I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
3
8
,
No
.
1
,
A
p
r
il
20
2
5
,
p
p
.
535
~
545
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
3
8
.
i
1
.
pp
535
-
5
4
5
535
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Experime
ntal res
ea
rch on te
x
t
CA
PTCH
A of
fine
-
g
ra
ined
security f
ea
tures
Q
ia
n Wa
ng
1,
3
,
Sh
a
f
a
f
I
bra
him
1
,
Xing
Wa
n
2
,
Z
a
inu
ra
I
dr
us
1
1
C
o
l
l
e
g
e
o
f
C
o
m
p
u
t
i
n
g
,
I
n
f
o
r
mat
i
c
s
a
n
d
M
a
t
h
e
m
a
t
i
c
s
,
U
n
i
v
e
r
si
t
i
Te
k
n
o
l
o
g
i
M
A
R
A
,
S
h
a
h
A
l
a
m,
M
a
l
a
y
s
i
a
2
F
a
c
u
l
t
y
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
i
T
e
k
n
o
l
o
g
i
M
A
R
A
,
S
h
a
h
A
l
a
m,
M
a
l
a
y
si
a
3
S
c
h
o
o
l
o
f
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
L
e
s
h
a
n
V
o
c
a
t
i
o
n
a
l
a
n
d
T
e
c
h
n
i
c
a
l
C
o
l
l
e
g
e
,
Le
s
h
a
n
,
C
h
i
n
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ju
n
4
,
2
0
2
4
R
ev
is
ed
Oct
15
,
202
4
Acc
ep
ted
Oct
30
,
2
0
2
4
CAPT
CHA
is
a
c
y
b
e
rse
c
u
rit
y
m
e
a
su
re
th
a
t
d
isti
n
g
u
ish
e
s
b
e
twe
e
n
h
u
m
a
n
s
a
n
d
a
u
to
m
a
ted
sc
rip
ts.
Re
se
a
rc
h
e
rs
h
a
v
e
e
m
p
lo
y
e
d
v
a
ri
o
u
s
se
c
u
rit
y
fe
a
tu
re
s
to
th
wa
rt
a
u
t
o
m
a
ted
p
ro
g
ra
m
id
e
n
ti
fica
ti
o
n
b
y
h
a
c
k
e
rs.
Ho
we
v
e
r,
p
re
v
io
u
s
re
se
a
rc
h
o
n
th
e
a
tt
a
c
k
re
sista
n
c
e
o
f
CAPT
CHA
s
h
a
s
u
se
d
ro
u
g
h
l
y
q
u
a
n
t
it
a
ti
v
e
a
n
a
l
y
sis
in
ste
a
d
o
f
a
fin
e
-
g
ra
in
q
u
a
n
ti
tati
v
e
stu
d
y
.
Th
is
stu
d
y
imp
lem
e
n
ted
c
o
m
p
a
ra
ti
v
e
e
x
p
e
r
ime
n
ts
b
a
se
d
o
n
CAPT
CHA
re
c
o
g
n
i
ti
o
n
a
lg
o
rit
h
m
s
to
fi
n
d
th
e
b
e
st
-
m
ix
e
d
se
c
u
rit
y
fe
a
tu
re
s
.
A
m
u
lt
i
-
sta
g
e
b
e
st
p
a
ra
m
e
ter
se
lec
ti
o
n
(M
B
P
S
)
m
e
c
h
a
n
ism
wa
s
p
ro
p
o
se
d
in
t
h
is
st
u
d
y
.
Ex
p
e
rime
n
t
re
su
l
ts
in
d
ica
ted
t
h
a
t
m
ix
e
d
se
c
u
rit
y
fe
a
t
u
re
s
o
f
“
o
v
e
rl
a
p
+
sc
a
le
+
ro
tate
+
b
g
(b
a
c
k
g
ro
u
n
d
)”
we
re
th
e
b
e
st,
wit
h
a
n
a
v
e
ra
g
e
m
a
c
h
in
e
re
c
o
g
n
it
i
o
n
a
c
c
u
ra
c
y
o
f
o
n
ly
4
.
8
1
%
.
T
h
e
c
o
n
tras
t
e
x
p
e
ri
m
e
n
t
re
su
lt
il
lu
stra
ted
th
a
t
t
h
e
a
n
ti
-
a
tt
a
c
k
a
b
il
it
y
o
f
m
i
x
e
d
se
c
u
rit
y
fe
a
tu
re
s
wa
s
b
e
tt
e
r
th
a
n
a
d
d
i
n
g
a
d
v
e
rsa
rial
n
o
ise
,
wi
th
m
a
c
h
i
n
e
re
c
o
g
n
it
io
n
a
c
c
u
ra
c
y
d
e
c
re
a
se
d
b
y
2
.
