I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
,
p
p
.
1
0
2
2
~1
0
29
I
SS
N:
2
5
02
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs.v
2
1
.i
2
.
p
p
1
0
2
2
-
10
29
1022
J
o
ur
na
l
ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
A hybrid o
f
CN
N
a
nd LST
M
m
etho
ds for secu
ring
w
eb
a
pplica
tion a
g
a
inst cros
s
-
site
scrip
t
ing
attac
k
Ra
ed
Wa
heed
K
a
dh
i
m
1
,
M
et
ha
q T
a
lib
G
a
a
t
a
2
1
In
f
o
rm
a
ti
c
s In
stit
u
te f
o
r
P
o
stg
ra
d
u
a
te
S
tu
d
ies
/Iraq
i
Co
m
m
issi
o
n
f
o
r
Co
m
p
u
ter an
d
I
n
f
o
rm
a
ti
c
s,
Ira
q
1,
2
Co
m
p
u
ter S
c
ien
c
e
De
p
a
rtm
e
n
t,
Un
iv
e
rsity
o
f
M
u
sta
n
siriy
a
h
,
Ba
g
h
d
a
d
,
Ira
q
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 6
,
2
0
2
0
R
ev
i
s
ed
Sep
7
,
2
0
2
0
A
cc
ep
ted
Oct
2
1
,
2
0
2
0
Cro
ss
-
site
sc
rip
ti
n
g
(
X
S
S
)
is
t
o
d
a
y
o
n
e
o
f
th
e
b
ig
g
e
st
th
re
a
tt
h
a
t
c
o
u
ld
targ
e
ti
n
g
th
e
W
e
b
a
p
p
li
c
a
ti
o
n
.
B
a
se
d
o
n
stu
d
y
p
u
b
li
sh
e
d
b
y
th
e
o
p
e
n
w
e
b
a
p
p
li
c
a
ti
o
n
s
se
c
u
rit
y
p
ro
je
c
t
(O
WA
S
P
),
X
S
S
v
u
l
n
e
ra
b
il
it
y
h
a
s
b
e
e
n
p
re
se
n
t
a
m
o
n
g
th
e
TOP
1
0
W
e
b
a
p
p
li
c
a
ti
o
n
v
u
ln
e
ra
b
il
i
ti
e
s.
S
ti
ll
,
a
n
im
p
o
rtan
t
se
c
u
rit
y
-
r
e
late
d
issu
e
re
m
a
in
s
h
o
w
to
e
ff
e
c
ti
v
e
l
y
p
ro
tec
t
w
e
b
a
p
p
li
c
a
ti
o
n
s
f
ro
m
X
S
S
a
tt
a
c
k
s.
In
f
irst
p
a
rt
o
f
th
is
p
a
p
e
r,
a
m
e
th
o
d
f
o
r
d
e
te
c
ti
n
g
X
S
S
a
tt
a
c
k
w
a
s
p
ro
p
o
se
d
b
y
c
o
m
b
in
in
g
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(CNN
)
w
it
h
lo
n
g
sh
o
r
t
term
m
e
m
o
ries
(
L
S
TM
),
In
it
ial
ly
,
p
re
-
p
ro
c
e
ss
in
g
w
a
s
a
p
p
li
e
d
t
o
X
S
S
Da
ta
S
e
t
b
y
d
e
c
o
d
in
g
,
g
e
n
e
ra
li
z
a
ti
o
n
a
n
d
t
o
k
a
n
iza
ti
o
n
,
a
n
d
th
e
n
w
o
rd
2
v
e
c
w
a
s
a
p
p
li
e
d
t
o
c
o
n
v
e
rt
w
o
rd
s
in
to
w
o
rd
v
e
c
to
rs
in
X
S
S
p
a
y
lo
a
d
s.
A
n
d
th
e
n
w
e
u
se
th
e
c
o
m
b
in
a
ti
o
n
CNN
w
it
h
L
S
T
M
to
train
a
n
d
tes
t
w
o
rd
v
e
c
to
rs
to
p
ro
d
u
c
e
a
m
o
d
e
l
th
a
t
c
a
n
b
e
u
se
d
in
a
we
b
a
p
p
li
c
a
ti
o
n
.
Ba
se
d
o
n
th
e
o
b
tan
e
d
re
su
lt
s,
it
is
o
b
se
rv
e
d
th
a
t
th
e
p
ro
p
o
se
d
m
o
d
e
l
a
c
h
e
v
ied
a
n
e
x
c
e
l
len
t
re
su
lt
w
it
h
a
c
c
u
ra
c
y
o
f
9
9
,
4
%
.
K
ey
w
o
r
d
s
:
A
l
g
o
r
ith
m
s
C
B
OW
C
NN
an
d
L
ST
M
C
o
n
f
u
s
io
n
m
atr
i
x
W
o
r
d
2
v
ec
XSS
T
h
is
is
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
:
R
ae
d
W
ah
ee
d
Kad
h
i
m
I
n
f
o
r
m
a
tics
I
n
s
tit
u
te
f
o
r
P
o
s
tg
r
ad
u
ate
Stu
d
ies
I
r
aq
i
C
o
m
m
i
s
s
io
n
f
o
r
C
o
m
p
u
t
er
an
d
I
n
f
o
r
m
atic
s
,
I
r
aq
E
m
ail:
ra
a
d
7
4
2
0
0
2
@
y
ah
o
o
.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Mo
s
t
o
f
th
e
e
x
i
s
ti
n
g
w
eb
ap
p
licatio
n
s
ar
e
w
id
el
y
a
v
ailab
l
e
an
d
o
p
en
to
all
p
eo
p
le,
ev
e
n
h
ac
k
er
s
,
p
r
o
v
id
ed
b
y
co
m
p
an
ie
s
th
at
p
r
o
v
id
e
th
e
w
eb
s
er
v
ice
[1
-
4
]
.
A
lar
g
e
p
o
r
tio
n
o
f
t
h
ese
att
ac
k
s
ar
e
XSS
t
y
p
e,
w
h
ic
h
h
as
b
ee
n
co
n
s
id
er
ed
o
n
e
o
f
th
e
to
p
th
r
ea
t
s
i
n
th
e
OW
ASP
clas
s
if
icatio
n
f
o
r
s
ev
er
al
y
ea
r
s
[5
-
8
]
.
Hac
k
er
s
ca
n
,
f
o
r
ex
a
m
p
le,
ac
ce
s
s
m
u
ltip
le
r
eso
u
r
ce
s
,
s
en
s
it
iv
e
d
ata,
p
h
is
h
in
g
attac
k
s
,
m
i
s
i
n
f
o
r
m
atio
n
,
in
s
er
ti
n
g
m
a
licio
u
s
co
n
te
n
t,
t
a
m
p
er
in
g
W
eb
p
ag
es,
h
ij
ac
k
i
n
g
u
s
er
s
e
s
s
io
n
s
,
co
n
tr
o
llin
g
u
s
er
’
s
b
r
o
w
s
er
an
d
ac
ce
s
s
a
d
atab
ase
w
h
e
n
th
e
y
s
u
cc
ee
d
in
h
ac
k
i
n
g
th
e
m
i
n
to
web
ap
p
licatio
n
s
[
9
-
1
1
]
.
X
S
S
a
t
ta
ck
s
a
r
e
c
l
ass
if
i
e
d
in
to
t
h
r
e
e
m
a
in
ca
t
eg
o
r
i
es
:
n
o
n
-
p
e
r
s
i
s
t
en
t
X
SS
a
tt
a
ck
s
,
p
e
r
s
i
s
t
e
n
t
XS
S
a
t
t
a
ck
s
,
a
n
d
D
O
M
-
b
as
e
d
XS
S
a
t
t
a
ck
s
[
1
2
-
1
7
]
.
T
h
e
X
SS
a
tt
a
ck
is
e
asy
t
o
im
p
l
em
en
t
,
b
u
t
i
t
is
d
if
f
i
cu
lt
t
o
d
e
t
e
ct
a
n
d
p
r
ev
en
t
th
is
at
t
a
ck
b
e
c
au
s
e
o
f
d
i
f
f
e
r
e
n
t
b
r
o
w
s
e
r
s
u
s
e
d
.
Fu
r
th
e
r
m
o
r
e
,
F
ig
u
r
e
1
s
h
o
w
s
th
e
X
SS
a
tt
a
ck
a
r
c
h
it
e
c
tu
r
e
,
a
s
i
t
is
n
o
t
e
d
f
r
o
m
th
i
s
f
ig
u
r
e
th
at
th
e
in
t
r
u
d
e
r
a
d
d
s
an
in
je
c
t
e
d
c
o
d
e
t
o
a
w
e
b
s
i
t
e
an
d
u
s
e
s
th
is
s
i
te
a
s
a
p
r
ey
t
o
c
a
t
ch
a
v
ic
t
im
,
an
d
t
h
e
in
t
r
u
d
e
r
c
a
n
r
e
a
d
th
e
c
o
o
k
i
es
o
f
th
e
w
e
b
s
i
te
u
s
e
r
o
r
o
b
t
a
i
n
th
e
p
a
s
s
w
o
r
d
o
f
t
h
e
u
s
e
r
'
s
l
o
g
in
a
c
c
o
u
n
t
,
d
is
g
u
is
e
d
a
s
a
u
s
e
r
l
o
g
i
n
w
e
b
s
it
e
,
t
o
o
b
t
a
i
n
p
e
r
m
is
s
i
o
n
s
f
r
o
m
th
e
u
s
e
r
[
1
8
-
2
2
]
.
No
n
-
p
er
s
i
s
te
n
t
XS
S
attac
k
s
i
m
p
ac
t
t
h
e
cu
r
r
e
n
t
u
s
er
-
v
i
s
ite
d
p
ag
e,
allo
w
i
n
g
t
h
e
u
s
er
to
ac
ce
s
s
t
h
e
attac
k
er
'
s
ta
m
p
er
ed
co
n
n
ec
tio
n
.
W
h
en
t
h
e
u
s
er
v
is
i
ts
t
h
e
p
ag
e,
th
e
e
m
b
ed
d
ed
s
cr
ip
t
f
o
r
th
e
attac
k
is
e
x
ec
u
ted
b
y
th
e
b
r
o
w
s
er
o
f
t
h
e
User
[
2
3
,
2
4
]
.
P
er
s
is
ten
t
X
SS
s
to
r
ed
s
er
v
er
-
s
id
e
d
ata
o
f
t
h
e
at
tack
e
r
an
d
th
e
a
ttack
ca
n
s
tar
t
w
it
h
th
e
attac
k
d
ata.
