify
th
e
m
o
s
t
ef
f
icien
t
m
eth
o
d
s
.
T
h
e
em
p
i
r
ical
an
al
y
s
is
was
ca
r
r
ied
o
u
t
u
s
in
g
th
e
E
u
r
o
p
ea
n
ca
r
d
b
en
c
h
m
ar
k
d
ataset
f
o
r
f
r
au
d
d
etec
t
io
n
.
I
n
itially
,
ML
alg
o
r
ith
m
s
wer
e
ap
p
lied
to
th
e
d
ataset,
r
esu
ltin
g
in
s
o
m
e
im
p
r
o
v
em
en
t
in
f
r
au
d
d
etec
tio
n
a
cc
u
r
ac
y
.
Su
b
s
eq
u
en
tl
y
,
th
r
ee
ar
ch
itectu
r
es
b
ased
on
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN)
wer
e
u
s
ed
to
f
u
r
t
h
er
en
h
an
ce
d
etec
tio
n
p
er
f
o
r
m
an
ce
.
T
h
e
ad
d
itio
n
o
f
m
o
r
e
lay
er
s
led
to
in
cr
ea
s
e
d
ac
cu
r
ac
y
.
T
h
e
s
tu
d
y
also
ex
p
lo
r
ed
v
ar
iatio
n
s
in
th
e
n
u
m
b
er
o
f
h
i
d
d
en
la
y
er
s
,
ep
o
ch
s
,
an
d
th
e
latest
m
o
d
els
to
o
p
tim
ize
r
esu
lts
.
T
h
e
ev
al
u
atio
n
d
e
m
o
n
s
tr
ated
th
at
th
e
p
r
o
p
o
s
ed
DL
m
o
d
el
ac
h
iev
ed
im
p
r
o
v
ed
r
esu
lts
,
with
an
ac
cu
r
ac
y
o
f
9
9
.
9
%,
F1
-
s
co
r
e
o
f
8
5
.
7
1
%,
p
r
ec
is
io
n
o
f
9
3
%,
an
d
ar
ea
u
n
d
e
r
th
e
cu
r
v
e
(
AUC
)
o
f
9
8
%,
o
u
tp
er
f
o
r
m
i
n
g
ex
is
tin
g
ML
an
d
DL
alg
o
r
ith
m
s
f
o
r
c
r
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
.
Ho
wev
er
,
th
e
au
th
o
r
s
h
av
e
n
o
t
ev
alu
ated
th
e
p
r
esen
ted
al
g
o
r
ith
m
s
in
ter
m
s
o
f
G
-
m
ea
n
,
F
-
m
ea
s
u
r
e,
an
d
s
p
ec
if
icity
.
B
esid
es,
th
ey
h
av
e
n
o
t a
s
s
ess
ed
th
eir
m
eth
o
d
s
s
tatis
tically
.
I
leb
er
i
et
a
l.
[
2
2
]
h
as
b
ee
n
p
r
o
p
o
s
ed
a
ML
-
b
ased
cr
e
d
it
ca
r
d
f
r
au
d
d
etec
tio
n
en
g
in
e
th
at
u
tili
ze
s
a
g
en
etic
alg
o
r
ith
m
(
GA)
f
o
r
f
e
atu
r
e
s
elec
tio
n
.
On
ce
th
e
o
p
ti
m
al
f
ea
tu
r
es
ar
e
s
elec
ted
,
th
e
p
r
o
p
o
s
ed
d
etec
tio
n
en
g
in
e
em
p
lo
y
s
s
ev
er
al
ML
c
lass
if
ier
s
,
in
clu
d
in
g
DT
,
R
F,
L
R
,
ANN,
an
d
NB
.
T
o
v
alid
a
te
its
p
er
f
o
r
m
an
ce
,
th
e
en
g
i
n
e
was
ev
al
u
ated
u
s
in
g
a
d
ataset
g
en
e
r
ated
f
r
o
m
E
u
r
o
p
ea
n
ca
r
d
h
o
ld
er
s
.
T
h
e
r
esu
l
ts
s
h
o
wed
th
at
t
h
e
GA
-
RF
clas
s
if
ier
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
9
9
.
9
8
%,
f
o
llo
we
d
b
y
GA
-
DT
with
9
9
.
9
2
%,
GA
-
LR
with
9
9
.
9
1
%,
an
d
GA
-
NB
with
9
9
.
4
4
%.
A
lim
itatio
n
o
f
th
is
wo
r
k
is
th
at
it
h
as
n
o
t
b
ee
n
ev
alu
ated
i
n
ter
m
s
o
f
s
tatis
tica
l
an
aly
s
is
.
W
h
ile
th
e
s
tu
d
y
f
o
c
u
s
es
o
n
o
p
tim
izin
g
ML
class
if
ier
s
an
d
f
ea
tu
r
e
s
elec
tio
n
,
it
d
o
es
n
o
t
p
r
o
v
id
e
a
n
in
-
d
ep
th
s
tatis
tical
ev
alu
atio
n
o
f
t
h
e
r
esu
lts
.
T
h
is
o
m
is
s
io
n
lim
its
th
e
u
n
d
er
s
tan
d
i
n
g
o
f
t
h
e
m
o
d
el's
p
er
f
o
r
m
an
ce
in
ter
m
s
o
f
s
tatis
t
ical
s
ig
n
if
ican
ce
,
wh
ich
co
u
ld
o
f
f
er
m
o
r
e
r
o
b
u
s
t in
s
ig
h
ts
in
to
th
e
r
eliab
ilit
y
an
d
g
en
er
aliza
b
ilit
y
o
f
th
e
p
r
o
p
o
s
ed
f
r
au
d
d
etec
tio
n
en
g
in
e
.
An
o
th
er
s
tu
d
y
h
as
b
ee
n
p
r
esen
ted
an
ef
f
ec
tiv
e
a
p
p
r
o
ac
h
f
o
r
cr
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
u
s
in
g
a
NN
en
s
em
b
le
class
if
ier
p
air
ed
with
a
h
y
b
r
i
d
d
ata
r
esam
p
lin
g
te
ch
n
iq
u
e
[
2
3
]
.
T
h
e
en
s
em
b
le
c
lass
if
ier
is
b
u
ilt
b
y
in
teg
r
atin
g
a
lo
n
g
s
h
o
r
t
-
te
r
m
m
em
o
r
y
(
L
STM
)
NN
as
th
e
b
ase
m
o
d
el
with
in
th
e
ad
ap
tiv
e
b
o
o
s
tin
g
(
Ad
aBo
o
s
t)
f
r
am
ewo
r
k
.
T
h
e
h
y
b
r
id
r
esam
p
li
n
g
m
eth
o
d
is
im
p
lem
en
ted
u
s
in
g
th
e
SMOT
E
co
m
b
in
e
d
with
th
e
ed
ited
n
ea
r
est
n
eig
h
b
o
r
(
E
N
N)
m
eth
o
d
.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
test
ed
o
n
p
u
b
licly
av
ailab
le
r
ea
l
-
wo
r
ld
cr
ed
it
ca
r
d
tr
an
s
ac
tio
n
d
atasets
,
an
d
its
p
e
r
f
o
r
m
an
ce
is
co
m
p
ar
ed
with
s
ev
er
al
al
g
o
r
ith
m
s
,
in
clu
d
in
g
SVM,
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
(
ML
P),
DT
,
tr
ad
itio
n
al
Ad
aBo
o
s
t,
an
d
L
STM
.
T
h
e
ex
p
er
im
en
tal
f
in
d
in
g
s
in
d
icate
th
at
class
if
ier
s
tr
ain
ed
o
n
r
esam
p
l
ed
d
ata
ac
h
iev
e
d
b
etter
r
esu
lt
s
,
with
th
e
L
STM
en
s
em
b
le
o
u
tp
er
f
o
r
m
in
g
o
th
er
m
o
d
els,
attain
in
g
a
s
en
s
itiv
ity
o
f
0
.
9
9
6
an
d
s
p
ec
if
icity
o
f
0
.
9
9
8
.
Ho
we
v
er
,
th
e
lim
itatio
n
o
f
th
is
wo
r
k
is
th
at
it
h
as
b
ee
n
ev
alu
ated
i
n
ter
m
s
o
f
s
en
s
itiv
ity
,
s
p
ec
if
icity
,
an
d
AUC
o
n
ly
,
with
o
u
t
co
n
s
id
er
in
g
o
th
er
im
p
o
r
tan
t
p
er
f
o
r
m
an
ce
m
etr
ics o
r
a
m
o
r
e
co
m
p
r
e
h
en
s
iv
e
s
tatis
tical
an
aly
s
is
.
Ag
h
war
et
a
l.
[
2
4
]
h
as
b
ee
n
i
n
v
esti
g
ated
th
e
p
er
f
o
r
m
a
n
ce
o
f
f
iv
e
ML
alg
o
r
ith
m
s
,
tr
ain
ed
b
o
th
with
an
d
with
o
u
t
th
e
SMOT
E
,
to
ass
es
s
th
eir
ef
f
ec
tiv
en
ess
in
c
r
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
.
T
h
e
alg
o
r
ith
m
s
test
ed
in
clu
d
e
R
F,
KNN,
NB
,
SVM,
an
d
L
R
.
T
h
e
m
eth
o
d
o
lo
g
y
was
im
p
lem
e
n
ted
a
n
d
ev
alu
ated
th
r
o
u
g
h
an
ap
p
licatio
n
p
r
o
g
r
a
m
m
in
g
i
n
ter
f
ac
e
(
API
)
u
s
in
g
Flas
k
an
d
Str
ea
m
lit
in
Py
th
o
n
.
Prio
r
to
a
p
p
l
y
in
g
SMOT
E
,
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
7
4
9
-
4
7
6
2
4752
R
F
clas
s
if
ier
d
em
o
n
s
tr
ated
th
e
h
ig
h
est
p
er
f
o
r
m
an
ce
with
an
ac
cu
r
ac
y
o
f
0
.
9
8
0
2
,
wh
ile
L
R
,
KNN,
NB
,
an
d
SVM
alg
o
r
ith
m
s
h
av
e
b
ee
n
a
ch
iev
ed
ac
cu
r
ac
ies
o
f
0
.
9
2
1
9
,
0
.
9
4
3
5
,
0
.
9
5
0
8
,
an
d
0
.
9
0
0
8
,
r
esp
ec
tiv
ely
.
Af
ter
th
e
ap
p
licatio
n
o
f
SMOT
E
,
th
e
R
F
class
if
ier
ac
h
iev
ed
a
h
ig
h
er
ac
c
u
r
ac
y
o
f
0
.
9
9
1
9
,
wh
ile
L
R
,
KNN,
NB
,
an
d
SVM
alg
o
r
ith
m
s
r
ea
ch
ed
ac
cu
r
ac
ies
o
f
0
.
9
8
0
5
,
0
.
9
2
1
0
,
0
.
9
1
2
5
,
a
n
d
0
.
8
1
4
5
,
r
esp
ec
tiv
ely
.
T
h
ese
r
esu
lts
em
p
h
asize
th
e
ad
v
a
n
tag
e
o
f
c
o
m
b
in
in
g
RF
with
SMOT
E
t
o
im
p
r
o
v
e
p
r
ed
ictio
n
ac
c
u
r
ac
y
in
f
r
a
u
d
d
etec
tio
n
task
s
.
Nev
er
th
eless
,
th
e
lim
itatio
n
o
f
th
is
wo
r
k
is
th
at
th
e
ev
alu
atio
n
was
s
o
lely
b
ased
o
n
a
cc
u
r
ac
y
,
p
r
ec
is
io
n
,
an
d
r
ec
all,
with
o
u
t
co
n
s
id
er
in
g
o
th
e
r
im
p
o
r
tan
t
p
er
f
o
r
m
an
c
e
m
etr
ics
o
r
co
n
d
u
ctin
g
a
m
o
r
e
co
m
p
r
eh
e
n
s
iv
e
s
tatis
t
ical
an
aly
s
is
.
3.
M
E
T
H
O
D
T
h
is
s
ec
tio
n
d
escr
ib
es
th
e
m
eth
o
d
o
lo
g
y
ad
o
p
ted
f
o
r
cr
ed
i
t
ca
r
d
f
r
au
d
d
etec
tio
n
,
i
n
clu
d
in
g
d
etails
ab
o
u
t
th
e
d
atab
ase
an
d
th
e
SMOT
E
m
eth
o
d
.
I
t
also
co
v
er
s
th
e
im
p
lem
en
tatio
n
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
u
s
ed
f
o
r
d
etec
tin
g
c
r
ed
it
ca
r
d
f
r
au
d
.
Su
b
s
eq
u
en
t
s
ec
tio
n
s
will
ex
p
lain
th
e
cr
ed
it
ca
r
d
tr
an
s
m
is
s
io
n
d
atab
ase
an
d
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
.
3
.
1
.
Da
t
a
ba
s
e
T
h
e
c
r
e
d
i
t
c
a
r
d
f
r
a
u
d
d
e
t
e
c
t
i
o
n
d
a
t
a
s
e
t
u
s
e
d
i
n
t
h
i
s
s
t
u
d
y
w
a
s
s
o
u
r
c
e
d
f
r
o
m
K
a
g
g
l
e
[
2
5
]
,
a
w
i
d
e
l
y
r
e
c
o
g
n
i
z
e
d
p
l
a
t
f
o
r
m
f
o
r
d
a
t
a
s
c
i
e
n
c
e
r
es
e
a
r
c
h
a
n
d
c
o
m
p
e
t
i
ti
o
n
s
.