2
%
.
M
o
re
o
v
e
r,
b
y
i
n
v
e
st
ig
a
ti
n
g
t
h
e
e
ffica
c
y
o
f
se
c
u
rit
y
fe
a
tu
re
p
a
ra
m
e
ters
,
th
is
st
u
d
y
p
ro
v
i
d
e
s
p
ra
c
ti
c
a
l
g
u
i
d
e
li
n
e
s
fo
r
d
e
si
g
n
in
g
ro
b
u
st
CAPT
CHA
s.
F
u
rth
e
rm
o
re
,
th
is
st
u
d
y
a
lso
p
re
se
n
ts
v
a
lu
a
b
le
in
si
g
h
t
s
in
t
o
th
e
se
c
u
rit
y
o
f
ima
g
e
g
e
n
e
ra
ti
o
n
tec
h
n
o
l
o
g
y
.
K
ey
w
o
r
d
s
:
An
ti
-
r
ec
o
g
n
itio
n
C
APT
C
HA
g
en
er
atio
n
C
y
b
er
s
ec
u
r
ity
Secu
r
ity
m
ec
h
an
is
m
s
User
-
f
r
ien
d
lin
ess
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
:
Z
ain
u
r
a
I
d
r
u
s
C
o
lleg
e
o
f
C
o
m
p
u
tin
g
,
I
n
f
o
r
m
atics a
n
d
Ma
th
em
atics,
Un
iv
er
s
iti T
ek
n
o
lo
g
i M
AR
A
Sh
ah
Alam
,
Ma
lay
s
ia
E
m
ail:
za
in
u
r
a@
tm
s
k
.
u
itm
.
ed
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
Per
s
o
n
al
in
f
o
r
m
atio
n
s
ec
u
r
it
y
is
a
cr
u
cial
is
s
u
e
in
cy
b
e
r
s
ec
u
r
ity
,
esp
ec
ially
with
th
e
s
u
r
g
e
in
web
s
ites
an
d
in
ter
n
et
u
s
er
s
,
esp
ec
ially
in
h
ig
h
ly
d
e
v
elo
p
e
d
d
ee
p
lear
n
in
g
tech
n
o
lo
g
y
[
1
]
,
[
2
]
.
C
o
m
p
letely
au
to
m
ated
p
u
b
lic
tu
r
in
g
test
to
tell
co
m
p
u
ter
s
an
d
h
u
m
an
s
ap
ar
t
(
C
APTCHA
)
was
in
itia
l
ly
p
r
o
p
o
s
ed
in
[
3
]
.
C
APT
C
HA
i
s
a
to
o
l
th
at
ef
f
ec
tiv
ely
d
is
tin
g
u
is
h
es
b
etwe
en
au
to
m
ated
p
r
o
g
r
am
s
an
d
h
u
m
an
u
s
er
s
,
h
elp
in
g
to
b
lo
ck
m
alicio
u
s
r
eg
is
tr
atio
n
s
ch
em
es
an
d
s
af
eg
u
ar
d
u
s
er
d
a
ta.
Vis
u
al
-
b
ased
C
APT
C
HA
i
s
b
r
o
ad
ly
u
tili
ze
d
to
p
r
o
tect
u
s
er
s
'
p
er
s
o
n
al
in
f
o
r
m
atio
n
s
ec
u
r
ity
i
n
s
h
o
p
p
in
g
,
g
a
m
es,
an
d
v
ar
i
o
u
s
web
s
ites
'
r
eg
is
tr
atio
n
an
d
lo
g
i
n
.
T
ex
t
-
b
ased
C
APTCHA
is
wi
d
ely
u
s
ed
o
n
web
s
ites
to
r
e
g
is
ter
o
r
lo
g
in
b
ec
a
u
s
e
o
f
its
lo
w
co
s
t
a
n
d
r
ich
d
iv
er
s
ity
.
T
ak
e
f
o
u
r
E
n
g
lis
h
c
h
ar
ac
ter
s
(
in
clu
d
in
g
ca
s
e
-
in
s
e
n
s
itiv
e
letter
s
an
d
d
ig
its
)
a
s
an
ex
am
p
le,
an
d
th
er
e
ar
e
3
6
4
or
ap
p
r
o
x
i
m
ately
1
.
7
m
illi
o
n
p
o
s
s
ib
le
co
m
b
in
atio
n
s
[
4
]
.
Ho
wev
er
,
a
n
ad
v
an
tag
e
in
q
u
a
n
tity
is
n
o
t
t
h
e
co
r
e
s
o
lu
tio
n
o
f
C
APTCHA;
b
r
u
te
f
o
r
ce
an
d
s
eq
u
en
tial
s
ea
r
ch
in
g
ca
n
ea
s
ily
b
r
ea
k
th
e
s
im
p
le
tex
t
C
APT
C
HA
[
5
]
.