X
SS
attac
k
s
t
h
at
ar
e
n
o
t
p
er
s
i
s
ten
t
a
n
d
XSS
at
tack
s
th
a
t
ar
e
p
er
s
is
ten
t
ca
n
b
e
d
iv
id
ed
in
to
th
r
ee
ca
teg
o
r
ies
:
s
to
r
ed
XSS attac
k
s
,
r
ef
lec
tiv
e
XSS attac
k
s
,
o
r
d
o
m
-
r
elate
d
X
SS
attac
k
s
[
2
5
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
A
h
yb
r
id
o
f CN
N
a
n
d
LS
TM
meth
o
d
s
fo
r
s
ec
u
r
in
g
w
eb
a
p
p
lica
tio
n
a
g
a
in
s
t…
(
R
a
ed
W
a
h
ee
d
K
a
d
h
im
)
1023
Fig
u
r
e
1
.
T
h
e
ar
ch
itectu
r
e
o
f
XSS attac
k
A
n
u
m
b
er
o
f
r
esear
c
h
er
s
h
a
v
e
p
r
o
p
o
s
ed
a
s
et
o
f
m
et
h
o
d
s
to
d
etec
t
X
SS
t
h
r
ea
ts
d
is
co
v
er
ed
b
y
ar
tif
icial
i
n
tell
ig
e
n
ce
in
r
ec
e
n
t
d
ec
ad
es su
ch
a
s
:
B
A
V
is
h
n
u
e
t
al.
[
4
]
u
s
ed
t
h
r
ee
ML
al
g
o
r
ith
m
s
(
SVM,
Na
iv
e
B
a
y
e
s
,
J
4
8
d
ec
is
io
n
tr
ee
s
)
to
d
etec
t
w
h
et
h
er
o
r
n
o
t t
h
er
e
ar
e
X
SS
t
h
r
ea
ts
i
n
th
e
UR
L
an
d
j
av
a
s
cr
ip
t.
S R
a
th
o
r
e
et
a
l.
[
5
]
ten
tr
ad
itio
n
al
al
g
o
r
ith
m
s
o
f
ML
ar
e
u
s
ed
to
d
etec
t
w
h
et
h
er
o
r
n
o
t
th
e
SN
S
w
eb
s
ite
s
co
n
tain
X
SS
attac
k
s
.
L
i
k
ar
is
h
et
al.
[
6
]
p
r
o
p
o
s
ed
a
ML
al
g
o
r
th
i
m
s
in
w
eb
ap
p
lic
atio
n
s
to
d
etec
t
t
h
e
o
b
s
cu
r
ed
m
alicio
u
s
J
av
aScr
ip
t,
t
h
is
m
et
h
o
d
estab
lis
h
ed
6
5
f
ea
t
u
r
es
a
n
d
tr
ain
ed
f
o
u
r
c
lasi
f
icatio
n
clas
s
i
f
ier
s
.
Ah
m
ed
an
d
A
li
[
7
]
P
r
o
p
o
s
ed
u
s
e
o
f
g
e
n
etic
al
g
o
r
ith
m
s
t
o
d
etec
t X
SS
attac
k
s
,
th
e
y
tes
te
d
th
eir
s
o
lu
tio
n
to
w
eb
ap
p
licatio
n
s
b
ased
o
n
M
y
SQ
L
an
d
P
HP
.
W
an
g
a
n
d
Z
h
o
u
[
8
]
No
v
el
h
as
p
r
o
p
o
s
ed
r
ely
i
n
g
o
n
n
e
w
t
ec
h
n
o
lo
g
ies.
T
o
d
etec
t
XSS
a
ttack
s
,
f
o
r
ex
a
m
p
le,
HT
ML
5
ev
a
lu
ated
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
u
s
i
n
g
XS
Ser
,
a
co
m
m
o
n
to
o
l
cr
ea
ted
b
y
OW
A
SP
[
9
]
.
B
u
ilt
an
S2
XS2
to
o
l
to
d
etec
t
s
er
v
er
-
s
id
e
XS
S
attac
k
s
b
ased
o
n
t
h
e
a
u
to
m
at
ic
b
o
r
d
er
in
j
ec
tio
n
s
y
s
te
m
a
n
d
t
h
e
p
r
in
c
ip
le
o
f
p
o
lic
y
g
en
er
atio
n
,
to
test
t
h
is
p
r
o
ce
s
s
,
th
e
y
u
s
e
f
o
u
r
J
SP
p
r
o
g
r
a
m
m
es.
Nat
h
an
e
t
al.
[
1
0
]
p
r
o
p
o
s
ed
s
y
s
te
m
to
d
e
tect
XXS
attac
k
s
b
a
s
ed
o
n
r
a
n
d
o
m
f
o
r
est
as
c
lass
if
ier
.
Var
to
u
h
i
[
1
1
]
P
r
o
p
o
s
ed
s
y
s
te
m
f
o
r
an
a
l
y
zi
n
g
HT
T
P
tr
af
f
ic
to
ex
tr
ac
t
v
ec
to
r
f
ea
tu
r
es
b
y
u
s
in
g
n
-
g
r
a
m
a
n
d
class
i
f
y
i
n
g
b
a
s
ed
o
n
f
o
r
es
t
is
o
latio
n
(
Fo
r
est)
m
e
th
o
d
.
Gr
o
s
s
e
Kath
r
i
n
[
1
2
]
p
r
o
p
o
s
ed
an
ad
v
er
s
ar
ial
ex
a
m
p
le,
Ma
l
w
ar
e
d
etec
tio
n
al
g
o
r
ith
m
g
en
er
atio
n
w
h
ic
h
allo
w
ed
6
3
p
er
ce
n
t
o
f
m
al
w
ar
e
d
etec
to
r
s
d
es
ig
n
ed
t
o
s
u
cc
es
s
f
u
l
l
y
b
y
p
ass
b
y
n
e
u
r
al
Net
w
o
r
k
s
.
Yao
W
an
g
,
W
an
-
d
o
n
g
C
ai,
an
d
P
en
g
-
c
h
en
g
W
ei,
2
0
1
6
[
1
3
]
Pre
s
en
t
a
m
o
d
er
n
,
d
ee
p
lea
r
n
in
g
s
y
s
te
m
f
o
r
d
etec
tin
g
J
av
aScr
ip
t
m
alicio
u
s
co
d
e.
[
1
4
]
,
2
0
1
8
,
Scr
ip
tNet
s
y
s
te
m
p
r
o
p
o
s
ed
to
d
etec
t
XSS
attac
k
s
u
s
in
g
s
tatic,
d
y
n
a
m
ic
an
al
y
s
i
s
a
n
d
d
ee
p
r
ec
u
r
r
en
t
n
eu
r
al
c
lass
if
i
ca
tio
n
.
I
n
t
h
is
p
ap
er
,
t
h
e
r
es
ea
r
ch
er
p
r
esen
ted
a
m
et
h
o
d
f
o
r
d
etec
ti
n
g
XS
S
at
t
ac
k
s
b
y
r
el
y
i
n
g
o
n
m
ix
i
n
g
t
w
o
m
et
h
o
d
s
o
f
d
ee
p
lear
n
i
n
g
,
n
a
m
el
y
t
h
e
C
NN
m
et
h
o
d
an
d
t
h
e
L
ST
M
m
e
th
o
d
b
ased
o
n
XS
S
g
lo
b
al
d
atase
t,
an
d
a
f
ter
o
b
tain
i
n
g
a
tr
ai
n
e
d
m
o
d
el,
it
i
s
u
s
ed
f
o
r
s
ec
u
r
e
ch
at
w
eb
ap
p
licatio
n
.
T
h
e
co
n
tr
ib
u
tio
n
o
f
t
h
is
r
esear
ch
in
cl
u
d
es t
h
r
ee
m
ai
n
co
n
tr
ib
u
tio
n
s
:
a)
Featu
r
e
ex
tar
ac
tio
n
b
ased
o
n
e
x
tr
ac
t in
f
o
r
m
a
tio
n
o
n
s
e
m
a
n
ti
cs b
y
u
s
in
g
W
o
r
d
2
v
ec
m
eth
o
d
.
b)
Mix
i
n
g
o
f
t
h
e
C
N
N
m
et
h
o
d
w
it
h
t
h
e
L
ST
M
m
et
h
o
d
to
d
et
ec
t
XSSattac
k
s
,
w
h
ich
ac
h
iev
ed
an
ac
cu
r
ate
99,
4
%.
T
h
e
r
est
o
f
th
e
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
.
R
esear
ch
m
eth
o
d
s
is
p
r
esen
ted
i
n
Sectio
n
2
,
an
d
in
Sectio
n
3
w
e
g
iv
e
a
d
etailed
d
escr
ip
tio
n
o
f
p
r
o
p
o
s
ed
s
y
s
te
m
.
I
n
Sec
tio
n
4
,
w
e
co
n
d
u
c
t
t
h
e
ex
p
er
i
m
en
ts
a
n
d
ev
alu
a
tio
n
r
es
u
lt
s
.
Fin
a
ll
y
,
i
n
Sectio
n
5
,
w
e
s
u
m
m
ar
ize
o
u
r
w
o
r
k
a
n
d
d
is
cu
s
s
f
u
r
t
h
er
w
o
r
k
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
Secu
r
i
n
g
o
f
W
eb
ap
p
licatio
n
i
s
cr
itical
is
s
u
e,
b
ec
au
s
e
o
f
in
c
r
ea
s
in
g
c
y
b
er
cr
i
m
es.
I
n
th
i
s
s
ec
tio
n
,
t
h
e
r
esear
ch
m
et
h
o
d
s
h
a
v
e
b
ee
n
d
i
s
cu
s
s
ed
.
2
.
1
.
Co
nv
o
lutio
na
l neura
l net
w
o
rk
(
CNN)
Net
w
o
r
k
attac
k
d
etec
tio
n
h
as
r
ec
en
tl
y
b
ec
o
m
e
m
o
r
e
i
m
p
o
r
t
an
t
to
s
o
cial
n
et
w
o
r
k
i
n
g
d
ata
p
r
o
tectio
n
,
as
s
ec
u
r
it
y
t
h
r
ea
ts
s
u
ch
a
s
cr
o
s
s
-
s
ite
s
cr
ip
tin
g
,
click
j
ac
k
i
n
g
,
DDOS,
p
r
o
b
e
an
d
id
en
tity
t
h
ef
t
h
a
v
e
i
n
cr
ea
s
ed
.
T
h
e
aim
o
f
d
etec
tin
g
n
et
w
o
r
k
attac
k
s
is
to
d
if
f
er
e
n
tiate
b
etw
ee
n
h
o
s
ti
le
ac
tiv
it
ies
an
d
n
e
t
w
o
r
k
tr
a
f
f
ic
[
1
6
]
.