T
h
e
d
a
t
as
e
t
c
o
m
p
r
i
s
es
2
8
4
,
8
0
7
t
r
a
n
s
a
c
t
i
o
n
s
c
o
n
d
u
c
t
e
d
b
y
E
u
r
o
p
e
a
n
c
a
r
d
h
o
l
d
e
r
s
i
n
S
e
p
te
m
b
e
r
2
0
1
3
.
I
t
i
s
c
h
a
r
a
ct
e
r
i
z
e
d
b
y
a
s
t
r
o
n
g
c
l
a
s
s
i
m
b
a
l
a
n
c
e
,
c
o
n
t
a
i
n
i
n
g
o
n
l
y
4
9
2
f
r
a
u
d
u
l
e
n
t
t
r
a
n
s
ac
t
i
o
n
s
,
w
h
i
c
h
r
e
p
r
ese
n
t
a
p
p
r
o
x
i
m
a
t
e
l
y
0
.
1
7
2
%
o
f
t
h
e
t
o
t
al
r
e
c
o
r
d
s
.
E
a
c
h
t
r
a
n
s
a
ct
i
o
n
i
s
d
es
c
r
i
b
e
d
b
y
3
0
n
u
m
e
r
i
c
a
l
f
e
a
t
u
r
e
s
,
i
n
clu
d
i
n
g
t
h
e
a
t
t
r
i
b
u
t
es
t
i
m
e
a
n
d
a
m
o
u
n
t
,
w
h
i
l
e
t
h
e
r
e
m
a
i
n
i
n
g
2
8
f
e
a
t
u
r
e
s
w
e
r
e
d
e
r
i
v
e
d
t
h
r
o
u
g
h
p
r
i
n
c
i
p
a
l
c
o
m
p
o
n
e
n
t
a
n
a
l
y
s
is
(
PC
A
)
to
p
r
o
t
e
c
t
s
e
n
s
i
ti
v
e
i
n
f
o
r
m
a
t
i
o
n
a
n
d
r
e
d
u
c
e
d
i
m
e
n
s
i
o
n
a
l
i
t
y
.
T
h
e
t
a
r
g
e
t
v
a
r
i
a
b
le
,
c
la
s
s
,
d
e
n
o
t
es
w
h
e
t
h
e
r
a
t
r
a
n
s
a
c
ti
o
n
i
s
l
e
g
i
ti
m
a
t
e
‘
0
’
o
r
f
r
a
u
d
u
l
e
n
t
‘
1
’
.
T
o
m
i
t
i
g
a
t
e
th
e
i
s
s
u
e
o
f
c
l
as
s
i
m
b
a
l
a
n
ce
,
t
h
e
S
M
O
T
E
w
as
a
p
p
l
ie
d
t
o
t
h
e
tr
a
i
n
i
n
g
s
e
t
.
S
MO
T
E
g
e
n
e
r
a
t
e
s
s
y
n
t
h
e
t
i
c
s
a
m
p
l
es
f
o
r
t
h
e
m
i
n
o
r
i
t
y
c
l
a
s
s
,
a
ll
o
w
i
n
g
f
o
r
a
m
o
r
e
b
a
l
a
n
c
e
d
a
n
d
r
e
p
r
e
s
e
n
t
a
ti
v
e
d
a
t
as
e
t
d
u
r
i
n
g
m
o
d
e
l
t
r
a
i
n
i
n
g
.
A
d
d
i
t
i
o
n
a
l
l
y
,
t
h
e
t
i
m
e
a
n
d
a
m
o
u
n
t
f
e
atu
r
e
s
w
e
r
e
n
o
r
m
a
l
i
z
e
d
u
s
i
n
g
m
in
–
m
a
x
s
c
a
li
n
g
,
a
n
d
a
l
l
i
n
p
u
t
f
e
a
t
u
r
es
we
r
e
s
t
a
n
d
a
r
d
i
z
e
d
t
o
e
n
s
u
r
e
c
o
m
p
a
ti
b
i
li
t
y
a
n
d
s
t
a
b
l
e
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
E
L
M
a
l
g
o
r
i
t
h
m
.
3.
1.
1
.
Da
t
a
prepro
ce
s
s
ing
I
n
th
is
s
tu
d
y
,
th
e
SMOT
E
tech
n
iq
u
e
is
ap
p
lied
ex
cl
u
s
iv
ely
to
th
e
tr
ain
in
g
d
ataset
in
o
r
d
er
to
ad
d
r
ess
class
im
b
alan
ce
b
etwe
en
le
g
itima
te
an
d
f
r
au
d
u
len
t
tr
a
n
s
ac
tio
n
s
.
T
h
e
SMOT
E
tech
n
iq
u
e
o
p
er
ates
b
y
g
en
er
atin
g
s
y
n
th
etic
in
s
tan
ce
s
o
f
th
e
m
i
n
o
r
ity
class
(
f
r
a
u
d
)
u
s
in
g
th
e
KNN
s
ap
p
r
o
ac
h
.
Sp
ec
if
ically
,
f
o
r
ea
c
h
m
in
o
r
ity
in
s
tan
ce
,
n
ew
s
am
p
le
s
ar
e
cr
ea
ted
b
y
in
ter
p
o
latin
g
b
etwe
en
it a
n
d
its
n
ea
r
est n
eig
h
b
o
r
s
in
th
e
f
ea
t
u
r
e
s
p
ac
e,
th
er
eb
y
en
r
ich
i
n
g
th
e
tr
ain
in
g
d
ata
with
o
u
t
s
im
p
ly
d
u
p
licatin
g
o
r
ig
in
al
ex
am
p
les.
T
o
av
o
id
o
v
er
f
itti
n
g
,
th
e
SMOT
E
tech
n
iq
u
e
was
n
o
t
ap
p
lied
t
o
th
e
test
d
ata,
en
s
u
r
in
g
th
at
m
o
d
el
e
v
alu
atio
n
was
co
n
d
u
cted
o
n
o
r
ig
in
al,
u
n
s
ee
n
d
ata.
W
h
ile
s
y
n
th
etic
s
am
p
lin
g
im
p
r
o
v
es
th
e
m
o
d
el’
s
ab
ilit
y
to
lear
n
f
r
o
m
r
ar
e
f
r
au
d
ca
s
es,
ca
r
e
was
tak
en
to
a
p
p
ly
SMO
T
E
o
n
ly
b
ef
o
r
e
th
e
m
o
d
el
wa
s
tr
ain
ed
to
p
r
eser
v
e
th
e
i
n
teg
r
ity
an
d
r
ea
lis
m
o
f
th
e
p
er
f
o
r
m
an
ce
ev
alu
atio
n
.
T
o
p
r
ep
ar
e
th
e
cr
ed
it
ca
r
d
f
r
au
d
d
ataset
f
o
r
f
u
r
th
er
a
n
aly
s
is
,
th
e
f
o
llo
win
g
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
wer
e
p
er
f
o
r
m
ed
:
i)
No
r
m
aliza
tio
n
:
th
e
‘
tim
e’
an
d
‘
am
o
u
n
t’
f
ea
tu
r
es
wer
e
n
o
r
m
alize
d
u
s
in
g
m
in
-
m
ax
s
ca
lin
g
to
en
s
u
r
e
all
v
ar
iab
les ar
e
o
n
a
s
im
ilar
s
ca
le.
ii)
Featu
r
e
s
ca
lin
g
: a
ll f
ea
tu
r
es
wer
e
r
escaled
to
a
r
a
n
g
e
b
etwe
en
0
an
d
1
u
s
in
g
th
e
m
i
n
–
m
ax
s
ca
ler
.
iii)
C
las
s
im
b
alan
ce
h
an
d
lin
g
:
th
e
SMOT
E
was
ap
p
lied
to
th
e
tr
ain
in
g
d
ataset
in
o
r
d
e
r
to
b
al
an
ce
th
e
class
d
is
tr
ib
u
tio
n
,
en
a
b
lin
g
th
e
m
o
d
el
to
lear
n
ef
f
ec
ti
v
ely
f
r
o
m
b
o
t
h
m
ajo
r
ity
a
n
d
m
in
o
r
ity
s
am
p
l
es.
iv
)
T
r
ain
-
test
s
p
lit:
th
e
d
ataset
was
d
iv
id
ed
in
t
o
tr
ain
in
g
(
7
0
%
)
an
d
test
in
g
(
3
0
%)
s
u
b
s
ets
u
s
in
g
s
tr
atif
ied
s
am
p
lin
g
to
m
ain
tain
t
h
e
o
r
i
g
in
al
class
d
is
tr
ib
u
tio
n
ac
r
o
s
s
b
o
th
s
ets.
3
.
2
.
E
x
t
re
m
e
lea
rning
ma
chine
cla
s
s
if
ier
C
r
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
p
l
ay
s
a
v
ital
r
o
le
in
h
elp
in
g
f
in
an
cial
in
s
titu
tio
n
s
p
r
ev
en
t
u
n
au
th
o
r
ize
d
tr
an
s
ac
tio
n
s
an
d
m
in
im
ize
f
in
an
cial
lo
s
s
es.
I
n
th
is
s
tu
d
y
,
th
e
E
L
M
is
em
p
lo
y
e
d
as
th
e
p
r
i
m
ar
y
ML
alg
o
r
ith
m
d
u
e
to
its
ef
f
icien
cy
i
n
h
an
d
l
in
g
lar
g
e
d
atasets
an
d
its
h
ig
h
co
m
p
u
tatio
n
al
s
p
ee
d
.
T
h
e
E
L
M
alg
o
r
ith
m
is
p
r
o
p
o
s
ed
as
an
e
f
f
ec
tiv
e
an
d
s
ca
lab
le
ap
p
r
o
ac
h
f
o
r
d
etec
tin
g
f
r
au
d
u
len
t
ac
tiv
ities
in
c
r
ed
i
t
ca
r
d
tr
an
s
ac
tio
n
s
.
I
t
is
b
ased
o
n
th
e
c
o
n
ce
p
t
o
f
SLFNs,
wh
ich
ar
e
k
n
o
wn
f
o
r
th
eir
ab
ilit
y
to
m
an
ag
e
h
i
g
h
-
d
im
e
n
s
io
n
al
an
d
im
b
alan
ce
d
d
atasets
[
2
6
]
.
I
n
o
th
er
wo
r
d
s
,
th
e
E
L
M
alg
o
r
ith
m
is
a
ty
p
e
o
f
SLFN
ch
ar
ac
ter
ized
b
y
f
ast
tr
ain
in
g
an
d
lo
w
co
m
p
u
tatio
n
al
c
o
s
t.
W
ith
in
th
e
E
L
M
f
r
am
ewo
r
k
,
t
h
e
in
p
u
t
weig
h
ts
an
d
b
iases
o
f
th
e
h
i
d
d
en
lay
er
ar
e
r
an
d
o
m
l
y
ass
ig
n
ed
an
d
r
e
m
ain
f
ix
ed
,
w
h
ile
th
e
o
u
tp
u
t
weig
h
ts
ar
e
ca
lcu
lated
an
aly
tically
u
s
in
g
a
clo
s
ed
-
f
o
r
m
s
o
lu
tio
n
,
o
f
te
n
th
r
o
u
g
h
th
e
M
o
o
r
e
–
Pen
r
o
s
e
p
s
eu
d
o
in
v
er
s
e
[
2
7
]
.
T
h
is
d
esig
n
r
em
o
v
es
t
h
e
n
ee
d
f
o
r
iter
ativ
e
o
p
tim
izatio
n
o
f
h
id
d
e
n
lay
er
s
,
en
ab
lin
g
th
e
m
o
d
el
t
o
ac
h
iev
e
f
aster
co
n
v
er
g
en
ce
.
Owin
g
t
o
its
co
m
p
u
tatio
n
al
ef
f
icien
cy
an
d
ca
p
ac
ity
t
o
p
r
o
ce
s
s
lar
g
e
-
s
ca
le
d
ata,
E
L
M
is
p
ar
ticu
lar
ly
well
-
s
u
ited
f
o
r
f
r
au
d
d
etec
tio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
cred
it c
a
r
d
fr
a
u
d
d
etec
tio
n
w
ith
s
yn
th
etic
min
o
r
ity
o
ve
r
…
(
I
ma
n
K
a
d
h
im
A
jla
n
)
4753
ap
p
licatio
n
s
th
at
r
eq
u
ir
e
r
e
al
-
tim
e
o
r
n
ea
r
-
r
ea
l
-
tim
e
p
er
f
o
r
m
a
n
ce
.
T
h
e
p
r
o
p
o
s
ed
E
L
M
f
r
am
ewo
r
k
d
em
o
n
s
tr
ates
r
ap
id
tr
ain
in
g
c
ap
ab
ilit
y
,
s
tr
o
n
g
g
en
er
aliza
tio
n
p
er
f
o
r
m
an
ce
,
an
d
th
e
in
teg
r
atio
n
o
f
s
y
n
th
etic
o
v
er
s
am
p
lin
g
m
eth
o
d
s
to
ad
d
r
ess
class
im
b
alan
ce
is
s
u
es
.
T
h
e
in
itial
s
tag
e
o
f
th
e
f
r
am
ewo
r
k
in
v
o
lv
es
p
r
ep
ar
in
g
th
e
d
ataset
to
en
s
u
r
e
o
p
tim
al
p
e
r
f
o
r
m
an
ce
o
f
th
e
E
L
M
alg
o
r
ith
m
.
T
h
e
cr
ed
it
ca
r
d
d
ataset
is
f
ir
s
t
lo
ad
ed
,
an
d
th
e
tim
e
an
d
am
o
u
n
t
f
ea
tu
r
es
ar
e
n
o
r
m
alize
d
u
s
in
g
th
e
m
in
–
m
ax
s
ca
lin
g
tech
n
iq
u
e
to
b
r
in
g
th
eir
v
alu
es
with
in
t
h
e
r
an
g
e
o
f
[
0
,
1
]
.