Ma
ch
i
n
e
lear
n
in
g
(
ML
)
h
a
s
b
ee
n
s
u
cc
e
s
s
f
u
ll
y
ap
p
lied
in
d
i
f
f
er
en
t
ar
ea
s
o
f
co
m
p
u
ter
s
cie
n
ce
,
s
u
c
h
a
s
i
m
a
g
e
p
r
o
ce
s
s
in
g
a
n
d
co
m
p
u
ter
v
is
i
o
n
.
On
e
o
f
th
e
p
o
p
u
lar
ML
tech
n
iq
u
es
i
n
th
e
f
ield
o
f
in
tr
u
s
io
n
d
etec
tio
n
is
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
0
2
2
-
10
29
1024
K
-
Nea
r
e
s
t
Nei
g
h
b
o
r
(
K
-
NN)
[
1
7
]
.
Ho
w
ev
er
,
co
n
v
o
l
u
tio
n
s
n
e
u
r
al
n
et
w
o
r
k
(
C
NN)
r
e
m
a
in
s
th
e
co
m
m
o
n
l
y
u
s
ed
d
ee
p
lear
n
i
n
g
m
o
d
el
f
o
r
lar
g
e
-
s
ca
le
i
m
a
g
e
r
ec
o
g
n
it
io
n
a
n
d
clas
s
if
icatio
n
.
T
h
e
C
N
N
w
o
r
k
s
b
y
d
ir
ec
tl
y
ex
tr
ac
ti
n
g
i
m
a
g
es.
C
N
Ns
lear
n
to
d
etec
t
v
ar
io
u
s
i
m
ag
e
f
ea
t
u
r
es
u
s
i
n
g
te
n
s
o
r
h
u
n
d
r
ed
s
o
f
h
id
d
en
la
y
er
s
.
E
ac
h
h
id
d
en
la
y
er
i
n
cr
ea
s
es
th
e
co
m
p
lex
i
t
y
o
f
t
h
e
i
m
a
g
e
c
h
ar
ac
t
er
is
tics
lear
n
ed
.
T
h
e
w
o
r
k
in
g
p
r
in
cip
le
o
f
t
h
e
C
NN
r
elies
o
n
d
ir
ec
t
ex
tr
ac
tin
o
f
i
m
ag
e
f
ea
tu
r
e
s
.
C
NNs
ar
e
tr
ai
n
ed
to
d
etec
t
v
ar
io
u
s
i
m
a
g
e
f
ea
tu
r
es
w
it
h
v
ar
y
i
n
g
n
u
m
b
er
s
o
f
h
id
d
en
la
y
er
s
an
d
ea
ch
h
id
d
en
la
y
er
ad
d
s
to
th
e
co
m
p
le
x
it
y
o
f
t
h
e
lear
n
ed
i
m
a
g
e
f
ea
t
u
r
es [
1
8
]
.
2
.
2
.
L
o
ng
s
ho
rt
t
er
m
m
e
m
o
r
ies
(
L
S
T
M
)
L
o
n
g
-
f
o
r
m
s
h
o
r
t
m
e
m
o
r
y
(
L
S
T
M)
n
et
w
o
r
k
s
ar
e
a
c
u
tti
n
g
-
e
d
g
e
s
eq
u
e
n
ce
lear
n
i
n
g
tec
h
n
i
q
u
e.
T
h
e
y
ar
e
less
co
m
m
o
n
f
o
r
ti
m
e
s
er
ies
f
o
r
ec
asts
,
b
u
t
ar
e
in
tr
in
s
ic
all
y
s
u
itab
le
f
o
r
th
i
s
d
o
m
ai
n
.
L
ST
M
is
an
R
N
N
v
ar
ian
t
ca
p
ab
le
o
f
lear
n
i
n
g
l
o
n
g
-
ter
m
d
ep
en
d
en
cie
s
.
T
h
e
Ho
ch
r
eiter
an
d
Sch
m
id
h
u
b
er
f
ir
s
t
p
r
o
p
o
s
ed
an
d
r
ef
in
ed
L
ST
Ms
b
y
m
a
n
y
o
t
h
e
r
r
esear
ch
er
s
.
T
h
ey
w
o
r
k
w
el
l
o
n
a
w
id
e
r
an
g
e
o
f
p
r
o
b
lem
s
an
d
ar
e
th
e
m
o
s
t
u
s
ed
R
NN
t
y
p
e.
T
h
e
m
at
h
e
m
atica
l
f
o
r
m
u
lat
io
n
o
f
L
ST
M
u
n
its
p
r
io
r
to
d
iv
in
g
i
n
t
h
e
t
h
eo
r
y
b
eh
i
n
d
t
h
e
co
n
ce
p
t sh
o
w
s
h
o
w
th
e
h
id
d
en
s
tate
h
t
-
t ti
m
e
t is d
eter
m
i
n
ed
f
r
o
m
t
h
e
p
r
ev
io
u
s
h
id
d
en
s
tate
h
t
-
1
[
1
9
]
:
T
h
e
I
n
p
u
t g
a
te
is
d
y
s
p
laied
as,
i(
t)
=
σ
(
W
(
i
)
x
(
t)
+
u
(
i)
h
(
t
-
1
)
)
(
1
)
Fo
r
g
et
Gate
ca
lcu
lated
as,
f
(
t)
=
σ
(
W
(
f
)
x
(
t)
+
u
(
f
)
h
(
t
-
1
)
)
(
2
)
an
d
,
th
e
O
u
tp
u
t/ E
x
p
o
s
u
r
e
g
a
t
e
as,
o
(
t)
=
σ
(
W
(
o
)
x
(
t)
+
u
(
o
)
h
(
t
-
1
)
)
(
3
)
an
d
th
e
Ne
w
m
e
m
o
r
y
ce
ll a
s
,
c˜
(
t)
=
tan
h
(
W
(
c)
x
(
t)
+
u
(
c)
h
(
t
-
1
)
)
(
4
)
L
ast
l
y
,
t
h
e
f
in
al
m
e
m
o
r
y
ce
ll i
s
ca
lcu
lated
as,
c(
t)
=
f
(
t)
o
c˜
(
t
-
1
)
+
i(
t)
o
c
˜
(
t
)
(
5
)
An
d
,
h
(
t)
=
o
(
t)
o
tan
h
(
c(
t)
)
(
6
)
2
.
3
.
Wo
rd2
v
ec
Go
o
g
le
r
elea
s
ed
W
o
r
d
2
v
ec
in
2
0
1
3
as
a
d
ee
p
lear
n
in
g
to
o
l
th
at
co
m
p
r
is
es
o
f
t
h
e
C
o
n
ti
n
u
o
u
s
B
ag
o
f
W
o
r
d
s
(
C
B
OW
)
m
o
d
el
&
t
h
e
Sk
ip
-
Gr
a
m
m
o
d
el.
T
h
e
r
o
le
o
f
t
h
e
C
B
OW
m
o
d
el
is
to
p
er
f
o
r
m
co
n
tex
t
-
b
ased
p
r
ed
ictio
n
o
f
th
e
c
u
r
r
en
t
w
o
r
d
w
h
ile
t
h
e
s
u
r
r
o
u
n
d
i
n
g
w
o
r
d
s
f
o
r
th
e
g
i
v
e
n
cu
r
r
en
t
w
o
r
d
a
r
e
p
r
ed
icted
b
y
t
h
e
Sk
ip
-
Gr
a
m
m
o
d
el
[
2
0
]
.
2
.
4
.
E
v
a
lua
t
i
o
n
m
et
ho
d
s
T
h
er
e
ar
e
m
an
y
t
y
p
es
o
f
e
v
al
u
atio
n
m
et
h
o
d
s
th
a
t
h
a
v
e
b
ee
n
d
esi
g
n
ed
f
o
r
p
er
f
o
r
m
a
n
ce
e
v
alu
a
tio
n
o
f
n
e
w
m
eth
o
d
s
.
I
n
th
i
s
w
o
r
k
,
t
h
e
m
o
s
t
co
m
m
o
n
t
y
p
es
o
f
ev
alu
atio
n
m
et
h
o
d
s
w
h
ich
ar
e
a
r
e
A
cc
u
r
ac
y
,
R
ec
al
l
an
d
F1
Sco
r
e.
Mo
r
eo
v
er
,
th
er
e
ar
e
f
o
u
r
c
lass
e
s
i
n
t
h
e
co
n
f
u
s
io
n
m
atr
ix
,
th
e
s
e
ar
e
t
h
e
T
r
u
e
P
o
s
itiv
e
s
w
h
ic
h
s
p
ec
if
ie
s
a
m
o
u
n
t
o
f
X
SS
s
a
m
p
les
th
at
ar
e
co
r
r
ec
tly
clas
s
i
f
i
ed
,
th
e
Fals
e
P
o
s
itiv
es
w
h
ic
h
p
o
r
tr
ay
s
t
h
e
n
u
m
b
er
o
f
No
n
-
XSS
s
a
m
p
le
s
w
r
o
n
g
l
y
clas
s
if
ied
as
XSS,
t
h
e
T
r
u
e
Neg
ati
v
e
th
a
t
in
d
icate
t
h
e
n
u
m
b
er
o
f
co
r
r
ec
tl
y
cl
ass
i
f
ied
No
n
-
X
SS
s
a
m
p
les,
an
d
th
e
Fals
e
Ne
g
ati
v
es
th
at
ca
p
tu
r
e
th
e
t
h
e
n
u
m
b
er
o
f
X
S
S
s
a
m
p
le
s
w
r
o
n
g
l
y
class
i
f
ied
as No
n
-
X
SS
.
I
n
t
h
is
w
o
r
k
,
t
h
e
ca
lcu
latio
n
o
f
ac
c
u
r
ac
y
is
p
r
esen
ted
as in
(
7
)
,
(
7
)
w
h
er
ein
t
h
e
p
r
ec
is
io
n
w
a
s
ca
lc
u
lated
as i
n
(
8
)
,
(
8
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
A
h
yb
r
id
o
f CN
N
a
n
d
LS
TM
meth
o
d
s
fo
r
s
ec
u
r
in
g
w
eb
a
p
p
lica
tio
n
a
g
a
in
s
t…
(
R
a
ed
W
a
h
ee
d
K
a
d
h
im
)
1025
A
l
s
o
,
th
e
s
e
n
s
i
tiv
it
y
w
a
s
ca
lcu
lated
as in
(
9
)
,
(
9
)
3.