T
h
is
n
o
r
m
aliza
tio
n
s
tep
en
s
u
r
es
th
at
all
i
n
p
u
t
f
ea
tu
r
es
co
n
tr
ib
u
te
u
n
if
o
r
m
l
y
d
u
r
in
g
m
o
d
el
tr
ain
in
g
.
T
h
e
d
at
aset is
th
en
d
iv
id
ed
in
to
in
p
u
t
f
ea
tu
r
es (
x
)
an
d
t
h
e
tar
g
et
v
ar
i
ab
le
(
y
)
,
wh
er
e
th
e
class
at
tr
ib
u
te
id
en
tifie
s
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
‘
1
’
an
d
leg
it
im
ate
tr
an
s
ac
tio
n
s
‘
0
’
.
T
o
av
o
id
d
ata
leak
ag
e,
th
e
d
ata
is
f
u
r
th
er
s
p
lit in
to
t
r
ain
in
g
an
d
test
in
g
s
u
b
s
ets u
s
in
g
7
0
:
3
0
r
atio
.
On
e
o
f
th
e
m
ajo
r
ch
allen
g
es
in
cr
ed
it
ca
r
d
f
r
a
u
d
d
etec
tio
n
is
th
e
e
x
tr
em
e
class
im
b
alan
ce
,
wh
er
e
f
r
au
d
u
le
n
t
tr
an
s
ac
tio
n
s
r
ep
r
es
en
t
less
th
an
1
%
o
f
th
e
to
tal
d
ataset.
T
h
is
im
b
alan
ce
o
f
ten
b
iases
ML
m
o
d
els
to
war
d
p
r
ed
ictin
g
th
e
m
aj
o
r
ity
class
,
r
esu
lt
in
g
in
h
ig
h
o
v
er
al
l
ac
cu
r
ac
y
b
u
t
p
o
o
r
f
r
au
d
d
ete
ctio
n
p
er
f
o
r
m
an
ce
.
T
o
o
v
e
r
co
m
e
t
h
is
is
s
u
e,
th
e
SMOT
E
is
in
teg
r
ated
in
to
th
e
tr
ain
in
g
p
r
o
ce
s
s
.
SMOT
E
g
en
er
ates
s
y
n
th
etic
s
am
p
les
o
f
th
e
m
in
o
r
ity
class
b
y
in
ter
p
o
latin
g
b
etwe
en
e
x
is
tin
g
m
in
o
r
ity
in
s
tan
ce
s
,
th
er
eb
y
in
cr
ea
s
in
g
th
eir
r
ep
r
esen
tatio
n
with
in
th
e
d
at
aset
[
2
8
]
.
B
alan
cin
g
th
e
d
ataset
th
r
o
u
g
h
SMOT
E
en
ab
les
th
e
m
o
d
el
to
lear
n
m
o
r
e
e
f
f
ec
tiv
ely
f
r
o
m
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
p
atter
n
s
.
T
h
is
ap
p
r
o
ac
h
is
p
a
r
ticu
lar
ly
b
en
ef
icial
f
o
r
t
h
e
E
L
M
alg
o
r
ith
m
,
as
it
en
h
an
ce
s
th
e
m
o
d
el’
s
ab
ilit
y
to
d
if
f
er
en
tiate
b
etwe
en
m
in
o
r
ity
an
d
m
aj
o
r
ity
class
es,
im
p
r
o
v
es
r
ec
all,
an
d
r
e
d
u
ce
s
f
alse
n
eg
at
iv
es.
Sin
ce
f
r
a
u
d
u
le
n
t
tr
an
s
ac
t
io
n
s
ar
e
s
ig
n
if
ican
tly
f
ewe
r
t
h
an
leg
itima
te
o
n
es,
SMOT
E
is
ap
p
lied
o
n
ly
to
th
e
tr
ain
in
g
d
ata
to
m
ain
tain
r
e
alis
tic
ev
alu
atio
n
co
n
d
itio
n
s
.
T
h
e
E
L
M
m
o
d
el
is
in
itialized
b
y
d
ef
i
n
in
g
a
r
an
g
e
o
f
h
i
d
d
en
n
o
d
es,
t
y
p
ically
b
etwe
en
1
0
a
n
d
1
0
0
,
an
d
b
y
s
elec
tin
g
s
u
itab
le
ac
tiv
atio
n
f
u
n
ctio
n
s
s
u
ch
as
s
ig
m
o
id
o
r
R
eL
U.
T
h
ese
co
n
f
ig
u
r
atio
n
s
allo
w
ex
p
er
im
en
tat
io
n
to
id
en
tify
th
e
o
p
tim
al
co
m
b
i
n
atio
n
o
f
h
id
d
en
n
o
d
es
a
n
d
ac
tiv
atio
n
f
u
n
ctio
n
s
f
o
r
ac
h
ie
v
in
g
h
ig
h
er
ac
cu
r
ac
y
in
f
r
au
d
d
etec
tio
n
.
As
a
b
in
ar
y
class
if
icatio
n
task
ch
ar
ac
ter
ized
b
y
h
ig
h
d
ata
im
b
alan
ce
,
th
is
s
etu
p
en
s
u
r
es
th
at
th
e
E
L
M
m
o
d
el
ca
n
ac
h
iev
e
b
alan
ce
d
an
d
r
eliab
le
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
b
o
th
class
es.
i)
Sig
m
o
id
:
th
e
s
ig
m
o
id
f
u
n
ctio
n
is
o
n
e
o
f
th
e
m
o
s
t
wid
ely
u
s
ed
ac
tiv
atio
n
f
u
n
ctio
n
s
in
class
if
icatio
n
task
s
,
esp
ec
ially
wh
en
th
e
g
o
al
is
to
o
u
tp
u
t
p
r
o
b
a
b
ilit
ies
[
2
9
]
.
I
t
m
ap
s
an
y
r
ea
l
-
v
al
u
ed
n
u
m
b
er
in
to
a
r
an
g
e
b
etwe
en
0
an
d
1
,
wh
ic
h
alig
n
s
well
with
b
in
ar
y
class
if
icatio
n
o
b
jectiv
es,
s
u
c
h
as
d
is
tin
g
u
is
h
in
g
b
etwe
en
f
r
a
u
d
u
len
t
an
d
leg
itima
te
tr
an
s
ac
tio
n
s
[
3
0
]
.
Ho
we
v
er
,
it
m
a
y
s
u
f
f
e
r
f
r
o
m
th
e
v
a
n
is
h
in
g
g
r
ad
ien
t
p
r
o
b
lem
,
wh
ich
ca
n
h
in
d
e
r
tr
a
in
in
g
in
d
ee
p
er
o
r
m
o
r
e
co
m
p
l
ex
m
o
d
els.
ii)
R
eL
U:
th
e
R
eL
U
is
a
p
iece
w
i
s
e
lin
ea
r
ac
tiv
atio
n
f
u
n
ctio
n
t
h
o
s
e
o
u
tp
u
ts
ze
r
o
f
o
r
n
e
g
ativ
e
in
p
u
t
v
alu
es
an
d
r
etu
r
n
s
th
e
in
p
u
t
v
alu
e
its
elf
f
o
r
p
o
s
itiv
e
in
p
u
ts
.
T
h
is
p
r
o
p
er
ty
in
t
r
o
d
u
ce
s
s
p
ar
s
ity
in
n
eu
r
al
ac
tiv
atio
n
s
,
lead
in
g
to
f
aster
c
o
n
v
er
g
en
ce
an
d
lo
wer
co
m
p
u
t
atio
n
al
co
m
p
lex
ity
.
T
h
ese
ad
v
an
tag
es
m
ak
e
R
eL
U
h
ig
h
ly
ef
f
ec
tiv
e
f
o
r
lar
g
e
-
s
ca
le
an
d
r
ea
l
-
tim
e
f
r
au
d
d
et
ec
tio
n
ap
p
licatio
n
s
[
3
1
]
.
I
n
a
d
d
itio
n
,
R
eL
U
h
elp
s
r
ed
u
ce
th
e
im
p
ac
t
o
f
th
e
v
an
is
h
in
g
g
r
a
d
ien
t
p
r
o
b
lem
,
t
h
er
eb
y
im
p
r
o
v
in
g
th
e
lea
r
n
in
g
ef
f
icien
cy
o
f
m
o
d
els s
u
ch
as E
L
M
[
3
2
]
.
T
h
ese
two
ac
tiv
atio
n
f
u
n
ctio
n
s
wer
e
s
elec
ted
to
ass
ess
th
e
E
L
M
m
o
d
el’
s
a
d
ap
tab
ilit
y
to
d
if
f
er
e
n
t
ac
tiv
atio
n
b
eh
av
io
r
s
.
T
h
e
s
ig
m
o
id
f
u
n
ctio
n
p
r
o
v
id
es
s
m
o
o
t
h
p
r
o
b
a
b
ilis
tic
m
ap
p
in
g
,
aid
in
g
in
th
e
d
etec
tio
n
o
f
m
in
o
r
ity
class
es
s
u
ch
as
f
r
au
d
ca
s
es,
wh
ile
R
eL
U
em
p
h
asizes
co
m
p
u
tatio
n
al
ef
f
icien
cy
a
n
d
s
ca
lab
ilit
y
,
wh
ich
ar
e
ess
en
tial
f
o
r
lar
g
e
-
s
ca
le
f
in
an
cial
ap
p
licatio
n
s
.
T
h
e
E
L
M
alg
o
r
ith
m
tr
ain
s
th
e
s
in
g
le
-
lay
er
f
ee
d
f
o
r
war
d
n
etwo
r
k
u
s
in
g
r
an
d
o
m
izatio
n
an
d
least
-
s
q
u
ar
es o
p
tim
izatio
n
,
as o
u
tlin
ed
in
t
h
e
f
o
llo
win
g
m
ath
em
atica
l step
s
.
i)
Hid
d
en
lay
e
r
co
m
p
u
tatio
n
:
let
th
e
in
p
u
t
d
ataset
b
e
X
∈
ℝ
×
,
wh
er
e
is
th
e
n
u
m
b
er
o
f
s
am
p
les
an
d
is
th
e
n
u
m
b
er
o
f
f
ea
tu
r
es.
R
an
d
o
m
ly
in
itialize
th
e
i
n
p
u
t
weig
h
t
m
atr
ix
W
∈
ℝ
×
an
d
b
ias
v
ec
t
o
r
b
∈
ℝ
,
wh
er
e
is
th
e
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es.
T
h
e
h
i
d
d
en
la
y
er
o
u
tp
u
t,
H
∈
ℝ
×
is
ca
lcu
lated
as (
1
)
.
=
(
+
)
(
1
)
Her
e
,
(
⋅
)
r
ep
r
esen
ts
th
e
ac
tiv
atio
n
f
u
n
ctio
n
(
e.
g
.
,
s
ig
m
o
id
o
r
R
eL
U:)
.
W
h
er
e
th
e
ac
tiv
atio
n
f
u
n
ctio
n
s
o
f
s
ig
m
o
id
an
d
R
eL
U
ar
e
c
o
m
p
u
ted
as sh
o
wn
in
(
2
)
an
d
(
3
)
,
r
e
s
p
ec
tiv
ely
.
=
(
)
=
1
1
+
−
(
2
)
=
(
)
=
(
0
,
)
(
3
)
ii)
Ou
tp
u
t
weig
h
t
ca
lcu
latio
n
:
th
e
o
u
tp
u
t
weig
h
ts
∈
ℝ
×
1
ar
e
co
m
p
u
ted
u
s
in
g
th
e
Mo
o
r
e
-
Pe
n
r
o
s
e
p
s
eu
d
o
-
in
v
er
s
e
o
f
t
h
e
h
id
d
en
l
ay
er
o
u
t
p
u
t m
atr
ix
H
,
as sh
o
wn
in
(
4
)
.
=
†
(
4
)
W
h
er
e
,
H
†
is
th
e
p
s
eu
d
o
-
in
v
er
s
e
o
f
H
,
an
d
T
is
th
e
tar
g
et
v
ec
to
r
,
T
∈
ℝ
×
1
(
class
lab
el
s
: 0
f
o
r
n
o
n
-
f
r
au
d
an
d
1
f
o
r
f
r
au
d
)
.
T
h
e
p
s
eu
d
o
-
i
n
v
er
s
e
o
f
H
,
H
†
is
co
m
p
u
ted
as sh
o
wn
in
(
5
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
7
4
9
-
4
7
6
2
4754
†
=
(
⊤
)
−
1
⊤
(
5
)
iii)
Pre
d
i
cti
o
n
:
Fo
r
a
g
iv
en
t
est
i
n
p
u
t
tes
t
∈
tes
t
×
,
t
h
e
p
r
ed
ict
ed
o
u
t
p
u
t
pr
e
d
∈
tes
t
×
1
is
ca
l
cu
l
at
ed
as
(
6
)
.
p
r
e
d
=
(
t
e
st
+
)
(
6
)
T
h
r
esh
o
ld
in
g
is
ap
p
lied
to
pr
e
d
to
o
b
tain
b
in
a
r
y
p
r
ed
ictio
n
s
(
f
r
au
d
o
r
n
o
n
-
f
r
au
d
)
,
as g
iv
e
n
in
(
7
)
.
̂
=
{
1
,
if
p
re
d
,
≥
0
.
5
0
,
othe
r
wi
s
e
(
7
)
W
h
er
e,
̂
is
th
e
b
in
ar
y
p
r
e
d
ictio
n
f
o
r
t
h
e
i
-
th
s
am
p
le,
wh
er
e
1
in
d
icate
s
f
r
au
d
a
n
d
0
in
d
icate
s
n
o
n
-
f
r
au
d
.
Fu
r
th
er
m
o
r
e
,
th
e
p
r
o
p
o
s
ed
E
L
M
m
o
d
el
is
tr
ain
ed
u
s
in
g
th
e
b
alan
ce
d
tr
ain
in
g
d
ata
s
et.
Du
r
in
g
ev
alu
atio
n
,
th
e
test
d
ata
is
p
ass
ed
th
r
o
u
g
h
th
e
m
o
d
el
to
c
o
m
p
u
te
p
r
ed
ictio
n
s
.
B
in
ar
y
lab
els
ar
e
g
en
er
ated
b
y
ap
p
ly
in
g
a
th
r
esh
o
ld
o
f
0
.
5
to
th
e
p
r
ed
icted
v
al
u
es.
Key
ev
al
u
atio
n
m
etr
ics ar
e
ca
lcu
lated
,
i
n
clu
d
in
g
ac
c
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
(
s
en
s
itiv
ity
)
,
s
p
ec
if
icity
,
F
-
m
ea
s
u
r
e,
an
d
G
-
m
ea
n
.
T
h
ese
m
etr
ics
co
m
p
r
eh
en
s
iv
ely
ass
ess
th
e
m
o
d
el’
s
p
e
r
f
o
r
m
an
ce
in
id
e
n
tify
in
g
f
r
a
u
d
u
len
t
tr
an
s
ac
ti
o
n
s
.
I
n
a
d
d
itio
n
,
th
e
r
esu
lt
s
ar
e
s
u
b
jecte
d
to
d
escr
ip
tiv
e
s
tatis
tical
an
aly
s
i
s
to
s
u
m
m
ar
ize
th
e
p
er
f
o
r
m
an
ce
ac
r
o
s
s
v
ar
io
u
s
co
n
f
ig
u
r
atio
n
s
o
f
h
id
d
en
n
o
d
es
an
d
ac
tiv
atio
n
f
u
n
ctio
n
s
.
Sig
n
if
ican
ce
test
in
g
is
co
n
d
u
cte
d
to
ascer
tain
wh
eth
er
o
b
s
er
v
ed
d
if
f
e
r
en
ce
s
in
m
etr
ics
ar
e
s
tatis
tically
m
ea
n
in
g
f
u
l.
Vis
u
aliza
tio
n
s
,
s
u
ch
as
lin
e
p
lo
ts
,
co
m
p
ar
e
m
etr
ics
(
e.
g
.
,
s
p
ec
if
icity
v
s
.
h
id
d
en
n
o
d
es
)
to
id
e
n
tify
o
p
tim
al
co
n
f
ig
u
r
atio
n
s
.
Fig
u
r
e
1
i
llu
s
tr
ates
th
e
f
lo
wch
ar
t
an
d
th
e
k
ey
s
tep
s
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
in
d
etec
tin
g
cr
ed
it c
ar
d
f
r
au
d
.
I
n
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
,
th
e
d
ataset
will
b
e
lo
ad
an
d
p
r
ep
r
o
ce
s
s
.
T
h
en
,
th
e
SMOT
E
m
eth
o
d
is
ap
p
lied
to
b
alan
ce
t
h
e
tr
ain
i
n
g
d
ata.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
is
in
itialized
b
y
d
ef
in
in
g
th
e
h
id
d
en
n
o
d
es
(
1
0
to
1
0
0
)
a
n
d
ac
tiv
atio
n
f
u
n
ctio
n
s
(
s
ig
m
o
id
an
d
R
eL
U)
.
Af
ter
th
e
in
itializatio
n
s
tep
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
will
b
e
tr
ain
ed
b
y
co
m
p
u
tin
g
h
id
d
en
lay
er
o
u
tp
u
ts
an
d
o
u
tp
u
t
weig
h
ts
.
Su
b
s
eq
u
en
tly
,
p
r
e
d
ictio
n
s
an
d
b
in
a
r
y
lab
els
ar
e
g
en
e
r
ated
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
will
b
e
e
v
alu
ated
b
y
ca
lcu
latin
g
m
etr
ics
an
d
s
av
e
r
esu
lts
.
L
astl
y
,
b
ased
o
n
t
h
e
ex
p
er
im
e
n
tal
r
esu
lt,
th
e
s
tatis
tical
an
aly
s
is
an
d
v
is
u
aliza
ti
o
n
ar
e
p
er
f
o
r
m
ed
.
Fig
u
r
e
1
.
T
h
e
f
lo
wch
a
r
t o
f
t
h
e
p
r
o
p
o
s
ed
E
L
M
m
o
d
el
f
o
r
cr
e
d
it c
ar
d
f
r
a
u
d
d
etec
tio
n
4.
E
XP
E
R
I
M
E
N
T
A
L
SE
T
UP
AND
RE
SUL
T
S AN
AL
Y
SI
S
T
h
e
p
r
o
p
o
s
ed
E
L
M
m
o
d
el
was
im
p
lem
en
ted
to
d
etec
t
c
r
ed
it
ca
r
d
f
r
a
u
d
in
a
h
ig
h
ly
im
b
alan
ce
d
d
ataset.
T
h
e
d
ataset
co
n
s
is
ted
o
f
leg
itima
te
an
d
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
,
wh
er
e
th
e
latter
co
n
s
titu
ted
o
n
l
y
a
s
m
all
f
r
ac
tio
n
o
f
th
e
t
o
tal
s
am
p
les.
T
o
ad
d
r
ess
th
is
im
b
alan
c
e,
th
e
SMOT
E
tech
n
i
q
u
e
was a
p
p
lied
ex
cl
u
s
iv
ely
to
th
e
tr
ain
in
g
s
et,
en
s
u
r
in
g
an
eq
u
al
r
ep
r
esen
tatio
n
o
f
b
o
th
class
e
s
wh
ile
m
ain
tain
in
g
th
e
in
teg
r
ity
o
f
th
e
test
in
g
s
et.
T
h
e
d
ataset
was
p
r
ep
r
o
ce
s
s
ed
b
y
n
o
r
m
alizin
g
n
u
m
e
r
ical
f
ea
tu
r
es,
in
clu
d
in
g
th
e
tr
an
s
ac
tio
n
"
am
o
u
n
t"
an
d
"tim
e,
"
u
s
in
g
m
in
-
m
ax
s
ca
lin
g
.
Su
b
s
eq
u
en
t
ly
,
th
e
f
ea
tu
r
es
wer
e
s
ca
led
t
o
en
s
u
r
e
u
n
if
o
r
m
ity
.
T
h
e
d
ata
was
s
p
lit
in
to
tr
ain
i
n
g
a
n
d
test
in
g
s
ets
with
a
7
0
:3
0
r
atio
.
T
h
e
p
r
o
p
o
s
ed
E
L
M
m
o
d
el
em
p
l
o
y
ed
two
ac
tiv
atio
n
f
u
n
ctio
n
s
f
o
r
t
h
e
h
id
d
en
lay
er
,
s
ig
m
o
id
an
d
R
eL
U,
to
ev
alu
ate
th
eir
im
p
ac
t
o
n
p
er
f
o
r
m
an
ce
.
T
h
e
n
u
m
b
er
o
f
h
id
d
e
n
n
o
d
es
was
v
ar
ied
f
r
o
m
1
0
t
o
1
0
0
,
with
in
cr
em
en
ts
o
f
1
0
,
t
o
an
al
y
ze
th
e
in
f
l
u
en
ce
o
f
th
e
h
id
d
e
n
lay
er
'
s
co
m
p
lex
ity
.
I
n
o
th
e
r
wo
r
d
s
,
to
d
eter
m
i
n
e
th
e
o
p
tim
al
co
n
f
ig
u
r
atio
n
o
f
h
id
d
en
n
o
d
es
in
th
e
E
L
M,
we
c
o
n
d
u
cted
v
ar
io
u
s
e
x
p
er
im
en
ts
.
Fo
r
ea
c
h
c
o
n
f
ig
u
r
atio
n
,
th
e
E
L
M
m
o
d
el
was
tr
a
in
ed
an
d
e
v
alu
ated
u
s
in
g
th
e
s
am
e
d
ataset
an
d
e
x
p
er
im
en
tal
co
n
d
itio
n
s
.
T
h
e
s
elec
tio
n
cr
iter
io
n
f
o
r
th
e
b
est
co
n
f
ig
u
r
atio
n
was
p
r
im
ar
ily
th
e
ac
c
u
r
ac
y
ac
h
iev
ed
o
n
th
e
test
s
et,
as
it
r
ef
lects
th
e
m
o
d
el'
s
o
v
er
all
p
er
f
o
r
m
an
ce
.
T
h
e
co
n
f
i
g
u
r
atio
n
th
at
y
ield
e
d
th
e
h
ig
h
est
ac
cu
r
ac
y
was
s
elec
ted
as
th
e
o
p
tim
al
s
etu
p
.
W
h
ile
o
th
er
m
etr
ics
s
u
ch
as
r
ec
all,
s
p
ec
if
icity
,
a
n
d
G
-
m
ea
n
wer
e
also
co
n
s
id
er
ed
,
ac
cu
r
ac
y
s
er
v
e
d
as
th
e
d
ec
is
iv
e
f
ac
to
r
in
f
in
alizin
g
th
e
b
est
n
u
m
b
e
r
o
f
h
id
d
e
n
n
o
d
es.
T
h
e
p
r
o
p
o
s
e
d
m
o
d
el
h
as
b
ee
n
e
v
alu
ated
in
ter
m
s
o
f
m
an
y
ev
alu
atio
n
m
etr
ics wh
ich
ca
n
b
e
lis
ted
as f
o
llo
ws
[
3
3
]
–
[
3
6
]
.
‒
Acc
u
r
ac
y
:
th
e
p
r
o
p
o
r
tio
n
o
f
c
o
r
r
ec
tly
class
if
ied
tr
an
s
ac
tio
n
s
am
o
n
g
all
tr
an
s
ac
tio
n
s
.
I
t
is
ca
lcu
lated
as
s
h
o
wn
in
(
8
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
cred
it c
a
r
d
fr
a
u
d
d
etec
tio
n
w
ith
s
yn
th
etic
min
o
r
ity
o
ve
r
…
(
I
ma
n
K
a
d
h
im
A
jla
n
)
4755
=
+
+
+
+
(
8
)
‒
Pre
cisi
o
n
:
th
e
p
r
o
p
o
r
tio
n
o
f
c
o
r
r
ec
tly
class
if
ied
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
am
o
n
g
all
p
r
ed
ic
ted
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
.
T
h
is
m
etr
ic
is
co
m
p
u
ted
as g
iv
e
n
in
(
9
)
.
=
+
(
9
)
‒
R
ec
all
(
s
en
s
itiv
ity
)
:
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tl
y
class
if
ied
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
am
o
n
g
all
ac
tu
al
f
r
au
d
u
le
n
t tr
an
s
ac
tio
n
s
.
I
t is c
o
m
p
u
ted
as sh
o
wn
in
(
1
0
)
.
=
+
(
1
0
)
‒
Sp
ec
if
icity
:
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tly
class
if
ied
leg
itima
t
e
tr
an
s
ac
tio
n
s
am
o
n
g
all
ac
t
u
al
leg
itima
te
tr
an
s
ac
tio
n
s
.
T
h
is
m
etr
ic
is
ca
lcu
lated
as sh
o
wn
in
(
1
1
)
.
=
+
(
1
1
)
‒
F
-
m
ea
s
u
r
e
:
th
e
h
ar
m
o
n
ic
m
ea
n
s
o
f
p
r
ec
is
io
n
an
d
r
ec
all
m
etr
ics.
T
h
e
p
er
f
o
r
m
an
ce
m
etr
ic
ca
n
b
e
co
m
p
u
ted
as sh
o
wn
in
(
1
2
)
.
−
=
2
×
×
+
(
1
2
)
‒
G
-
m
ea
n
:
th
e
g
eo
m
etr
ic
m
ea
n
o
f
s
en
s
itiv
ity
an
d
s
p
ec
if
icity
m
etr
ics.
T
h
e
G
-
m
ea
n
m
etr
ic
i
s
ca
lcu
lated
as
s
h
o
wn
in
(
1
3
)
.
−
=
√
×
2
(
1
3
)
I
n
th
e
co
n
tex
t
o
f
cr
ed
it
ca
r
d
f
r
au
d
d
etec
tio
n
,
ev
alu
atin
g
m
o
d
el
p
e
r
f
o
r
m
an
ce
u
s
in
g
a
d
iv
er
s
e
s
et
o
f
m
etr
ics
is
e
s
s
en
tial
d
u
e
to
t
h
e
in
h
er
en
t
class
im
b
alan
ce
.
W
h
ile
ac
cu
r
ac
y
p
r
o
v
id
es
a
g
en
er
al
m
ea
s
u
r
e
o
f
co
r
r
ec
tn
ess
,
it
ca
n
b
e
m
is
lead
in
g
in
im
b
alan
ce
d
d
atasets
wh
er
e
th
e
m
ajo
r
ity
class
d
o
m
in
ates.
Fo
r
ex
am
p
le,
a
m
o
d
el
p
r
e
d
ictin
g
all
tr
an
s
ac
tio
n
s
as
leg
itima
te
m
ay
s
till
ac
h
iev
e
h
ig
h
ac
c
u
r
ac
y
.