P
RO
P
O
SE
D
SYS
T
E
M
Dee
p
lear
n
in
g
h
as
p
r
o
g
r
ess
ed
g
r
ea
tl
y
i
n
co
m
p
u
ter
v
is
io
n
,
t
h
e
p
r
o
ce
s
s
i
n
g
o
f
n
at
u
r
al
la
n
g
u
ag
e
s
a
n
d
ar
tif
icial
in
tel
lig
e
n
ce
,
a
n
d
it
h
as
al
s
o
b
eg
u
n
to
d
ev
elo
p
in
t
h
e
f
ield
o
f
s
ec
u
r
i
t
y
a
n
d
m
o
v
e
to
p
r
ac
tical
ap
p
licatio
n
s
.
T
h
e
e
x
p
er
i
m
e
n
ts
ca
r
r
ied
o
u
t
u
s
i
n
g
t
h
i
s
p
r
o
p
o
s
ed
m
et
h
o
d
m
ai
n
l
y
u
s
e
te
x
t
cla
s
s
i
f
icatio
n
m
et
h
o
d
s
to
d
ete
ct
XSS
attac
k
s
b
y
d
ee
p
lear
n
in
g
.
Firstl
y
,
w
e
p
r
o
ce
s
s
in
p
u
t
d
ata,
in
clu
d
i
n
g
Dec
o
d
in
g
,
g
en
er
aliza
tio
n
an
d
to
k
a
n
iztio
n
I
n
p
u
t
d
ata.
T
h
en
,
w
it
h
a
w
o
r
d
2
v
ec
m
o
d
el,
w
e
ex
tr
ac
t
f
ea
t
u
r
es.
Fi
n
all
y
,
th
ese
f
ea
t
u
r
es
h
a
v
e
b
ee
n
p
lace
d
in
to
C
N
N
-
L
STM
to
tr
ain
a
clas
s
i
f
icati
o
n
to
d
if
f
er
e
n
tia
te
X
SS
f
r
o
m
n
o
r
m
al
s
a
m
p
le
s
.
Fi
g
u
r
e
2
s
h
o
w
s
th
e
ar
c
h
itect
u
r
e
o
f
t
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
.
I
n
t
h
e
n
e
x
t
s
u
b
s
ec
tio
n
,
t
h
e
p
r
o
ce
s
s
in
g
s
te
p
es
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
s
h
a
v
e
b
ee
n
ill
u
s
tr
ated
.
Fig
u
r
e
2
.
T
h
e
ar
ch
itectu
r
e
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
3
.
1
.
P
re
pro
ce
s
s
ing
s
t
a
g
e
T
h
is
is
t
h
e
f
ir
s
t
s
ta
g
e
o
f
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
s
af
ter
i
m
p
o
r
t
d
ata
to
th
e
s
y
s
te
m
,
i
t
is
co
n
s
i
s
tes
o
f
t
h
r
ee
s
tep
s
w
h
ich
ar
e
d
ec
o
d
ed
,
g
en
er
alize
d
an
d
to
k
en
ized
,
as e
x
p
l
ain
ed
b
elo
w
:
3
.
1
.
1
.
Dec
o
der
A
ttac
k
er
s
m
a
y
av
o
id
co
n
v
e
n
ti
o
n
al
f
iltra
ti
o
n
o
r
v
alid
atio
n
b
ased
o
n
r
e
g
u
lar
ex
p
r
es
s
io
n
s
b
y
r
el
y
in
g
o
n
en
co
d
in
g
tec
h
n
iq
u
es
s
u
c
h
a
s
Hex
e
n
co
d
in
g
,
UR
L
en
co
d
in
g
,
Un
ico
d
e
en
co
d
in
g
,
HT
ML
e
n
tit
y
e
n
co
d
in
g
,
etc.
Hen
ce
,
th
i
s
s
tu
d
y
p
r
o
p
o
s
esa
d
ec
o
d
er
f
o
r
r
ec
u
r
r
en
t
test
in
g
an
d
co
n
v
er
s
io
n
o
f
all
th
e
d
in
p
u
t
d
ata
ec
o
d
in
g
o
p
tio
n
s
i
n
to
th
e
o
r
ig
i
n
al
f
o
r
m
.
W
o
r
d
s
co
n
s
is
ti
n
g
o
f
ch
ar
ac
te
r
s
an
d
n
u
m
b
er
s
R
esp
ec
ti
v
el
y
,
f
o
r
th
e
r
ed
u
ctio
n
o
f
th
e
n
u
m
b
er
o
f
p
ar
ticip
les
t
h
e
n
u
m
b
er
o
f
n
u
m
b
er
s
a
n
d
th
e
h
y
p
er
li
n
k
s
m
u
s
t
b
e
s
i
m
p
li
f
ied
t
o
r
ep
lace
n
u
m
b
er
s
b
y
"
0
"
an
d
th
e
h
y
p
er
li
n
k
s
to
h
ttp
:
/
u
.
Af
ter
t
h
e
p
r
o
ce
s
s
i
n
g
o
f
th
e
d
ec
o
d
er
,
f
o
r
ex
a
m
p
le,
an
o
b
s
cu
r
ed
XSS
p
ay
lo
ad
u
s
i
n
g
HT
ML
:e
n
c
o
d
in
g
“
<scr
ip
t>
&
#
9
7
;
&
#
1
0
8
;
&
#
1
0
1
;
&
#
1
1
4
;
&
#
1
1
6
;
&
#
4
0
;
&
#
4
9
;
&
#
4
1
;</scr
ip
t>”
is
r
esto
r
ed
to
“
<scr
ip
t>a
ler
t (
1
)
</scr
ip
t>”
.
3
.
1
.
2
.
G
ener
a
liza
t
io
n
T
o
r
ed
u
ce
th
e
i
n
ter
f
er
en
ce
o
f
r
ed
u
n
d
an
t
a
n
d
ir
r
elev
a
n
t
i
n
f
o
r
m
atio
n
,
t
h
e
d
ec
o
d
ed
d
ata
is
g
en
er
alis
ed
b
y
t
h
e
f
o
llo
w
in
g
s
tep
s
:
Firs
t,
w
e
r
ep
lace
s
e
v
er
al
UR
L
s
o
f
t
h
e
i
n
p
u
t
d
ata
u
s
in
g
'h
ttp
:/
w
eb
s
ite.
T
h
e
n
u
m
b
er
s
i
n
th
e
d
ata
ar
e
th
en
r
ep
lace
d
b
y
"
0
.
"
A
n
d
th
e
s
tr
in
g
is
r
ep
lac
ed
w
it
h
"
p
ar
am
s
tr
i
n
g
"
as
a
f
u
n
ct
io
n
p
ar
a
m
eter
.
Fu
r
t
h
er
m
o
r
e,
o
th
er
s
p
ec
ial
ch
a
r
ac
ter
is
tics
,
s
u
ch
a
s
co
n
tr
o
l a
n
d
b
lan
k
ch
ar
ac
ter
s
w
er
e
eli
m
i
n
ated
.
3
.
1
.
3
.
T
o
k
eniza
t
io
n
T
h
is
s
tu
d
y
d
es
ig
n
ed
a
s
et
o
f
r
eg
u
lar
e
x
p
r
ess
io
n
s
f
o
r
in
p
u
t
d
ata
cu
s
to
m
izatio
n
b
ased
o
n
t
h
e
s
cr
ip
tin
g
lan
g
u
a
g
e
f
ea
tu
r
es.
T
h
e
T
o
k
en
i
za
tio
n
C
la
s
s
i
f
icatio
n
is
s
h
o
w
n
in
T
ab
le
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
0
2
2
-
10
29
1026
T
ab
le
1
.
C
lass
if
icatio
n
o
f
to
k
e
n
izatio
n
C
l
a
ssi
f
i
c
a
t
i
o
n
L
i
st
S
t
a
r
t
L
a
b
e
l
<
scri
p
t
>
,
<
f
r
a
me
>
,
<
i
mg
,
<
b
o
d
y
>
,
e
t
c
En
d
L
a
b
e
l
<
/
scr
i
p
t
>
,
<
/
f
r
a
me
>
,
<
/
b
o
d
y
>
,
e
t
c
W
i
n
d
o
w
s E
v
e
n
t
o
n
e
r
r
o
r
=
,
o
n
l
o
a
d
=
,
o
n
b
l
u
r
=
,
o
n
c
u
t
=
,
e
t
c
F
u
n
c
t
i
o
n
N
a
me
a
l
e
r
t
(
,
p
r
o
mp
t
(
,
S
t
r
i
n
g
.
f
r
o
mC
h
a
r
C
o
d
e
(
,
e
t
c
S
c
r
i
p
t
U
R
L
j
a
v
a
scr
i
p
t
:
,
v
b
scr
i
p
t
:
,
e
t
c
O
t
h
e
r
s
>
,
)
,
#
,
e
t
c
A
ll
s
tep
o
f
p
r
ep
r
o
ce
s
s
in
g
s
ta
g
e
ab
o
v
e
ar
e
s
u
m
m
ar
ized
in
a
n
alg
o
r
ith
m
(
3
.
1
)
o
f
as sh
o
w
n
:
Algorithm 3.1: Preprocessing Stage Algorithm
Input
Read vector (V).
Output
Number of scripts.
Step1
Read each script until reach the end of file
Step2
Converts the input data into the original form( Decoder)
Step3
Replaced
numbers
in
the
input
data
wi
th
“0
”,
an
d
th
e
st
ri
ng
as
a
fu
nc
ti
on
parameter with “param_string”. (Generalization).
Step4
Separate each word from the scripts to obtain a tokens (Tokenization).
Step5
Return scripts list (Data)
Step6
Return token list from previous step (Word)
Step7
Calculate the frequency for each token (Vocabulary)
Step8
For each token in VocabularyList
If token found in list
Then
Take token
Else
replace it with (
-
1)
End if
Step9
Return scripts list
3
.
2
.
Wo
rd2
Vec
t
o
r
I
n
th
e
W
o
r
d
2
Vec
s
tag
e
w
e
'
r
e
g
o
in
g
to
h
a
v
e
f
ea
t
u
r
es
ex
tr
ac
tio
n
,
w
e
n
ee
d
to
cr
ea
te
a
v
o
ca
b
u
lar
y
o
f
th
e
m
o
s
t
co
m
m
o
n
w
o
r
d
s
i
n
t
h
e
s
cr
ip
t.
3
,
0
0
0
to
k
en
s
h
a
v
e
b
ee
n
u
s
ed
i
n
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
I
n
W
o
r
d
2
Vec
,
a
k
e
y
m
ea
s
u
r
e
o
f
m
ea
n
in
g
i
s
t
h
e
co
n
tex
t
o
f
t
h
e
w
o
r
d
.
Her
e
t
h
e
b
u
il
t
-
in
w
o
r
d
v
ec
to
r
m
o
d
el
is
u
s
ed
to
b
u
i
ld
a
s
e
m
a
n
ticized
XSS
m
o
d
el
to
a
llo
w
th
e
s
y
s
te
m
to
u
n
d
er
s
ta
n
d
HT
ML
lan
g
u
ag
e
s
l
ik
e
<
s
cr
ip
t
>
a
n
d
aler
tin
g
)
.