T
o
a
d
d
r
ess
th
is
,
r
ec
all
(
s
en
s
itiv
ity
)
is
p
ar
ticu
lar
ly
i
m
p
o
r
tan
t
as
it
m
ea
s
u
r
es
th
e
m
o
d
el’
s
a
b
ilit
y
to
c
o
r
r
ec
tly
id
en
tify
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
,
a
cr
itical
asp
ec
t
in
m
in
im
izin
g
f
in
an
cial
lo
s
s
es.
Sp
ec
if
icity
,
o
n
th
e
o
th
er
h
an
d
,
ev
al
u
ates
h
o
w
well
th
e
m
o
d
el
id
en
tifie
s
leg
it
im
ate
tr
an
s
ac
tio
n
s
,
wh
ich
h
el
p
s
r
ed
u
ce
f
alse
alar
m
s
.
A
d
d
itio
n
ally
,
t
h
e
G
-
m
ea
n
m
etr
ic
o
f
f
er
s
a
b
alan
ce
d
m
ea
s
u
r
e
b
y
co
m
b
in
in
g
r
ec
all
an
d
s
p
ec
if
icity
,
p
r
o
v
id
in
g
in
s
ig
h
t
in
to
th
e
m
o
d
el’
s
ab
ilit
y
to
p
er
f
o
r
m
well
o
n
b
o
t
h
class
es.
B
y
u
s
in
g
m
u
ltip
le
m
etr
ics,
th
is
s
tu
d
y
e
n
s
u
r
es
a
h
o
lis
tic
ev
alu
atio
n
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
'
s
p
er
f
o
r
m
an
ce
an
d
its
p
r
ac
tical
r
elev
a
n
ce
in
r
ea
l
-
wo
r
ld
f
r
a
u
d
d
etec
tio
n
s
y
s
tem
s
.
T
h
e
ac
tiv
atio
n
f
u
n
ctio
n
p
lay
s
a
cr
itical
r
o
le
in
d
eter
m
in
in
g
t
h
e
lear
n
in
g
an
d
r
ep
r
esen
tatio
n
ca
p
ac
ity
o
f
th
e
E
L
M
m
o
d
el.
R
esu
lts
s
h
o
w
th
at
b
o
th
s
ig
m
o
id
an
d
R
e
L
U
ex
h
ib
it
d
is
tin
ct
p
atter
n
s
in
p
er
f
o
r
m
an
ce
as
th
e
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es
in
cr
ea
s
es.
T
ab
le
1
s
h
o
ws
th
e
ex
p
er
im
en
tal
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
in
d
etec
tin
g
cr
e
d
it
ca
r
d
f
r
au
d
.
T
h
e
s
ig
m
o
id
f
u
n
ctio
n
d
em
o
n
s
tr
ates
co
n
s
is
ten
t
im
p
r
o
v
em
e
n
ts
in
k
e
y
m
etr
ics,
p
ar
ticu
lar
ly
ac
cu
r
ac
y
an
d
r
ec
all,
as
th
e
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es
in
cr
ea
s
es.
Fo
r
in
s
tan
c
e,
with
3
0
h
id
d
en
n
o
d
es,
th
e
s
ig
m
o
i
d
ac
tiv
atio
n
ac
h
iev
es
an
ac
c
u
r
ac
y
o
f
9
9
.
2
2
5
%,
r
ec
all
o
f
8
1
.
7
5
7
%,
an
d
a
G
-
m
ea
n
o
f
9
0
.
0
8
2
%,
in
d
icatin
g
a
b
ala
n
ce
d
p
e
r
f
o
r
m
an
ce
i
n
d
etec
tin
g
b
o
th
leg
itima
te
an
d
f
r
a
u
d
u
len
t
tr
an
s
ac
tio
n
s
.
I
n
co
n
tr
ast,
th
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
s
h
o
ws
a
m
o
r
e
g
r
a
d
u
al
im
p
r
o
v
em
e
n
t
ac
r
o
s
s
th
e
m
etr
ics,
ac
h
iev
in
g
co
m
p
etitiv
e
r
esu
lts
at
h
ig
h
er
h
id
d
en
n
o
d
e
c
o
u
n
ts
.
W
h
ile
R
eL
U’
s
p
r
ec
is
io
n
is
s
lig
h
tly
lo
wer
co
m
p
ar
e
d
to
s
ig
m
o
id
at
lo
wer
n
o
d
e
co
u
n
ts
,
it
g
ain
s
s
tab
ilit
y
an
d
ac
h
iev
es
co
m
p
ar
ab
le
r
ec
all
an
d
G
-
m
ea
n
v
alu
es
as
th
e
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es
in
cr
e
ases
.
T
h
is
tr
en
d
h
ig
h
lig
h
ts
R
eL
U’
s
p
o
ten
tial
f
o
r
s
ca
lab
ilit
y
i
n
co
m
p
lex
m
o
d
els.
T
ab
le
2
s
h
o
ws
th
e
co
n
f
u
s
io
n
m
atr
ix
v
alu
es
f
o
r
th
e
h
ig
h
est
r
esu
lts
ac
h
iev
ed
b
y
th
e
p
r
o
p
o
s
ed
m
eth
o
d
co
n
ce
r
n
in
g
tr
u
e
p
o
s
itiv
es
,
tr
u
e
n
eg
ativ
es
,
f
alse
p
o
s
itiv
es
,
an
d
f
alse
n
eg
ativ
es
.
Fig
u
r
e
2
s
h
o
ws
th
e
b
est
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
f
o
r
d
etec
tin
g
cr
e
d
it c
ar
d
f
r
a
u
d
.
T
h
e
r
esu
lts
d
em
o
n
s
tr
ate
th
at
t
h
e
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es
s
ig
n
if
ican
tly
in
f
lu
en
ce
s
th
e
p
e
r
f
o
r
m
an
ce
o
f
th
e
E
L
M
m
o
d
el.
W
h
en
u
s
in
g
th
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
,
ac
cu
r
ac
y
im
p
r
o
v
ed
m
ar
k
ed
l
y
f
r
o
m
9
5
.
9
0
6
%
at
1
0
h
id
d
e
n
n
o
d
es to
a
p
ea
k
o
f
9
9
.
2
2
5
% a
t 3
0
n
o
d
es.
B
ey
o
n
d
t
h
is
p
o
in
t,
th
e
p
er
f
o
r
m
an
ce
b
eg
an
to
p
latea
u
,
with
ac
cu
r
ac
y
f
lu
ctu
atin
g
s
lig
h
tly
with
in
th
e
r
an
g
e
o
f
9
8
.
7
to
9
8
.
9
%,
in
d
icatin
g
d
im
i
n
is
h
in
g
r
etu
r
n
s
f
r
o
m
a
d
d
in
g
m
o
r
e
n
o
d
es.
Similar
ly
,
with
th
e
R
eL
U
ac
tiv
atio
n
f
u
n
cti
o
n
,
t
h
e
h
ig
h
est
ac
c
u
r
ac
y
o
f
9
9
.
7
1
0
%
was
also
ac
h
iev
ed
at
3
0
h
id
d
en
n
o
d
es.
Su
b
s
eq
u
e
n
t
in
cr
ea
s
es
in
h
i
d
d
en
n
o
d
es
p
r
o
d
u
ce
d
m
ar
g
in
al
v
ar
iatio
n
s
,
with
ac
cu
r
ac
y
r
a
n
g
in
g
b
etwe
en
9
8
.
3
an
d
9
9
.
3
%.
T
h
ese
f
in
d
in
g
s
s
u
g
g
est
th
at
an
o
p
tim
al
co
n
f
ig
u
r
atio
n
lik
ely
ex
is
ts
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
7
4
9
-
4
7
6
2
4756
with
in
th
e
3
0
–
4
0
h
i
d
d
en
n
o
d
e
r
an
g
e.
I
n
cr
ea
s
in
g
th
e
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es
b
ey
o
n
d
t
h
is
th
r
esh
o
ld
d
o
es
n
o
t
r
esu
lt in
s
u
b
s
tan
tial p
er
f
o
r
m
an
ce
g
ain
s
an
d
m
a
y
lead
to
u
n
n
e
ce
s
s
ar
y
co
m
p
u
tatio
n
al
c
o
m
p
le
x
ity
.
B
ased
o
n
th
e
ex
p
er
im
en
tal
r
esu
lts
o
b
tain
ed
f
r
o
m
t
h
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
,
t
h
e
b
est
p
er
f
o
r
m
a
n
ce
f
o
r
ea
ch
ev
alu
atio
n
m
etr
ic
an
d
im
p
o
r
tan
t in
s
ig
h
ts
in
to
th
e
ac
h
iev
ed
r
esu
lts
ca
n
b
e
s
u
m
m
ar
i
ze
d
as f
o
llo
ws:
i)
Acc
u
r
ac
y
:
th
e
o
v
er
all
ac
cu
r
a
cy
o
f
th
e
m
o
d
el
im
p
r
o
v
es
with
an
in
c
r
ea
s
e
in
h
id
d
e
n
n
o
d
es
f
o
r
b
o
t
h
ac
tiv
atio
n
f
u
n
ctio
n
s
,
r
ea
c
h
in
g
o
v
er
9
9
%
in
m
u
ltip
le
co
n
f
i
g
u
r
atio
n
s
.
T
h
is
h
ig
h
ac
c
u
r
ac
y
r
ef
lects
th
e
m
o
d
el’
s
ef
f
ec
tiv
e
n
ess
in
co
r
r
e
ctly
class
if
y
in
g
tr
an
s
ac
tio
n
s
.
ii)
Pre
cisi
o
n
:
p
r
ec
is
io
n
,
wh
ich
m
ea
s
u
r
es
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tly
id
en
tifie
d
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
,
r
em
ain
s
r
elativ
ely
lo
w
f
o
r
b
o
th
ac
tiv
atio
n
f
u
n
ctio
n
s
at
s
m
aller
n
o
d
e
co
u
n
ts
b
u
t
im
p
r
o
v
es
as
h
id
d
e
n
n
o
d
es
in
cr
ea
s
e.
Fo
r
ex
am
p
le,
s
ig
m
o
id
ac
h
iev
es
a
p
r
ec
is
io
n
o
f
1
6
.
0
0
5
%
with
3
0
h
id
d
e
n
n
o
d
es,
wh
ile
R
eL
U
ac
h
iev
es 3
5
.
2
0
7
% p
r
ec
i
s
io
n
at
s
im
ilar
h
ig
h
er
n
o
d
e.
iii)
R
ec
all:
r
ec
all,
a
cr
itical
m
etr
ic
f
o
r
f
r
au
d
d
etec
tio
n
,
c
o
n
s
is
ten
tly
im
p
r
o
v
es
with
th
e
n
u
m
b
er
o
f
h
id
d
e
n
n
o
d
es.
T
h
e
s
ig
m
o
id
ac
tiv
atio
n
r
ea
ch
es
a
r
ec
all
o
f
8
5
.
8
1
1
%
with
8
0
h
id
d
en
n
o
d
es,
s
h
o
wca
s
in
g
its
r
o
b
u
s
tn
ess
in
id
en
tify
i
n
g
f
r
a
u
d
u
len
t
tr
an
s
ac
tio
n
s
.
R
eL
U
f
o
llo
ws
a
s
im
ilar
p
atter
n
b
u
t
lag
s
s
lig
h
tl
y
b
eh
in
d
i
n
lo
wer
c
o
n
f
ig
u
r
atio
n
s
.
I
n
o
th
er
wo
r
d
s
,
th
e
p
r
o
p
o
s
e
d
E
L
M
alg
o
r
ith
m
ac
h
iev
e
d
8
4
.
4
5
9
%
r
ec
all
at
7
0
h
id
d
en
n
o
d
es u
s
in
g
t
h
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
.
iv
)
Sp
ec
if
icity
:
s
p
ec
if
icity
r
em
ain
s
h
ig
h
ac
r
o
s
s
all
co
n
f
ig
u
r
atio
n
s
,
in
d
icatin
g
t
h
e
m
o
d
el’
s
a
b
ilit
y
to
co
r
r
ec
tly
class
if
y
leg
itima
te
tr
an
s
ac
tio
n
s
.
B
o
th
ac
tiv
atio
n
f
u
n
ctio
n
s
m
ain
tain
s
p
ec
if
icity
ab
o
v
e
9
8
% in
m
o
s
t
ca
s
es,
en
s
u
r
in
g
a
l
o
w
f
alse
-
p
o
s
itiv
e
r
ate.
v)
F
-
m
ea
s
u
r
e:
th
e
F
-
m
ea
s
u
r
e
b
alan
ce
s
p
r
ec
is
io
n
an
d
r
ec
all,
an
d
its
tr
en
d
s
r
ef
lect
th
e
tr
ad
e
-
o
f
f
s
b
etwe
e
n
th
ese
m
etr
ics.
Sig
m
o
id
ac
h
iev
es
h
ig
h
er
F
-
m
ea
s
u
r
e
v
alu
es
at
m
o
d
er
ate
h
id
d
en
n
o
d
e
co
u
n
ts
,
wh
ile
R
eL
U
clo
s
es th
e
g
ap
at
h
ig
h
er
co
u
n
ts
.
Ho
wev
er
,
th
e
h
ig
h
est F
-
m
ea
s
u
r
e
is
4
8
.
9
7
1
%,
o
b
tain
ed
u
s
in
g
R
eL
U
at
3
0
h
id
d
en
n
o
d
es.
v
i)
G
-
m
ea
n
:
as
a
co
m
b
in
ed
m
ea
s
u
r
e
o
f
r
ec
all
an
d
s
p
ec
if
icity
,
G
-
m
ea
n
h
ig
h
lig
h
ts
th
e
o
v
er
all
b
alan
ce
o
f
th
e
m
o
d
el’
s
p
er
f
o
r
m
an
ce
.
B
o
th
ac
tiv
atio
n
f
u
n
ctio
n
s
ex
h
ib
it
a
s
tead
y
in
cr
ea
s
e
in
G
-
m
ea
n
with
m
o
r
e
h
id
d
e
n
n
o
d
es,
r
ea
c
h
in
g
ab
o
v
e
9
0
%
in
o
p
tim
al
co
n
f
i
g
u
r
atio
n
s
.
I
n
ad
d
itio
n
,
th
e
h
ig
h
est
G
-
m
ea
n
r
esu
lt
is
9
2
.
0
6
8
%,
wh
er
e
it h
as
b
ee
n
o
b
tain
ed
u
s
in
g
s
ig
m
o
id
at
h
i
d
d
e
n
n
o
d
es
o
f
8
0
.
T
ab
le
1
.
T
h
e
ex
p
er
im
en
tal
r
es
u
lts
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
A
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
H
.
N
.
A
c
c
u
r
a
c
y
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
F
-
mea
su
r
e
(
%)
G
-
mea
n
(
%)
S
i
g
m
o
i
d
10
9
5
.
9
0
6
2
.
8
9
7
6
9
.
5
9
5
9
5
.
9
5
2
5
.
5
6
2
8
1
.
7
1
7
20
9
8
.
6
1
8
9
.
6
8
0
8
3
.
7
8
4
9
8
.
6
4
4
1
7
.
3
5
5
9
0
.
9
1
1
30
9
9
.
2
2
5
1
6
.
0
0
5
8
1
.
7
5
7
9
9
.
2
5
6
2
6
.
7
7
0
9
0
.
0
8
2
40
9
8
.
9
6
2
1
2
.
3
3
4
8
1
.
7
5
7
9
8
.
9
9
2
2
1
.
4
3
5
8
9
.
9
6
2
50
9
8
.
8
5
4
1
1
.
5
6
3
8
4
.
4
5
9
9
8
.
8
7
9
2
0
.
3
4
2
9
1
.
3
8
5
60
9
8
.
9
0
3
1
1
.
8
1
0
8
2
.
4
3
2
9
8
.
9
3
2
2
0
.
6
6
0
9
0
.
3
0
6
70
9
8
.
7
1
0
1
0
.
5
1
3
8
5
.
8
1
1
9
8
.
7
3
3
1
8
.
7
3
2
9
2
.
0
4
5
80
9
8
.
7
5
8
1
0
.
8
8
3
8
5
.
8
1
1
9
8
.
7
8
1
1
9
.
3
1
6
9
2
.
0
6
8
90
9
8
.
7
4
8
1
0
.
5
3
1
8
3
.
1
0
8
9
8
.
7
7
5
1
8
.
6
9
3
9
0
.
6
0
3
1
0
0
9
8
.
6
8
9
1
0
.
2
9
4
8
5
.
1
3
5
9
8
.
7
1
3
1
8
.
3
6
7
9
1
.
6
7
3
R
e
LU
10
9
7
.
3
1
2
4
.
2
5
7
6
7
.
5
6
8
9
7
.
3
6
3
8
.
0
1
0
8
1
.
1
0
9
20
9
7
.
3
0
5
4
.
7
0
8
7
5
.
6
7
6
9
7
.
3
4
2
8
.
8
6
4
8
5
.
8
2
8
30
9
9
.
7
1
0
3
5
.
2
0
7
8
0
.
4
0
5
9
9
.
7
4
3
4
8
.
9
7
1
8
9
.
5
5
4
40
9
9
.
3
0
2
1
7
.
1
5
5
7
9
.
0
5
4
9
9
.
3
3
8
2
8
.
1
9
3
8
8
.
6
1
7
50
9
9
.
2
2
4
1
6
.
2
5
2
8
3
.
7
8
4
9
9
.
2
5
1
2
7
.
2
2
3
9
1
.
1
9
0
60
9
8
.
3
3
2
7
.
9
1
0
8
1
.
0
8
1
9
8
.
3
6
2
1
4
.
4
1
4
8
9
.
3
0
5
70
9
8
.
5
5
0
9
.
3
2
1
8
4
.
4
5
9
9
8
.
5
7
4
1
6
.
7
9
0
9
1
.
2
4
4
80
9
9
.
1
7
8
1
5
.
3
7
5
8
3
.
1
0
8
9
9
.
2
0
6
2
5
.
9
4
9
9
0
.
8
0
1
90
9
8
.
9
8
6
1
2
.
8
3
6
8
3
.
7
8
4
9
9
.
0
1
3
2
2
.
2
6
2
9
1
.
0
8
1
1
0
0
9
9
.
2
4
6
1
6
.
6
6
7
8
3
.
7
8
4
9
9
.
2
7
3
2
7
.
8
0
3
9
1
.
2
0
0
T
ab
le
2
.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
v
alu
es f
o
r
t
h
e
h
ig
h
est ac
h
iev
e
d
r
esu
lts
A
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
M
e
t
r
i
c
h
i
g
h
l
i
g
h
t
s
T
r
u
e
p
o
s
i
t
i
v
e
s
T
r
u
e
n
e
g
a
t
i
v
e
s
F
a
l
se
p
o
si
t
i
v
e
s
F
a
l
se
n
e
g
a
t
i
v
e
s
S
i
g
m
o
i
d
H
i
g
h
e
s
t
r
e
c
a
l
l
a
n
d
G
-
me
a
n
1
2
7
8
4
2
5
5
1
0
4
0
21
R
e
LU
H
i
g
h
e
s
t
a
c
c
u
r
a
c
y
,
p
r
e
c
i
si
o
n
,
sp
e
c
i
f
i
c
i
t
y
,
a
n
d
F
-
m
e
a
s
u
r
e
1
1
9
8
5
0
7
6
2
1
9
29
Acc
o
r
d
in
g
to
th
e
s
tatis
tical
an
aly
s
is
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
m
o
d
el,
th
e
ex
p
er
im
en
tal
r
esu
lts
d
em
o
n
s
tr
ate
n
o
tab
le
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
v
ar
io
u
s
m
etr
ics
f
o
r
b
o
th
ac
tiv
atio
n
f
u
n
ctio
n
s
.
T
ab
le
3
p
r
esen
ts
th
e
s
tatis
t
ical
an
aly
s
is
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
in
d
etec
tin
g
cr
e
d
it
ca
r
d
f
r
au
d
.
I
n
ter
m
s
o
f
a
cc
u
r
ac
y
,
th
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
ac
h
iev
e
d
a
m
ea
n
ac
cu
r
ac
y
o
f
9
8
.
7
1
5
%
,
ac
co
m
p
an
ied
b
y
a
s
m
all
s
ta
n
d
ar
d
d
ev
iatio
n
o
f
0
.
8
3
6
%,
r
e
f
lectin
g
co
n
s
is
ten
t
an
d
r
eliab
le
p
er
f
o
r
m
an
ce
ac
r
o
s
s
tr
ials
.
T
h
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
cred
it c
a
r
d
fr
a
u
d
d
etec
tio
n
w
ith
s
yn
th
etic
min
o
r
ity
o
ve
r
…
(
I
ma
n
K
a
d
h
im
A
jla
n
)
4757
p
er
f
o
r
m
ed
s
lig
h
tly
lo
wer
,
with
a
m
ea
n
ac
cu
r
ac
y
o
f
9
8
.
5
3
7
%
an
d
a
s
tan
d
ar
d
d
ev
iatio
n
o
f
0
.
9
4
1
%,
in
d
icatin
g
m
ar
g
in
ally
h
ig
h
e
r
v
ar
iab
ilit
y
co
m
p
ar
ed
to
R
eL
U.
Pre
cisi
o
n
r
esu
lts
h
ig
h
lig
h
t
s
o
m
e
ch
alle
n
g
es
in
id
en
tify
i
n
g
f
r
au
d
u
le
n
t
tr
a
n
s
ac
tio
n
s
.
T
h
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
ac
h
iev
ed
a
n
a
v
er
ag
e
p
r
ec
is
io
n
o
f
1
3
.
9
6
9
%,
with
a
r
elativ
ely
h
ig
h
s
tan
d
a
r
d
d
e
v
iatio
n
o
f
8
.
9
1
5
%,
s
u
g
g
esti
n
g
v
a
r
iab
ilit
y
in
its
ab
ilit
y
to
m
in
im
ize
f
alse
p
o
s
itiv
es.
I
n
co
n
tr
ast,
th
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
r
ec
o
r
d
ed
a
lo
wer
av
er
ag
e
p
r
ec
is
io
n
o
f
1
0
.
6
5
1
%
b
u
t
ex
h
i
b
ited
g
r
ea
ter
co
n
s
is
ten
cy
,
as e
v
id
en
ce
d
b
y
its
s
m
aller
s
tan
d
ar
d
d
e
v
iatio
n
o
f
3
.
2
4
8
%.
R
ec
all,
a
cr
itical
m
etr
ic
f
o
r
d
etec
tin
g
f
r
a
u
d
u
len
t
ca
s
es,
s
h
o
wed
s
tr
o
n
g
p
er
f
o
r
m
an
c
e
f
o
r
b
o
th
ac
tiv
atio
n
f
u
n
ctio
n
s
.
T
h
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
ac
h
iev
e
d
an
a
v
er
ag
e
r
ec
all
o
f
8
0
.
2
7
0
%,
with
a
s
tan
d
ar
d
d
ev
iatio
n
o
f
5
.
2
4
2
%,
in
d
ica
tin
g
its
ef
f
ec
tiv
en
ess
in
i
d
e
n
tify
in
g
f
r
au
d
u
len
t
t
r
an
s
ac
tio
n
s
.
T
h
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
p
er
f
o
r
m
e
d
s
lig
h
tly
b
etter
,
with
an
av
er
a
g
e
r
ec
all
o
f
8
2
.
3
6
5
%
an
d
a
s
tan
d
ar
d
d
ev
iatio
n
o
f
4
.
7
4
0
%,
r
ef
lectin
g
its
r
o
b
u
s
tn
ess
an
d
s
lig
h
tly
s
u
p
er
io
r
d
etec
tio
n
ca
p
ab
ilit
y
.
Sp
ec
if
icity
r
em
ain
ed
co
n
s
is
ten
tly
h
ig
h
ac
r
o
s
s
b
o
th
ac
tiv
atio
n
f
u
n
ctio
n
s
,
d
em
o
n
s
tr
atin
g
th
e
m
o
d
el'
s
ab
ilit
y
to
ac
cu
r
ately
class
if
y
leg
itima
te
tr
an
s
ac
tio
n
s
.
T
h
e
R
eL
U
ac
ti
v
atio
n
f
u
n
cti
o
n
ac
h
iev
e
d
a
m
ea
n
s
p
ec
if
icity
o
f
9
8
.
7
4
7
%,
with
a
s
tan
d
ar
d
d
ev
iatio
n
o
f
0
.
8
3
1
%,
wh
ile
th
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
r
ec
o
r
d
e
d
a
s
lig
h
tly
lo
wer
m
ea
n
s
p
ec
if
icity
o
f
9
8
.
5
6
5
%,
with
a
s
tan
d
ar
d
d
e
v
iatio
n
o
f
0
.
9
3
5
%.
B
o
th
r
esu
lts
u
n
d
er
s
co
r
e
th
e
m
o
d
el'
s
s
tr
o
n
g
p
e
r
f
o
r
m
an
ce
in
m
in
im
izin
g
f
alse p
o
s
itiv
es.
T
h
e
F
-
m
ea
s
u
r
e,
wh
ich
b
alan
c
es
p
r
ec
is
io
n
an
d
r
ec
all,
r
ev
ea
l
ed
s
o
m
e
tr
ad
e
-
o
f
f
s
.
T
h
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
ac
h
ie
v
ed
a
m
ea
n
F
-
m
ea
s
u
r
e
o
f
2
2
.
8
4
8
%,
b
u
t
with
a
r
elativ
ely
h
ig
h
s
tan
d
ar
d
d
ev
iatio
n
o
f
1
1
.
9
6
4
%
,
in
d
icatin
g
v
ar
iab
ilit
y
in
b
al
an
cin
g
th
ese
two
m
etr
ics.
T
h
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
c
tio
n
,
with
a
m
ea
n
F
-
m
ea
s
u
r
e
o
f
1
8
.
7
2
3
%
an
d
a
s
m
aller
s
tan
d
ar
d
d
e
v
iatio
n
o
f
5
.
3
1
3
%,
s
h
o
wed
less
v
a
r
iab
ilit
y
b
u
t
a
s
lig
h
tly
less
o
p
tim
al
b
ala
n
ce
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
all.
Fin
ally
,
th
e
G
-
m
ea
n
,
w
h
ich
c
o
m
b
in
es
r
ec
all
an
d
s
p
ec
if
icity
to
ass
es
s
o
v
er
all
b
alan
ce
,
d
e
m
o
n
s
tr
ated
s
tr
o
n
g
p
er
f
o
r
m
an
ce
f
o
r
b
o
t
h
ac
tiv
atio
n
f
u
n
ctio
n
s
.
T
h
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
ac
h
iev
ed
a
m
ea
n
G
-
m
ea
n
o
f
8
8
.
9
9
3
%,
with
a
s
tan
d
ar
d
d
ev
iatio
n
o
f
3
.
2
5
1
%,
in
d
icatin
g
a
well
-
b
alan
ce
d
p
er
f
o
r
m
an
ce
.