Fo
r
m
o
d
elli
n
g
th
e
w
o
r
d
2
v
ec
class
o
f
t
h
e
g
e
n
i
s
m
m
o
d
u
le
i
s
u
s
ed
an
d
th
e
w
o
r
d
s
p
ac
e
is
1
2
8
d
i
m
en
s
io
n
s
.
T
h
e
co
n
ce
p
t
o
f
th
e
n
eu
r
al
n
et
w
o
r
k
i
n
w
o
r
d
2
v
ec
to
r
is
to
p
r
o
v
id
e
o
u
r
tar
g
et
in
p
u
t
w
o
r
d
s
as
o
n
e
-
h
o
t
v
ec
to
r
s
.
T
h
en
w
e
w
an
t
th
e
n
e
u
r
al
n
e
t
w
o
r
k
to
tr
ai
n
t
h
e
p
r
o
b
ab
ilit
y
o
f
a
v
alid
co
n
te
x
t
w
o
r
d
th
r
o
u
g
h
a
h
id
d
en
la
y
er
w
h
ile
r
ed
u
ci
n
g
t
h
e
p
r
o
b
ab
ilit
y
o
f
an
in
v
alid
co
n
tex
t
w
o
r
d
(
i.e
.
w
o
r
d
s
w
h
ich
n
e
v
er
ap
p
ea
r
in
t
h
e
s
u
r
r
o
u
n
d
in
g
s
o
f
t
h
e
tar
g
et
w
o
r
d
)
.
T
h
is
in
clu
d
es
t
h
e
u
s
e
o
f
a
s
o
f
t
m
a
x
o
n
th
e
o
u
tp
u
t
la
y
er
.
W
h
en
tr
ain
i
n
g
i
s
co
m
p
leted
,
t
h
e
o
u
tp
u
t
la
y
er
i
s
r
e
m
o
v
ed
an
d
t
h
e
w
e
ig
h
ts
o
f
t
h
e
h
id
d
en
la
y
er
ar
e
o
u
r
e
m
b
ed
d
in
g
v
ec
to
r
s
.
A
Sk
ip
-
g
r
a
m
m
a
y
b
e
d
ep
icted
a
s
a
n
eu
r
al
n
et
w
o
r
k
co
n
s
is
ti
n
g
o
f
a
m
ap
p
in
g
la
y
er
an
d
an
o
u
tp
u
t
la
y
er
.
I
n
p
u
t
r
ep
r
esen
tatio
n
:
E
v
er
y
w
o
r
d
in
t
h
e
in
p
u
t
la
y
er
i
s
a
s
i
n
g
le
h
ea
t
en
co
d
i
n
g
,
m
ea
n
in
g
t
h
at
all
w
o
r
d
s
ar
e
r
ep
r
esen
ted
as
N
-
d
i
m
e
n
s
io
n
a
l
v
ec
to
r
s
,
w
h
er
e
N
is
t
h
e
to
tal
n
u
m
b
er
o
f
w
o
r
d
s
o
f
t
h
e
v
o
c
ab
u
lar
y
.
E
ac
h
w
o
r
d
h
as a
s
ize
o
f
1
in
th
e
v
ec
to
r
an
d
th
e
r
e
m
ai
n
d
er
h
as a
v
al
u
e
o
f
0
.
Fo
r
w
ar
d
p
r
o
ce
s
s
o
f
p
r
o
p
ag
atio
n
in
th
e
n
et
w
o
r
k
:
T
h
e
o
u
tp
u
t
la
y
er
v
ec
to
r
v
al
u
e
ca
n
b
e
ca
lcu
lated
f
r
o
m
b
o
th
th
e
v
ec
to
r
o
f
th
e
h
i
d
d
en
lay
er
(
K
d
i
m
e
n
s
io
n
)
an
d
th
e
w
eig
h
t
m
atr
i
x
o
f
th
e
Kx
N
d
i
m
en
s
io
n
b
et
w
ee
n
th
e
h
id
d
en
la
y
er
an
d
t
h
e
o
u
tp
u
t
la
y
er
.
T
h
e
o
u
tp
u
t
la
y
er
is
also
an
N
-
d
i
m
en
s
io
n
al
v
ec
to
r
,
w
h
ic
h
m
a
tch
e
s
a
v
o
ca
b
u
lar
y
w
o
r
d
w
it
h
ea
ch
d
i
m
en
s
io
n
.
T
h
e
So
f
t
m
a
x
ac
tiv
at
io
n
f
u
n
ct
io
n
is
t
h
en
ap
p
lied
to
th
e
v
ec
to
r
o
f
th
e
o
u
tp
u
t
la
y
er
t
o
ca
lcu
late
ea
ch
w
o
r
d
'
s
p
r
o
b
ab
ilit
y
.
Fi
n
all
y
,
t
h
e
So
f
t
m
a
x
is
u
s
ed
f
o
r
ca
lcu
l
atin
g
th
e
li
k
eli
h
o
o
d
o
f
an
y
w
o
r
d
o
n
t
h
e
o
u
tp
u
t la
y
e
r
v
ec
to
r
.
3
.
3
.
CNN
ba
s
ed
o
n L
ST
M
T
h
e
f
ea
tu
r
e
v
ec
to
r
es
r
es
u
lted
f
r
o
m
w
o
r
d
2
v
ec
to
r
s
tep
w
ill
b
e
u
s
in
g
i
n
th
i
s
s
tep
,
w
h
ic
h
co
n
s
is
t
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
f
t
w
o
d
ee
p
lear
n
in
g
m
e
th
o
d
s
w
h
ic
h
ar
e
C
NN
a
n
d
L
ST
M.
Fu
r
t
h
er
m
o
r
e,
th
e
C
NN
b
ase
d
L
ST
M
u
s
ed
to
in
cr
ea
s
e
th
e
a
cc
u
r
ac
y
o
f
o
u
r
n
et
w
o
r
k
b
y
u
s
in
g
t
h
e
C
o
n
v
o
l
u
tio
n
ar
y
Ne
u
r
al
Net
w
o
r
k
(
C
N
N)
o
n
e
-
d
i
m
en
s
io
n
al
t
h
r
ee
-
ti
m
e
a
n
d
ze
r
o
p
ad
d
in
g
t
w
o
-
t
i
m
e,
r
es
u
lt
in
g
f
r
o
m
co
n
ca
te
n
atio
n
(
C
NN1
,
C
NN
2
a
n
d
C
NN)
u
s
e
i
n
th
e
L
ST
M.
Fig
u
r
e
3
s
h
o
w
s
t
h
e
ar
ch
itect
u
r
e
o
f
C
NN
-
L
ST
M
n
et
w
o
r
k
p
h
ase
s
Fig
u
r
e
3
.
th
e
ar
ch
itect
u
r
e
o
f
C
NN
-
L
ST
M
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
A
h
yb
r
id
o
f CN
N
a
n
d
LS
TM
meth
o
d
s
fo
r
s
ec
u
r
in
g
w
eb
a
p
p
lica
tio
n
a
g
a
in
s
t…
(
R
a
ed
W
a
h
ee
d
K
a
d
h
im
)
1027
Fig
u
r
e
3
.
T
h
e
ar
ch
itectu
r
e
o
f
C
NN
-
L
ST
M
T
h
e
m
ai
n
s
tep
s
o
f
C
N
N
-
L
ST
M
s
tag
e
ab
o
v
e
ar
e
s
h
o
w
n
i
n
a
n
alg
o
r
it
h
m
(
3
)
:
Algorithm 3.3: CNN
-
LSTM Stage Algorithm
Input
Input Layer
Output
Normal or abnormal
Step1
Convolutional
neural
network
one
dimension(CNN1)
Step2
Convolutional
neural
network
one
dimension(CNN2)
Step3
Zero padding 1
Step4
Convolutional
neural
network
one
dimension(CNN3)
Step5
Zero padding 2
Step6
Concatenate ( CNN1,CNN2,CNN3 )
Step7
LSTM
Step8
Dropout 0.5%
Step9
Fully Connected layer
Step10
Softmax Function
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
th
i
s
s
ec
tio
n
,
th
e
e
x
p
er
i
m
e
n
t
al
r
esu
lt
s
an
d
ev
al
u
atio
n
s
ta
g
e
h
av
e
b
ee
n
d
escr
ib
ed
.
4
.
1
.
da
t
a
s
et
s
et
t
ing
T
h
e
p
u
b
lic
s
ec
u
ir
t
y
d
ata
s
et
is
v
er
y
s
ca
r
ce
.
T
h
e
d
ata
f
r
o
m
t
h
e
ex
p
er
i
m
en
t
i
n
t
h
is
p
r
o
p
o
s
ed
m
e
th
o
d
co
n
tain
t
w
o
p
ar
ts
w
h
ich
ar
e
n
o
r
m
al
a
n
d
ab
n
o
r
m
al,
w
h
ile
ab
n
o
r
m
al
s
a
m
p
les
ar
e
m
o
r
e
th
a
n
4
0
,
0
0
0
f
r
o
m
www
.
x
s
s
ed
.
co
m
a
n
d
7
4
,
0
0
0
n
o
r
m
al
l
y
,
to
en
s
u
r
e
d
ata
s
ec
u
r
it
y
t
h
e
h
o
s
t
an
d
p
ath
in
f
o
r
m
at
io
n
i
s
r
e
m
o
v
ed
i
n
th
e
u
r
l,
an
d
o
n
l
y
t
h
e
p
a
y
lo
ad
p
ar
t
is
r
etain
ed
.
T
h
e
ab
o
v
e
d
ata
ar
e
s
av
ed
af
ter
u
r
l
e
n
co
d
in
g
i
n
t
h
e
csv
.
S
in
ce
u
r
l
is
en
co
d
ed
f
o
r
p
ar
t o
f
th
e
o
r
ig
in
a
l d
ata,
th
is
ca
n
o
n
l
y
b
e
u
s
ed
af
ter
t
w
o
u
r
l d
ec
o
d
in
g
.
4
.
2
.
Si
m
ula
t
io
n sp
ec
if
ica
t
io
n
T
h
is
w
a
s
co
n
d
u
cted
o
n
P
C
w
i
t
h
th
e
f
o
llo
w
i
n
g
f
ea
t
u
r
es: 2
.
6
0
GHz
,
1
6
G
R
A
M
a
n
d
NVI
DI
A
Ge
Fo
r
ce
GT
X
1
0
6
0
w
i
th
6
GB
,
I
n
te
l(
R
)
C
o
r
e(
T
M)
i7
-
7
7
0
0
C
P
U
@
.
Ker
as
i
s
a
h
ig
h
-
le
v
el,
P
y
t
h
o
n
-
d
ev
elo
p
ed
,
n
eu
r
al
n
et
w
o
r
k
A
P
I
in
ter
f
ac
e
t
h
at
is
i
m
p
le
m
e
n
tab
le
o
n
C
NT
K,
T
en
s
o
r
Flo
w
,
an
d
T
h
ea
n
o
[
1
5
]
.
B
ein
g
th
at
Ker
asi
s
ea
s
y
to
u
s
e
an
d
ex
h
ib
it
s
ea
s
y
ex
ten
s
ib
ilit
y
an
d
m
o
d
u
lar
it
y
,
I
w
as
u
s
ed
in
th
is
s
t
u
d
y
w
it
h
T
en
s
o
r
Flo
w
f
o
r
th
e
d
ev
elo
p
m
e
n
t a
n
d
ev
al
u
atio
n
o
f
th
e
p
r
p
o
s
ed
d
etec
tio
n
m
et
h
o
d
.
I
n
th
is
e
x
p
er
i
m
e
n
t,
1
0
-
f
o
ld
cr
o
s
s
v
a
lid
atio
n
w
a
s
u
s
ed
to
e
v
alu
ate
t
h
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
.
First,
th
e
d
ataset
w
as
r
an
d
o
m
l
y
p
ar
titi
o
n
ed
in
to
1
0
eq
u
iv
ale
n
t
s
izes
i
n
1
0
-
f
o
ld
cr
o
s
s
v
alid
atio
n
.
Nin
e
s
u
b
s
et
s
w
er
e
s
to
r
ed
as
th
e
tr
ain
i
n
g
s
et
w
h
ile
o
n
e
s
et
w
a
s
r
eser
v
d
f
o
r
th
e
v
alid
atio
n
o
f
t
h
e
m
o
d
el.
T
h
e
cr
o
s
s
-
v
alid
atio
n
p
r
o
ce
s
s
w
a
s
e
x
ec
u
ted
1
0
ti
m
e
s
,
an
d
ea
c
h
o
f
th
e
1
0
s
u
b
s
et
s
w
a
s
u
s
ed
j
u
s
t
o
n
ce
as
th
e
v
al
i
d
atio
n
d
ata
b
ef
o
r
e
av
er
ag
i
n
g
to
g
et
th
e
f
in
al
r
e
s
u
lt
s.
T
h
e
r
es
u
lt
o
f
ea
c
h
clas
s
if
ica
tio
n
ca
n
b
e
a
b
i
n
ar
y
d
e
cisi
o
n
p
r
o
b
le
m
b
e
p
o
s
itiv
e
o
r
n
eg
at
iv
e.
T
ab
le
2
s
h
o
w
ed
th
e
r
ep
r
esen
tatio
n
o
f
t
h
e
d
ec
is
io
n
as a
co
n
f
u
s
io
n
m
atr
i
x
in
a
s
tr
u
ct
u
r
e.
T
ab
le
2
.
C
o
n
f
u
s
io
n
m
atr
i
x
A
c
t
u
a
l
X
S
S
A
c
t
u
a
l
N
o
n
-
X
S
S
P
r
e
d
i
c
t
e
d
X
S
S
TP
FP
P
r
e
d
i
c
t
e
d
N
o
n
-
X
S
S
FN
FP
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
0
2
2
-
10
29
1028
Mo
r
eo
v
er
,
th
er
e
ar
e
f
o
u
r
cl
ass
es
in
t
h
e
co
n
f
u
s
io
n
m
atr
i
x
,
th
e
s
e
ar
e
t
h
e
T
r
u
e
P
o
s
itiv
es
w
h
ic
h
s
p
ec
if
ie
s
a
m
o
u
n
t
o
f
XS
S
s
a
m
p
les
th
at
ar
e
co
r
r
ec
tl
y
clas
s
i
f
i
ed
,
th
e
Fals
e
P
o
s
iti
v
es
w
h
ic
h
p
o
r
tr
ay
s
t
h
e
n
u
m
b
er
o
f
No
n
-
XSS
s
a
m
p
le
s
w
r
o
n
g
l
y
clas
s
if
ied
as
XSS,
t
h
e
T
r
u
e
Neg
ati
v
e
th
a
t
in
d
icate
t
h
e
n
u
m
b
er
o
f
co
r
r
ec
tl
y
class
i
f
ied
No
n
-
X
SS
s
a
m
p
les,
an
d
th
e
Fals
e
Ne
g
ati
v
es
th
at
ca
p
tu
r
e
th
e
t
h
e
n
u
m
b
er
o
f
X
S
S
s
a
m
p
le
s
w
r
o
n
g
l
y
class
i
f
ied
as
No
n
-
X
SS
.
W
e
ev
alu
ate
t
h
e
ex
p
er
i
m
e
n
tal
r
es
u
lt
s
o
n
th
e
b
as
i
s
o
f
a
co
n
f
u
s
io
n
m
atr
i
x
w
ith
p
r
ec
is
e
r
ec
all,
F1
s
co
r
e
an
d
th
e
f
o
r
m
u
la
cr
iter
ia
i
n
t
h
e
f
o
llo
w
i
n
g
eq
u
atio
n
s
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
w
a
s
o
b
j
ec
tiv
el
y
ev
alu
ae
d
b
y
co
m
p
ar
in
g
th
e
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
to
XSS
-
attac
k
s
w
it
h
th
e
ea
r
li
er
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
b
y
R
ath
o
r
e
[
5
]
.
W
hi
ch
r
elies o
n
1
0
tr
ad
itio
n
al
ML
m
et
h
o
d
s
to
d
etec
t X
SS
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
f
o
u
n
d
lear
n
i
n
g
m
o
d
el
is
b
etter
t
h
an
ten
ML
alg
o
r
it
h
m
s
i
n
T
ab
le
3
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
ie
v
ed
a
n
ac
cu
r
ac
y
r
ate
o
f
9
9
,
3
%,
r
ec
all
r
ate
o
f
ab
o
u
t
9
9
,
1
%,
F
1
s
co
r
e
o
f
9
9
,
5
%,
an
d
p
r
ec
is
io
n
v
al
u
e
o
f
ab
o
u
t
9
9
,
9
%.
Fig
r
ep
esen
ts
t
h
e
r
ec
eiv
er
o
p
er
atin
g
c
h
ar
ac
ter
is
tic
(
R
OC
)
cu
r
v
e
th
a
t
ca
p
tu
r
ed
th
e
d
etec
tio
n
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
.
T
h
e
o
b
tain
ed
R
O
C
c
u
r
v
e
w
a
s
al
m
o
s
t
p
er
f
ec
t,
ex
h
ib
it
in
g
a
c
u
r
v
e
ar
ea
(
A
UC
)
o
f
0
.
9
5
.
Fu
r
th
er
m
o
r
e,
th
e
co
r
r
esp
o
n
d
in
g
tr
u
e
p
o
s
iti
v
e
r
ate
an
d
f
al
s
e
p
o
s
iti
v
e
r
ate
w
er
e
ab
o
u
t 1
.
0
a
n
d
0
.
0
1
,
r
esp
ec
tv
el
y
.
T
h
is
i
m
p
l
ies
th
at
t
h
e
p
r
o
p
o
s
ed
d
ee
p
-
lear
n
i
n
g
ap
p
r
o
ac
h
in
th
i
s
s
t
u
d
y
w
a
s
e
f
f
ec
t
iv
e
in
XSS
attac
k
s
d
etec
tio
n
,
Fi
g
u
r
e
4.
s
h
o
w
ed
T
h
e
co
m
p
ar
is
o
n
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
i
th
th
e
r
elate
d
w
o
r
k
.
T
ab
le
3
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
f
o
u
n
d
lear
n
in
g
m
o
d
el
T
e
c
h
n
i
q
u
e
s
R
e
c
a
l
l
P
r
e
c
i
si
o
n
F
-
m
e
a
su
r
e
A
c
c
u
r
a
c
y
R
a
n
d
o
mF
o
r
e
st
0
.
9
7
1
0
.
9
7
7
0
.
9
7
4
0
.
9
7
2
A
D
T
r
e
e
0
.
9
7
2
0
.
9
7
0
0
.
9
7
1
0
.
9
7
1
R
a
n
d
o
mS
u
b
S
p
a
c
e
0
.
9
6
9
0
.
9
7
2
0
.
9
7
0
0
.
9
7
0
D
e
c
o
r
a
t
e
0
.
9
6
8
0
.
9
6
5
0
.
9
6
6
0
.
9
6
6
A
d
a
B
o
o
st
.
M
1
0
.
9
6
5
0
.
9
6
4
0
.
9
6
4
0
.
9
6
6
JR
i
p
0
.
9
6
2
0
.
9
6
0
0
.
9
6
1
0
.
9
6
4
N
a
i
v
e
B
a
y
e
s
0
.
9
6
4
0
.
9
6
6
0
.
9
6
5
0
.
9
6
3
S
u
p
p
o
r
t
V
e
c
t
o
r
M
a
c
h
i
n
e
0
.
9
5
8
0
.
9
5
5
0
.
9
5
6
0
.
9
5
8
L
o
g
i
st
i
c
R
e
g
r
e
ssi
o
n
0
.
9
5
5
0
.
9
5
0
0
.
9
5
2
0
.
9
5
6
k
-
N
e
a
r
e
st
N
e
i
g
h
b
o
r
s
0
.
9
5
0
0
.
9
5
4
0
.
9
5
2
0
.
9
4
7
P
r
o
p
o
se
d
a
p
p
r
o
a
c
h
0
.
9
9
1
0
.
9
9
9
0
.
9
9
5
0
.
9
9
3
Fig
u
r
e
4
.
T
h
e
co
m
p
ar
is
o
n
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
w
it
h
th
e
r
elate
d
w
o
r
k
5.
CO
NCLU
SI
O
N
T
h
e
d
e
t
e
c
ti
o
n
o
f
c
r
o
s
s
-
s
it
e
s
c
r
i
p
t
s
i
s
an
es
s
en
t
i
a
l
w
e
b
s
ec
u
r
ity
f
a
c
t
o
r
.
T
h
e
v
u
l
n
e
r
a
b
i
l
ity
o
f
w
e
b
a
p
p
l
i
c
a
t
i
o
n
s
t
o
X
SS
at
t
a
ck
s
c
an
c
au
s
e
t
h
ef
t
o
f
u
s
e
r
in
f
o
r
m
a
tio
n
an
d
o
th
e
r
r
el
a
t
e
d
s
e
cu
r
i
ty
is
s
u
e
s
.