T
h
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ct
io
n
ac
h
iev
ed
a
s
lig
h
tly
h
ig
h
e
r
m
ea
n
G
-
m
ea
n
o
f
9
0
.
0
7
5
%,
with
a
s
tan
d
ar
d
d
ev
i
atio
n
o
f
3
.
0
3
7
%,
r
ef
lectin
g
its
s
lig
h
tly
b
etter
o
v
er
all
b
alan
ce
b
etwe
en
d
etec
tin
g
f
r
au
d
u
le
n
t tr
an
s
ac
tio
n
s
an
d
m
i
n
im
izin
g
f
alse p
o
s
itiv
es.
Fig
u
r
e
2
.
T
h
e
h
ig
h
est o
b
tain
ed
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
alg
o
r
ith
m
f
o
r
d
etec
tin
g
cr
ed
it
ca
r
d
f
r
a
u
d
T
ab
le
3
.
T
h
e
s
tatis
tical
an
aly
s
i
s
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
e
l
Ev
a
l
u
a
t
i
o
n
mea
su
r
e
me
n
t
s
R
e
LU
S
i
g
m
o
i
d
M
e
a
n
(
%)
S
TD
(
%)
M
e
a
n
(
%)
S
TD
(
%)
A
c
c
u
r
a
c
y
9
8
.
7
1
5
0
.
8
3
6
9
8
.
5
3
7
0
.
9
4
1
P
r
e
c
i
s
i
o
n
1
3
.
9
6
9
8
.
9
1
5
1
0
.
6
5
1
3
.
2
4
8
R
e
c
a
l
l
8
0
.
2
7
0
5
.
2
4
2
8
2
.
3
6
5
4
.
7
4
0
S
p
e
c
i
f
i
c
i
t
y
9
8
.
7
4
7
0
.
8
3
1
9
8
.
5
6
5
0
.
9
3
5
F
-
mea
su
r
e
2
2
.
8
4
8
1
1
.
9
6
4
1
8
.
7
2
3
5
.
3
1
3
G
-
me
an
8
8
.
9
9
3
3
.
2
5
1
9
0
.
0
7
5
3
.
0
3
7
5.
DIS
CU
SS
I
O
N
T
h
e
ex
p
er
im
en
tal
r
e
s
u
l
t
s
d
em
o
n
s
t
r
a
t
e
t
h
at
t
h
e
p
r
o
p
o
s
e
d
E
L
M
m
o
d
e
l
e
f
f
e
c
t
i
v
el
y
a
d
d
r
e
s
s
es
t
h
e
c
h
a
l
l
e
n
g
e
s
o
f
c
r
e
d
i
t
c
a
r
d
f
r
a
u
d
d
e
t
e
c
t
i
o
n
i
n
i
m
b
a
l
a
n
c
e
d
d
a
t
as
et
s
.
A
p
p
l
y
i
n
g
t
h
e
S
M
O
T
E
t
e
c
h
n
i
q
u
e
t
o
t
h
e
t
r
a
i
n
i
n
g
s
e
t
h
e
l
p
s
b
a
l
a
n
c
e
cl
a
s
s
r
e
p
r
es
e
n
t
a
ti
o
n
,
w
h
i
c
h
s
i
g
n
i
f
i
c
a
n
tl
y
c
o
n
t
r
i
b
u
t
e
s
t
o
i
m
p
r
o
v
e
d
r
e
ca
l
l
a
n
d
s
p
e
ci
f
i
c
it
y
.
T
h
e
c
h
o
i
c
e
o
f
a
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
a
l
s
o
i
n
f
l
u
e
n
c
e
s
l
e
a
r
n
i
n
g
b
e
h
a
v
i
o
r
,
w
i
t
h
t
h
e
s
i
g
m
o
i
d
f
u
n
c
t
i
o
n
y
i
e
l
d
i
n
g
s
t
r
o
n
g
e
r
i
n
i
t
i
al
p
e
r
f
o
r
m
a
n
c
e
a
n
d
t
h
e
R
e
L
U
f
u
n
c
t
i
o
n
s
h
o
w
i
n
g
b
e
t
te
r
s
c
a
l
a
b
i
l
it
y
.
A
d
j
u
s
t
i
n
g
t
h
e
n
u
m
b
e
r
o
f
h
i
d
d
e
n
n
o
d
e
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
7
4
9
-
4
7
6
2
4758
r
e
v
e
a
l
e
d
t
h
a
t
m
o
d
e
l
c
o
m
p
l
e
x
i
ty
p
l
a
y
s
a
k
e
y
r
o
l
e
i
n
a
c
h
i
e
v
i
n
g
o
p
t
i
m
a
l
r
es
u
l
ts
.
I
n
c
r
e
as
i
n
g
th
e
n
u
m
b
e
r
o
f
h
i
d
d
e
n
n
o
d
e
s
e
n
h
a
n
c
e
s
t
h
e
m
o
d
el
’
s
a
b
i
l
i
t
y
t
o
c
a
p
t
u
r
e
c
o
m
p
l
e
x
r
e
l
a
t
i
o
n
s
h
i
p
s
,
i
m
p
r
o
v
i
n
g
a
cc
u
r
a
c
y
a
n
d
o
v
e
r
a
l
l
p
e
r
f
o
r
m
a
n
c
e
.
H
o
w
e
v
e
r
,
e
x
c
e
s
s
i
v
e
c
o
m
p
l
e
x
i
t
y
c
a
n
l
e
a
d
t
o
d
i
m
i
n
i
s
h
i
n
g
r
e
t
u
r
n
s
o
r
o
v
e
r
f
i
t
t
i
n
g
,
h
i
g
h
l
i
g
h
t
i
n
g
t
h
e
i
m
p
o
r
t
a
n
c
e
o
f
c
a
r
e
f
u
l
h
y
p
e
r
p
a
r
a
m
e
t
e
r
t
u
n
i
n
g
.
S
t
a
ti
s
t
i
c
al
a
n
a
l
y
s
is
f
u
r
t
h
e
r
s
u
p
p
o
r
t
s
t
h
e
e
f
f
e
c
t
iv
e
n
e
s
s
o
f
t
h
e
E
L
M
m
o
d
e
l
u
s
i
n
g
b
o
t
h
R
e
L
U
a
n
d
s
i
g
m
o
i
d
a
c
t
i
v
at
i
o
n
f
u
n
c
t
i
o
n
s
.
B
o
t
h
f
u
n
c
t
i
o
n
s
a
c
h
i
e
v
e
d
h
i
g
h
a
c
c
u
r
a
c
y
a
n
d
s
p
e
c
i
f
i
ci
t
y
,
c
o
n
s
i
s
te
n
t
l
y
e
x
c
e
ed
i
n
g
9
8
%
,
w
h
i
c
h
i
n
d
i
c
a
t
es
th
e
m
o
d
e
l
’
s
s
t
r
o
n
g
a
b
i
l
it
y
t
o
i
d
e
n
t
i
f
y
l
e
g
i
t
i
m
a
t
e
t
r
a
n
s
a
c
ti
o
n
s
a
n
d
m
a
i
n
t
ai
n
a
l
o
w
f
a
ls
e
-
p
o
s
it
i
v
e
r
at
e
,
w
h
ic
h
is
a
n
e
s
s
e
n
ti
a
l
as
p
e
c
t
o
f
r
e
l
ia
b
l
e
f
r
a
u
d
d
e
t
e
c
t
i
o
n
.
I
n
t
e
r
m
s
o
f
r
e
c
a
l
l
,
w
h
i
c
h
m
e
a
s
u
r
es
t
h
e
d
e
t
e
c
t
i
o
n
o
f
f
r
a
u
d
u
l
e
n
t
t
r
a
n
s
a
c
t
i
o
n
s
,
t
h
e
s
i
g
m
o
i
d
f
u
n
c
t
i
o
n
p
e
r
f
o
r
m
e
d
s
l
i
g
h
t
l
y
b
e
t
t
e
r
,
a
c
h
i
e
v
i
n
g
a
n
a
v
e
r
a
g
e
r
ec
a
l
l
o
f
8
2
.
3
6
5
%
c
o
m
p
a
r
e
d
t
o
8
0
.
2
7
0
%
f
o
r
R
e
L
U
.
T
h
is
s
u
g
g
e
s
ts
t
h
a
t
s
i
g
m
o
i
d
is
m
o
r
e
s
u
i
t
a
b
l
e
w
h
e
n
m
i
n
i
m
i
z
in
g
f
a
l
s
e
n
e
g
a
t
i
v
es
is
c
r
i
t
ic
a
l
.
H
o
w
e
v
e
r
,
p
r
e
c
i
s
i
o
n
v
a
l
u
e
s
f
o
r
b
o
t
h
a
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
s
r
e
m
a
i
n
e
d
r
e
l
at
i
v
e
l
y
l
o
w
,
w
i
t
h
R
e
L
U
a
c
h
i
e
v
i
n
g
1
3
.
9
6
9
%
a
n
d
s
i
g
m
o
i
d
1
0
.
6
5
1
%
o
n
a
v
e
r
a
g
e
.
T
h
i
s
o
u
t
c
o
m
e
r
e
f
l
e
c
ts
t
h
e
i
n
h
e
r
e
n
t
d
i
f
f
i
c
u
l
t
y
o
f
d
et
e
c
ti
n
g
f
r
a
u
d
u
l
e
n
t
t
r
a
n
s
a
c
t
i
o
n
s
i
n
d
at
a
s
e
ts
d
o
m
in
a
t
e
d
b
y
l
e
g
it
i
m
a
te
o
n
e
s
a
n
d
u
n
d
e
r
s
c
o
r
e
s
t
h
e
p
e
r
s
is
t
e
n
t
c
h
al
l
e
n
g
e
o
f
a
c
h
i
e
v
i
n
g
h
i
g
h
p
r
e
c
i
s
i
o
n
i
n
h
i
g
h
l
y
i
m
b
a
la
n
c
e
d
f
i
n
a
n
c
i
a
l
d
a
ta
.
T
h
e
F
-
m
e
as
u
r
e
,
s
h
o
w
e
d
t
h
a
t
R
e
L
U
h
a
d
a
s
l
i
g
h
t
l
y
b
et
t
e
r
o
v
e
r
a
l
l
b
a
l
a
n
c
e
i
n
s
o
m
e
c
o
n
f
i
g
u
r
a
t
i
o
n
s
.
H
o
w
e
v
e
r
,
w
h
e
n
c
o
n
s
i
d
e
r
i
n
g
th
e
G
-
m
e
a
n
,
w
h
i
c
h
a
c
c
o
u
n
ts
f
o
r
b
o
t
h
r
e
c
a
l
l
a
n
d
s
p
e
c
i
f
i
ci
t
y
,
s
i
g
m
o
i
d
p
e
r
f
o
r
m
e
d
s
l
i
g
h
tl
y
b
e
t
t
e
r
w
i
t
h
a
m
e
a
n
G
-
m
e
a
n
o
f
9
0
.
0
7
5
%
c
o
m
p
a
r
e
d
t
o
R
e
L
U'
s
8
8
.
9
9
3
%
.
T
h
i
s
in
d
i
c
a
t
e
s
t
h
a
t
w
h
i
l
e
s
i
g
m
o
i
d
m
a
y
n
o
t
e
x
c
el
i
n
p
r
e
c
is
i
o
n
,
i
t
p
r
o
v
i
d
es
a
m
o
r
e
b
a
l
a
n
ce
d
o
v
e
r
a
l
l
p
e
r
f
o
r
m
a
n
c
e
a
c
r
o
s
s
m
e
t
r
i
c
s
.
V
a
r
i
a
b
i
l
it
y
i
n
p
e
r
f
o
r
m
a
n
c
e
w
a
s
a
n
o
t
h
e
r
i
m
p
o
r
t
a
n
t
a
s
p
e
c
t
t
o
c
o
n
s
i
d
e
r
.
T
h
e
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
s
f
o
r
p
r
e
c
i
s
i
o
n
a
n
d
F
-
m
e
a
s
u
r
e
w
e
r
e
h
i
g
h
e
r
f
o
r
R
e
L
U
,
s
u
g
g
e
s
t
i
n
g
g
r
e
a
t
e
r
i
n
c
o
n
s
i
s
t
e
n
c
y
i
n
i
ts
p
e
r
f
o
r
m
a
n
c
e
a
c
r
o
s
s
t
r
ia
l
s
.
I
n
c
o
n
t
r
a
s
t
,
S
i
g
m
o
id
d
e
m
o
n
s
t
r
a
t
e
d
m
o
r
e
s
ta
b
l
e
r
es
u
l
t
s
,
m
a
k
i
n
g
i
t
p
o
t
e
n
t
ia
l
l
y
m
o
r
e
r
e
l
i
a
b
l
e
f
o
r
c
o
n
s
i
s
te
n
t
f
r
a
u
d
d
e
t
e
c
t
i
o
n
i
n
d
i
f
f
e
r
e
n
t
s
c
e
n
a
r
i
o
s
.
B
a
s
e
d
o
n
t
h
e
e
x
p
e
r
i
m
e
n
t
a
l
r
es
u
lt
s
,
i
t
d
e
m
o
n
s
t
r
a
te
d
t
h
a
t
t
h
e
c
h
o
i
c
e
o
f
a
c
ti
v
a
t
io
n
f
u
n
c
t
i
o
n
s
h
o
u
ld
d
e
p
e
n
d
o
n
t
h
e
s
p
e
c
i
f
i
c
p
r
i
o
r
i
t
ie
s
o
f
t
h
e
f
r
a
u
d
d
e
t
e
c
t
i
o
n
s
y
s
t
em
.