I
n
th
is
p
a
p
e
r
,
a
d
e
e
p
l
e
a
r
n
in
g
-
b
as
e
d
a
p
p
r
o
a
ch
f
o
r
XS
S
at
t
a
ck
s
d
e
t
ec
t
o
n
w
as
p
r
o
p
o
s
e
d
.
F
i
r
s
t
,
w
e
p
r
o
p
o
s
e
d
th
e
r
es
t
o
r
a
ti
o
n
o
f
t
h
e
o
r
i
g
in
al
d
a
t
a
f
r
o
m
th
e
o
b
s
cu
r
e
d
v
e
r
s
i
o
n
a
s
th
is
is
a
n
im
p
o
r
t
a
n
t
p
r
e
p
r
o
c
e
s
s
in
g
s
te
p
;
s
ec
o
n
d
ly
,
w
e
u
t
i
li
z
e
d
w
o
r
d
2
v
e
c
f
o
r
ex
t
r
a
ct
i
o
n
o
f
th
e
s
em
an
t
i
c
f
e
at
u
r
es
o
f
th
e
p
ay
l
o
a
d
s
o
f
XS
S
a
t
ta
ck
s
b
ef
o
r
e
m
a
p
p
in
g
th
em
t
o
v
e
c
t
o
r
s
.
F
in
al
ly
,
w
e
u
s
e
d
C
NN
w
i
th
L
S
T
M
t
o
b
u
i
l
d
c
l
ass
if
ic
a
t
i
o
n
m
o
d
e
ls
.
T
h
e
s
u
g
g
es
t
e
d
s
o
lu
t
i
o
n
w
as
t
e
s
t
e
d
u
s
in
g
1
0
-
f
o
l
d
c
r
o
s
s
-
v
a
l
i
d
at
i
o
n
o
n
th
e
r
e
a
l d
a
t
a
c
o
l
le
c
t
i
o
n
.
T
h
e
e
v
a
lu
a
ti
o
n
o
f
th
e
p
r
p
o
s
e
d
a
p
p
r
o
a
c
h
s
h
o
w
e
d
th
a
t i
t
a
c
h
ie
v
e
d
a
n
a
c
cu
r
a
c
y
l
ev
el
o
f
9
9
,
3
%
,
r
e
c
a
l
l
r
at
e
o
f
9
9
,
1
%
,
p
r
e
ci
s
i
o
n
v
a
lu
e
o
f
9
9
,
9
%
,
an
d
F
1
s
c
o
r
e
o
f
9
9
,
5
%
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
A
h
yb
r
id
o
f CN
N
a
n
d
LS
TM
meth
o
d
s
fo
r
s
ec
u
r
in
g
w
eb
a
p
p
lica
tio
n
a
g
a
in
s
t…
(
R
a
ed
W
a
h
ee
d
K
a
d
h
im
)
1029
RE
F
E
R
E
NC
E
S
[1
]
T
ian
lo
n
g
L
.
,
Yu
Q.
,
L
ian
g
S
.
a
n
d
Jia
n
a
n
Y.,
“
L
o
c
a
te
-
T
h
e
n
-
De
tec
t:
Re
a
l
-
ti
m
e
W
e
b
A
tt
a
c
k
De
tec
ti
o
n
v
ia
A
tt
e
n
ti
o
n
-
b
a
se
d
De
e
p
Ne
u
ra
l
Ne
t
w
o
rk
s”
,
T
we
n
ty
-
Ei
g
h
th
In
ter
n
a
ti
o
n
a
l
J
o
i
n
t
Co
n
fer
e
n
c
e
o
n
Arti
f
icia
l
I
n
telli
g
e
n
c
e
M
a
in
tra
c
k
.
P
p
.
4
7
2
5
-
4
7
3
1
IJCA
I,
2
0
1
8
.
[2
]
Nith
y
a
V
.
,
P
a
n
d
ian
S
.
,
M
a
larv
izh
i
C.
“
A
su
rv
e
y
o
n
d
e
tec
ti
o
n
a
n
d
p
re
v
e
n
ti
o
n
o
f
c
ro
ss
-
site
sc
r
ip
ti
n
g
a
tt
a
c
k
”
,
In
ter
n
a
t
io
n
a
lJo
u
rn
a
l
o
f
S
e
c
u
ri
ty a
n
d
Its
A
p
p
li
c
a
ti
o
n
s
,
v
o
l
.
9
,
n
o
.
3
,
p
p
.
1
3
9
-
1
5
2
,
2
0
1
5
.
[3
]
G
u
p
ta
S
.
,
G
u
p
ta
B.
,
“
Cro
ss
-
S
it
e
S
c
rip
ti
n
g
(XSS
)
a
tt
a
c
k
s
a
n
d
d
e
fe
n
se
m
e
c
h
a
n
is
m
s:
c
las
si
f
ic
a
ti
o
n
a
n
d
sta
te
-
of
-
th
e
-
a
rt”,
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
o
f
S
y
st
e
m A
ss
u
ra
n
c
e
En
g
in
e
e
rin
g
a
n
d
M
a
n
a
g
e
me
n
t
,
v
o
l.
8
,
n
o
.
1
,
p
p
.
5
1
2
-
5
3
0
,
2
0
1
7
.
[4
]
V
ish
n
u
B
A
,
Je
v
it
h
a
K
P
.
,
”
P
r
e
d
ictio
n
o
f
Cro
ss
-
S
i
te
S
c
rip
ti
n
g
A
tt
a
c
k
Us
in
g
M
a
c
h
in
e
L
e
a
rn
in
g
A
l
g
o
rit
h
m
s”
,
Pro
c
e
e
d
in
g
s
o
f
th
e
2
0
1
4
I
n
ter
n
a
t
i
o
n
a
l
Co
n
fer
e
n
c
e
o
n
In
ter
d
isc
ip
li
n
a
ry
Ad
v
a
n
c
e
s
in
Ap
p
l
ied
Co
mp
u
ti
n
g
.
A
CM
,
p
.
5
5
,
2
0
1
4
.
[5
]
Ra
th
o
re
S
,
S
h
a
rm
a
P
K,
P
a
rk
J
H.,
”
X
S
S
Clas
sif
ier:
A
n
Eff
icie
n
t
X
S
S
A
tt
a
c
k
De
tec
ti
o
n
A
p
p
ro
a
c
h
Ba
se
d
o
n
M
a
c
h
in
e
L
e
a
rn
in
g
Clas
sif
ier o
n
S
NSs”
,
J
o
u
rn
a
l
o
f
In
fo
rm
a
t
io
n
Pro
c
e
ss
in
g
S
y
ste
ms
,
v
o
l.
1
3
,
n
o
.
4
,
2
0
1
7
.
[6
]
P
.
L
ik
a
rish
,
E.
L
u
n
g
.
a
n
d
I.
j
o
,
"
Ob
f
u
sc
a
ted
m
a
li
c
io
u
s
jav
a
S
c
ri
p
t
d
e
tec
ti
o
n
u
si
n
g
c
las
s
f
ica
ti
o
n
tec
h
n
iq
u
e
s,"
i
n
Pro
c
e
e
d
in
g
s
o
f
t
h
e
4
th
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
M
a
li
c
io
u
s
a
n
d
U
n
wa
n
ted
S
o
f
twa
re
(
M
AL
W
AR
E)
,
M
o
n
trea
l,
Ca
n
a
d
a
,
p
p
.
4
7
-
5
4
,
2
0
0
9
.
[7
]
M
.
A
.
A
lu
n
e
d
,
a
n
d
F
.
A
li
,
"
M
u
l
ti
p
le
-
p
a
t
h
tes
ti
n
g
f
o
r
c
ro
ss
site
sc
rip
ti
n
g
u
sin
g
g
e
n
e
ti
c
a
lg
o
rit
h
m
s,"
J
o
u
rn
a
l
o
f
S
y
ste
ms
Arc
h
it
e
c
tu
re
,
v
o
l.
6
4
,
p
p
.
5
0
-
6
2
,
2
0
1
6
.
[8
]
C.
H
.
Wan
g
a
n
d
Y.
S
.
Zh
o
u
,
"
A
n
e
w
c
ro
ss
-
site
s
c
rip
ti
o
g
d
e
tec
rio
n
m
e
c
h
a
n
is
m
in
teg
r
a
ted
w
it
h
H
TM
IL
5
a
n
d
CORS
p
ro
p
e
rti
e
s b
y
u
sin
g
b
ro
w
se
r
e
x
ten
sio
n
s,"
in
Pro
c
e
e
d
in
g
s o
f
th
e
2
0
1
6
In
ter
n
a
ti
o
n
a
l
Co
mp
u
ter
S
y
mp
o
si
u
m (
ICS
)
,
Ch
iay
i,
T
a
i
w
a
n
,
p
p
.
2
6
4
-
2
6
9
,
2
0
1
6
.
[9
]
H.
S
h
a
h
riar
a
n
d
M
.
Z
u
lk
e
rn
in
e
,
"
S
2
X
S
2
:
A
S
e
rv
e
r
S
id
e
A
p
p
ro
a
c
h
to
A
u
to
m
a
ti
c
a
ll
y
De
te
c
t
X
S
S
A
tt
a
c
k
s,"
2
0
1
1
IEE
ENi
n
t
h
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
De
p
e
n
d
a
b
le,
A
u
to
n
o
mic
a
n
d
S
e
c
u
re
Co
mp
u
ti
n
g
,
S
y
d
n
e
y
,
NSW
,
p
p
.
7
-
1
4
,
2
0
1
1
.
[1
0
]
N.
S
h
o
n
e
,
T
.
N.
Ng
o
c
,
V
.
D.
P
h
a
i,
a
n
d
Q.
S
h
i,
”
A
d
e
e
p
lea
rn
in
g
a
p
p
ro
a
c
h
t
o
n
e
tw
o
rk
in
tr
u
sio
n
d
e
tec
ti
o
n
,
'
'
IEE
E
T
ra
n
s.
Eme
rg
.
T
o
p
ics
C
o
mp
u
t.
I
n
tell.
,
v
o
l.
2
,
n
o
.
1
,
p
.
4
1
5
0
,
2
0
1
8
.
[1
1
]
A
.
M
.
V
a
rto
u
n
i,
S
.
S
.
Ka
sh
i,
a
n
d
M
.
T
e
sh
n
e
h
lab
,
“
A
n
a
n
o
m
a
ly
d
e
tec
ti
o
n
m
e
th
o
d
to
d
e
tec
t
We
b
a
tt
a
c
k
s
u
sin
g
sta
c
k
e
d
a
u
to
-
e
n
c
o
d
e
r,
'
'
in
Pr
o
c
.
6
th
Ira
n
ia
n
J
o
in
t
Co
n
g
r.