S
i
g
m
o
i
d
i
s
m
o
r
e
s
u
i
t
a
b
l
e
f
o
r
a
p
p
l
i
c
a
t
i
o
n
s
t
h
a
t
p
r
i
o
r
i
t
i
z
e
r
e
c
al
l
a
n
d
o
v
e
r
a
l
l
b
a
l
a
n
c
e
,
a
s
r
e
f
l
e
ct
e
d
i
n
i
ts
s
u
p
e
r
i
o
r
r
e
c
a
l
l
a
n
d
G
-
m
e
a
n
.
O
n
t
h
e
o
t
h
e
r
h
a
n
d
,
R
e
L
U
m
i
g
h
t
b
e
p
r
e
f
e
r
r
e
d
i
n
s
ce
n
a
r
i
o
s
w
h
e
r
e
p
r
e
ci
s
i
o
n
a
n
d
c
o
n
s
is
t
e
n
c
y
a
r
e
m
o
r
e
c
r
i
t
i
ca
l
.
I
n
g
e
n
e
r
al
,
t
h
e
p
r
o
p
o
s
e
d
E
L
M
m
o
d
e
l
d
e
m
o
n
s
t
r
at
e
s
s
t
r
o
n
g
p
o
t
e
n
t
i
a
l
f
o
r
r
e
al
-
w
o
r
l
d
f
r
a
u
d
d
ete
c
t
i
o
n
a
p
p
li
c
a
ti
o
n
s
.
T
h
e
i
n
s
i
g
h
t
s
g
ai
n
e
d
f
r
o
m
t
h
e
e
x
p
e
r
i
m
e
n
t
a
l
r
e
s
u
lt
s
p
r
o
v
i
d
e
v
a
l
u
a
b
l
e
g
u
i
d
a
n
c
e
f
o
r
o
p
t
i
m
iz
i
n
g
E
L
M
c
o
n
f
i
g
u
r
a
ti
o
n
s
a
n
d
a
d
d
r
e
s
s
i
n
g
t
h
e
u
n
i
q
u
e
c
h
a
l
l
e
n
g
e
s
p
o
s
e
d
b
y
i
m
b
a
l
a
n
c
ed
d
a
t
a
s
e
ts
.
H
o
w
e
v
e
r
,
al
t
h
o
u
g
h
t
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
a
c
h
i
e
v
e
s
h
i
g
h
r
e
s
u
l
ts
i
n
t
e
r
m
s
o
f
a
c
c
u
r
a
c
y
,
r
e
c
a
l
l
,
s
p
e
c
i
f
i
ci
ty
,
a
n
d
G
-
m
e
a
n
,
i
t
s
t
i
l
l
s
u
f
f
e
r
s
f
r
o
m
l
o
w
p
r
e
c
i
s
i
o
n
a
n
d
F
-
m
e
a
s
u
r
e
,
w
h
i
c
h
a
r
e
c
o
n
s
i
d
e
r
e
d
l
i
m
i
t
a
ti
o
n
s
o
f
t
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
i
n
d
e
t
e
c
ti
n
g
c
r
e
d
i
t
c
a
r
d
f
r
a
u
d
.
T
h
e
u
s
e
o
f
S
M
O
T
E
h
a
d
a
s
i
g
n
i
f
i
c
a
n
t
i
m
p
a
c
t
o
n
t
h
e
m
o
d
e
l
’
s
a
b
i
l
it
y
t
o
d
e
t
e
c
t
f
r
a
u
d
u
l
e
n
t
t
r
a
n
s
a
c
ti
o
n
s
.
B
y
g
e
n
e
r
a
ti
n
g
s
y
n
t
h
et
i
c
s
a
m
p
le
s
f
o
r
t
h
e
m
i
n
o
r
i
t
y
(
i
.
e
.
,
f
r
a
u
d
)
c
l
as
s
,
S
MO
T
E
e
f
f
e
ct
i
v
e
l
y
b
a
l
a
n
c
e
d
t
h
e
d
a
t
as
e
t,
e
n
a
b
l
i
n
g
t
h
e
E
L
M
al
g
o
r
i
t
h
m
t
o
b
e
t
t
e
r
le
a
r
n
a
n
d
r
e
c
o
g
n
i
z
e
p
a
t
t
e
r
n
s
a
s
s
o
c
i
at
e
d
w
it
h
f
r
a
u
d
u
l
e
n
t
b
e
h
a
v
i
o
r
.
T
h
i
s
r
es
u
l
t
e
d
i
n
n
o
t
a
b
l
e
i
m
p
r
o
v
e
m
e
n
t
s
i
n
r
ec
a
l
l
,
w
h
i
c
h
is
a
cr
u
c
i
a
l
m
e
t
r
i
c
f
o
r
f
r
a
u
d
d
e
t
e
c
ti
o
n
s
i
n
c
e
m
i
n
i
m
i
z
i
n
g
f
a
l
s
e
n
e
g
at
i
v
es
is
o
f
p
r
i
m
a
r
y
im
p
o
r
t
a
n
c
e
.
A
s
c
o
m
m
o
n
l
y
o
b
s
er
v
e
d
w
i
t
h
o
v
e
r
s
a
m
p
l
i
n
g
m
e
t
h
o
d
s
,
t
h
is
i
m
p
r
o
v
e
m
e
n
t
c
a
m
e
w
i
t
h
a
s
l
i
g
h
t
r
e
d
u
c
t
i
o
n
in
s
p
e
c
i
f
i
c
it
y
,
a
s
t
h
e
m
o
d
e
l
b
ec
a
m
e
s
o
m
e
w
h
a
t
m
o
r
e
l
i
k
el
y
to
c
l
a
s
s
i
f
y
l
e
g
i
t
i
m
a
t
e
t
r
a
n
s
a
c
ti
o
n
s
as
f
r
a
u
d
u
l
e
n
t
,
l
e
ad
i
n
g
t
o
a
n
i
n
c
r
e
as
e
i
n
f
a
ls
e
p
o
s
i
t
i
v
es
.
Ne
v
e
r
t
h
e
l
es
s
,
t
h
e
o
v
e
r
a
l
l
p
e
r
f
o
r
m
a
n
c
e
,
r
e
f
l
e
c
t
e
d
i
n
m
e
t
r
i
c
s
s
u
c
h
a
s
G
-
m
e
a
n
a
n
d
F
-
m
e
a
s
u
r
e
,
i
m
p
r
o
v
ed
,
s
u
g
g
e
s
t
i
n
g
t
h
a
t
S
M
O
T
E
e
n
h
a
n
c
e
d
t
h
e
m
o
d
e
l
’
s
r
o
b
u
s
t
n
e
s
s
w
h
e
n
d
e
a
l
i
n
g
w
i
th
i
m
b
a
l
a
n
c
e
d
d
a
t
a
.
T
h
e
s
e
f
i
n
d
i
n
g
s
h
i
g
h
l
i
g
h
t
t
h
e
i
m
p
o
r
t
a
n
ce
o
f
a
p
p
l
y
i
n
g
d
a
t
a
b
a
l
a
n
c
i
n
g
t
e
c
h
n
i
q
u
es
i
n
f
r
a
u
d
d
e
t
e
c
t
i
o
n
t
o
e
n
s
u
r
e
t
h
a
t
m
i
n
o
r
i
t
y
c
l
as
s
p
at
t
e
r
n
s
a
r
e
e
f
f
ec
t
i
v
el
y
c
a
p
t
u
r
e
d
d
u
r
i
n
g
t
h
e
t
r
a
i
n
i
n
g
p
r
o
c
e
s
s
.
F
u
r
t
h
e
r
m
o
r
e
,
t
h
e
p
r
o
p
o
s
e
d
E
L
M
a
l
g
o
r
i
t
h
m
h
a
s
b
e
e
n
c
o
m
p
a
r
e
d
w
i
t
h
t
h
e
s
ta
t
e
-
of
-
t
h
e
-
a
r
t
m
e
t
h
o
d
s
f
o
r
d
e
t
e
c
t
i
n
g
c
r
e
d
i
t
c
a
r
d
f
r
a
u
d
i
n
te
r
m
s
o
f
a
cc
u
r
a
c
y
.
T
a
b
l
e
4
p
r
e
s
e
n
t
s
t
h
e
c
o
m
p
a
r
a
ti
v
e
p
e
r
f
o
r
m
an
c
e
o
f
t
h
e
p
r
o
p
o
s
e
d
E
L
M
a
l
g
o
r
i
t
h
m
a
g
a
i
n
s
t
s
t
a
t
e
-
of
-
t
h
e
-
a
r
t
m
e
t
h
o
d
s
f
o
r
c
r
e
d
i
t
c
a
r
d
f
r
a
u
d
d
e
t
e
c
t
i
o
n
.
T
h
e
p
r
o
p
o
s
e
d
E
L
M
a
l
g
o
r
i
t
h
m
d
e
m
o
n
s
t
r
a
t
es
a
s
i
g
n
i
f
i
ca
n
t
i
m
p
r
o
v
e
m
e
n
t
i
n
d
e
t
e
c
ti
n
g
c
r
e
d
i
t
c
a
r
d
f
r
a
u
d
c
o
m
p
a
r
e
d
t
o
s
t
at
e
-
of
-
t
h
e
-
a
r
t
m
e
t
h
o
d
s
.
W
h
i
l
e
t
r
a
d
it
i
o
n
a
l
a
p
p
r
o
a
c
h
e
s
li
k
e
L
R
[
7
]
an
d
R
F
[
3
7
]
ac
h
iev
ed
m
o
d
er
ate
ac
cu
r
ac
y
lev
els
u
p
to
9
4
.
4
2
5
an
d
8
2
.
5
%,
r
esp
ec
tiv
ely
.
Me
a
n
wh
ile,
th
e
p
r
o
p
o
s
ed
E
L
M
s
u
r
p
ass
es
th
em
with
an
im
p
r
ess
iv
e
ac
cu
r
ac
y
o
f
9
9
.
7
1
0
%.
I
n
ad
d
itio
n
,
L
STM
n
etwo
r
k
s
[
3
8
]
an
d
ANN
[
3
9
]
s
h
o
w
lo
wer
p
er
f
o
r
m
a
n
ce
,
ac
h
i
ev
in
g
9
0
an
d
7
0
%,
r
esp
ec
tiv
ely
,
in
d
icatin
g
t
h
eir
li
m
ited
ef
f
ec
tiv
e
n
ess
in
th
is
co
n
tex
t.
No
tab
ly
,
th
e
s
tan
d
ar
d
E
L
M
im
p
lem
en
tatio
n
ac
h
iev
ed
8
7
.
3
2
%
[
4
0
]
,
h
ig
h
li
g
h
tin
g
th
e
e
n
h
an
ce
m
e
n
ts
m
ad
e
in
th
e
p
r
o
p
o
s
ed
E
L
M
al
g
o
r
ith
m
.
T
h
e
n
ea
r
-
p
er
f
ec
t
ac
cu
r
ac
y
ac
h
iev
ed
b
y
th
e
p
r
o
p
o
s
ed
m
eth
o
d
u
n
d
er
s
co
r
es
its
r
o
b
u
s
tn
ess
an
d
p
o
ten
tial
as
a
s
u
p
er
io
r
to
o
l
f
o
r
cr
e
d
it c
ar
d
f
r
au
d
d
etec
tio
n
,
o
u
tp
er
f
o
r
m
in
g
co
n
v
en
tio
n
al
a
n
d
s
tate
-
of
-
th
e
-
ar
t m
eth
o
d
s
ac
r
o
s
s
th
e
b
o
ar
d
.
T
h
e
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
E
L
M
m
o
d
el
is
attr
ib
u
ted
to
s
ev
er
al
k
ey
f
ac
to
r
s
.
First,
th
e
d
y
n
am
ic
a
d
ju
s
tm
en
t
o
f
h
id
d
en
n
o
d
es
allo
wed
f
o
r
f
in
d
i
n
g
an
o
p
tim
al
n
etwo
r
k
s
ize
(
ar
o
u
n
d
3
0
–
4
0
n
o
d
es)
th
at
b
alan
ce
s
lear
n
i
n
g
ca
p
ac
i
ty
an
d
p
r
e
v
en
ts
o
v
e
r
f
itti
n
g
.
Seco
n
d
,
th
e
in
teg
r
atio
n
o
f
S
MO
T
E
ef
f
ec
tiv
ely
h
an
d
led
th
e
s
ev
er
e
class
im
b
alan
ce
b
y
g
e
n
er
atin
g
s
y
n
th
etic
m
in
o
r
ity
s
am
p
les,
im
p
r
o
v
in
g
th
e
m
o
d
el’
s
ab
ilit
y
to
d
etec
t
f
r
au
d
u
len
t
tr
an
s
ac
tio
n
s
.
L
astl
y
,
co
m
p
ar
in
g
ac
tiv
ati
o
n
f
u
n
ctio
n
s
r
ev
ea
led
th
at
R
eL
U’
s
co
m
p
u
tatio
n
al
ef
f
icien
cy
an
d
ab
ilit
y
t
o
m
itig
ate
v
an
is
h
in
g
g
r
ad
ien
ts
f
u
r
t
h
er
en
h
a
n
ce
d
ac
cu
r
ac
y
.
T
h
ese
c
o
m
b
in
e
d
im
p
r
o
v
em
e
n
ts
en
ab
le
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
to
o
u
tp
e
r
f
o
r
m
tr
ad
itio
n
al
m
o
d
els
b
y
b
etter
ca
p
tu
r
in
g
co
m
p
lex
f
r
au
d
p
atter
n
s
an
d
m
ain
tain
in
g
h
ig
h
d
etec
tio
n
ac
cu
r
ac
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.