F
u
zz
y
In
tell
.
S
y
st.
(
CFIS
)
,
p
p
.
1
3
1
-
1
3
4
,
2
0
1
8
.
[1
2
]
K.
G
ro
ss
e
,
N.
P
a
p
e
rn
o
t
,
a
n
d
P
.
M
a
n
o
h
a
ra
n
,
`
`A
d
v
e
rs
a
rial
e
x
a
m
p
les
f
o
r
m
a
l
w
a
r
e
d
e
tec
ti
o
n
,
'
'
in
Pro
c
.
Eu
r.
S
y
mp
.
Res
.
Co
mp
u
t
.
S
e
c
u
r.
Ch
a
m,
S
w
it
ze
rla
n
d
:
S
p
rin
g
e
r
,
p
p
.
6
2
7
9
,
2
0
1
7
.
[1
3
]
S.
Ra
th
o
re
,
P
.
K.
S
h
a
rm
a
,
a
n
d
J.
H.
P
a
rk
,
“
X
S
S
Clas
sif
ier:
A
n
e
fficie
n
t
X
S
S
a
tt
a
c
k
d
e
tec
ti
o
n
a
p
p
r
o
a
c
h
b
a
se
d
o
n
m
a
c
h
in
e
lea
rn
in
g
c
las
sif
ier o
n
S
N
S
s,”
J
.
In
f.
Pro
c
e
ss
.
S
y
st.
,
v
o
l.
1
3
,
n
o
.
4
,
p
p
.
1
0
1
4
-
1
0
2
8
,
2
0
1
7
.
[1
4
]
S
to
k
e
s,
Ja
c
k
W
.
,
Ra
k
sh
it
Ag
ra
wa
l,
a
n
d
G
e
o
ff
M
c
Do
n
a
ld
.
"
Ne
u
ra
l
c
las
si
f
ica
ti
o
n
o
f
m
a
li
c
io
u
s
sc
rip
ts
:
A
stu
d
y
w
it
h
jav
a
sc
rip
t
a
n
d
v
b
sc
rip
t.
"
a
rXiv
p
r
e
p
rin
t
a
r
X
iv
:1
8
0
5
.
0
5
6
0
3
,
2
0
1
8
.
[1
5
]
Ya
n
p
e
n
g
C.
,
Ju
n
j
ieC.,
Jia
n
w
e
i
H
.
,
“
A
S
u
rv
e
y
o
n
X
S
S
A
tt
a
c
k
D
e
tec
ti
o
n
a
n
d
P
re
v
e
n
ti
o
n
in
W
e
b
A
p
p
li
c
a
ti
o
n
s”
,
ICM
L
C:
Pro
c
e
e
d
in
g
s
o
f
t
h
e
1
2
th
In
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
M
a
c
h
in
e
L
e
a
rn
i
n
g
a
n
d
Co
mp
u
ti
n
g
Fe
b
ru
a
ry
,
A
CM
,
p
p
.
4
4
3
-
4
4
9
,
2
0
2
0
.
[1
6
]
M
u
ru
g
a
n
,
P
u
sh
p
a
ra
ja.
"
I
m
p
le
m
e
n
tatio
n
o
f
d
e
e
p
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
in
m
u
lt
i
-
c
las
s
c
a
t
e
g
o
rica
l
i
m
a
g
e
c
las
si
f
ica
ti
o
n
.
"
a
rX
iv
p
re
p
r
in
t
a
r
Xiv
:1
8
0
1
.
0
1
3
9
7
,
2
0
1
8
.
[1
7
]
F
a
n
g
,
Yo
n
g
,
Ya
n
g
L
i,
L
ian
g
L
iu
,
a
n
d
C
h
e
n
g
Hu
a
n
g
.
"
De
e
p
X
S
S
:
Cro
ss
site
sc
rip
ti
n
g
d
e
tec
ti
o
n
b
a
se
d
o
n
d
e
e
p
lea
rn
in
g
.
"
In
Pro
c
e
e
d
in
g
s
o
f
th
e
2
0
1
8
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ti
n
g
a
n
d
Arti
fi
c
ia
l
I
n
tell
ig
e
n
c
e
,
p
p
.
4
7
-
51,
2
0
1
8
.
[1
8
]
JIA
N
G
,
F
e
n
g
,
e
t
a
l.
“
D
e
e
p
lea
r
n
in
g
b
a
se
d
m
u
lt
i
-
c
h
a
n
n
e
l
in
telli
g
e
n
t
a
tt
a
c
k
d
e
tec
ti
o
n
f
o
r
d
a
ta
se
c
u
rit
y
”
,
IEE
E
tra
n
sa
c
ti
o
n
s
o
n
S
u
sta
i
n
a
b
le Co
m
p
u
ti
n
g
,
2
0
1
8
.
[1
9
]
Ca
stig
li
o
n
e
,
A
rc
a
n
g
e
lo
,
F
lo
ri
n
P
o
p
,
M
a
ss
im
o
F
icc
o
,
a
n
d
F
ra
n
c
e
sc
o
P
a
lm
ieri,
e
d
s.
“
Cy
b
e
rsp
a
c
e
S
a
fe
t
y
a
n
d
S
e
c
u
rit
y
:
1
0
t
h
In
tern
a
ti
o
n
a
l
S
y
m
p
o
siu
m
”
,
CS
S
2
0
1
8
,
Am
a
l
f
i,
Ital
y
,
O
c
t
o
b
e
r
2
9
-
3
1
,
2
0
1
8
,
P
r
o
c
e
e
d
in
g
s,
v
o
l.
1
1
1
6
1
.
S
p
rin
g
e
r
,
2
0
1
8
.
[2
0
]
M
ik
o
lo
v
,
T
o
m
a
s,
Ka
i
Ch
e
n
,
G
re
g
Co
rra
d
o
,
a
n
d
Je
ff
re
y
D
e
a
n
.
"
Eff
icie
n
t
e
stim
a
ti
o
n
o
f
w
o
rd
re
p
re
se
n
tatio
n
s
i
n
v
e
c
to
r
sp
a
c
e
.
"
a
rX
iv
p
re
p
rin
t
a
r
X
i
v
:1
3
0
1
.
3
7
8
1
,
2
0
1
3
.
[2
1
]
Z.
H.
S
a
li
h
,
G
.
T
.
H
a
sa
n
,
a
n
d
M
.
A
.
M
o
h
a
m
m
e
d
,
"
In
v
e
stig
a
te
a
n
d
a
n
a
l
y
z
e
th
e
le
v
e
ls
o
f
e
lec
tro
m
a
g
n
e
ti
c
ra
d
iatio
n
s
e
m
it
ted
f
ro
m
u
n
d
e
rg
ro
u
n
d
p
o
w
e
r
c
a
b
les
e
x
ten
d
e
d
in
m
o
d
e
rn
c
it
ies
,
"
in
2
0
1
7
9
t
h
In
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
El
e
c
tro
n
ics
,
Co
m
p
u
ter
s a
n
d
Arti
fi
c
ia
l
In
tell
ig
e
n
c
e
(
ECA
I)
,
2
0
1
7
.
[2
2
]
Z.
H.
S
a
li
h
,
G
.
T
.
Ha
sa
n
,
M
.
A
.
M
o
h
a
m
m
e
d
,
M
.
A
.
S
.
Kli
b
,
A
.
H.
A
li
,
a
n
d
R.
A
.
Ib
ra
h
im
,
"
S
tu
d
y
th
e
Eff
e
c
t
o
f
In
teg
ra
ti
n
g
th
e
S
o
lar
E
n
e
rg
y
S
o
u
rc
e
o
n
S
tab
il
it
y
o
f
El
e
c
tri
c
a
l
Distrib
u
t
io
n
S
y
ste
m
,
"
in
2
0
1
9
2
2
n
d
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
n
tro
l
S
y
ste
ms
a
n
d
C
o
mp
u
ter
S
c
ien
c
e
(
CS
CS
)
,
p
p
.
4
4
3
-
4
4
7
,
2
0
1
9
.
[2
3
]
N.
D.
Zak
i
,
N.
Y.
Ha
sh
im
,
Y.
M
.
M
o
h
iald
e
n
,
M
.
A
.
M
o
h
a
m
m
e
d
,
T
.
S
u
ti
k
n
o
,
a
n
d
A
.
H.
A
li
,
"
A
re
a
l
-
ti
m
e
b
ig
d
a
ta
se
n
ti
m
e
n
t
a
n
a
l
y
sis
f
o
r
iraq
i
tw
e
e
ts
u
sin
g
sp
a
rk
stre
a
m
in
g
,
"
Bu
ll
e
ti
n
o
f
El
e
c
trica
l
E
n
g
i
n
e
e
rin
g
a
n
d
In
fo
rm
a
ti
c
s
(
BE
EI)
,
v
o
l.
9
,
p
p
.
1
4
1
1
-
1
4
1
9
,
2
0
2
0
.
[2
4
]
M
.
A
.
M
o
h
a
m
m
e
d
,
R.
A
.
Ha
s
a
n
,
M
.
A
.
A
h
m
e
d
,
N.
T
a
p
u
s,
M
.
A
.
S
h
a
n
a
n
,
M
.
K.
Kh
a
lee
l,
e
t
a
l.
,
"
A
F
o
c
a
l
lo
a
d
b
a
lan
c
e
r
b
a
se
d
a
lg
o
rit
h
m
f
o
r
tas
k
a
ss
i
g
n
m
e
n
t
in
c
lo
u
d
e
n
v
iro
n
m
e
n
t,
"
in
2
0
1
8
1
0
th
I
n
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
El
e
c
tro
n
ics
,
Co
m
p
u
ter
s a
n
d
Arti
fi
c
ia
l
In
tell
ig
e
n
c
e
(
E
CAI)
,
p
p
.
1
-
4
,
2
0
1
8
.
[2
5
]
M
.
A
.
M
o
h
a
m
m
e
d
,
A
.
A
.
Ka
m
il
,
R.
A
.
Ha
sa
n
,
a
n
d
N.
T
a
p
u
s,
"
A
n
Ef
fe
c
ti
v
e
Co
n
tex
t
S
e
n
siti
v
e
Off
l
o
a
d
in
g
S
y
ste
m
f
o
r
M
o
b
il
e
Clo
u
d
En
v
iro
n
m
e
n
ts
u
sin
g
S
u
p
p
o
rt
V
a
lu
e
-
b
a
se
d
Clas
sif
ic
a
ti
o
n
,
"
S
c
a
la
b
le
Co
mp
u
ti
n
g
:
Pra
c
ti
c
e
a
n
d
Exp
e
rie
n
c
e
,
v
o
l.
2
0
,
p
p
.
6
8
7
-
6
9
8
,
2
0
1
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.