Pre
v
io
u
s
C
APTC
H
A
r
esear
ch
f
o
cu
s
ed
o
n
f
iv
e
s
ec
u
r
ity
m
ec
h
an
is
m
s
,
in
clu
d
in
g
v
ar
io
u
s
ty
p
es
o
f
C
APTCH
A
,
co
m
p
lex
im
ag
e
id
e
n
tific
atio
n
,
ad
v
an
ce
d
im
ag
e
g
en
er
atio
n
tech
n
iq
u
es
,
ad
d
e
d
p
er
tu
r
b
atio
n
s
,
an
d
in
teg
r
a
tio
n
th
eo
r
ies
f
r
o
m
o
th
er
d
is
cip
lin
es
to
p
r
o
tect
C
APT
C
HA.
F
ir
s
tly
,
v
ar
io
u
s
ty
p
es
o
f
C
APT
C
HA
,
in
clu
d
in
g
m
ath
,
d
o
ts
,
h
o
llo
w
,
an
d
two
-
lay
er
s
,
wer
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
1
,
Ap
r
il
20
2
5
:
5
3
5
-
5
4
5
536
in
tr
o
d
u
ce
d
[
6
]
−
[
8
]
.
Seco
n
d
ly
,
o
th
er
co
m
p
lex
im
ag
e
id
e
n
tific
atio
n
,
in
clu
d
in
g
g
eo
m
etr
ic
s
h
ap
es,
f
ac
e
im
ag
e
s,
p
u
zz
le
s
an
d
s
lid
er
s
,
wer
e
p
r
o
p
o
s
ed
in
C
APTCH
A
g
en
er
atio
n
m
o
d
els
[
9
]
−
[
1
2
]
.
T
h
i
r
d
ly
,
r
e
s
ea
r
ch
er
s
ex
p
lo
r
ed
ad
v
an
ce
d
im
ag
e
g
en
er
atio
n
t
ec
h
n
iq
u
es
,
i
n
clu
d
in
g
s
ty
le
tr
a
n
s
f
er
an
d
g
en
er
ativ
e
ad
v
e
r
s
ar
ial
n
etwo
r
k
(
GAN)
tech
n
o
lo
g
y
,
to
d
esig
n
r
o
b
u
s
t
C
APTC
HA
s
[
1
3
]
−
[
1
7
]
.
Ot
h
er
s
ad
d
ed
p
er
tu
r
b
atio
n
s
to
th
e
ir
C
APTCHA
g
en
er
atio
n
m
o
d
e
l,
s
u
ch
as
ad
v
er
s
ar
ial
ex
am
p
les
to
en
h
an
c
e
s
ec
u
r
ity
[
1
8
]
−
[
2
2
]
.
L
astl
y
,
in
teg
r
atio
n
t
h
eo
r
ies
f
r
o
m
o
th
e
r
d
is
cip
lin
es,
s
u
ch
as
v
is
u
al
r
ea
s
o
n
in
g
,
s
em
an
tic
u
n
d
er
s
tan
d
i
n
g
,
an
d
co
g
n
itiv
e
ab
ilit
y
tech
n
o
lo
g
y
,
wer
e
also
in
tr
o
d
u
ce
d
in
n
o
v
at
iv
el
y
[
2
3
]
−
[
2
5
]
.
All
th
ese
s
e
cu
r
ity
m
ec
h
an
is
m
s
ca
n
m
ak
e
C
APTCH
A
m
o
r
e
s
ec
u
r
e.
Ho
wev
e
r
,
s
in
ce
im
ag
e
-
tex
t
C
APTCHA
s
ap
p
ly
co
m
p
lex
im
a
g
es,
th
is
ch
allen
g
es
m
ac
h
in
e
r
ec
o
g
n
itio
n
an
d
b
r
in
g
s
d
if
f
icu
lties
to
ac
tu
al
h
u
m
an
u
s
er
s
.
Mo
r
eo
v
e
r
,
im
a
g
e
an
d
im
ag
e
-
tex
t CAP
T
C
HA
s
wil
l o
cc
u
p
y
m
o
r
e
s
to
r
ag
e
s
p
ac
e
an
d
h
av
e
a
s
lo
w
er
p
r
o
ce
e
d
in
g
s
p
ee
d
th
an
te
x
t
C
APT
C
HAs.
I
n
co
n
tr
ast
,
t
ex
t
-
b
ased
C
APTCHA
co
n
s
is
ts
o
f
ch
ar
ac
ter
s
o
r
d
ig
its
,
is
lo
w
c
o
s
t,
ea
s
y
to
d
ep
lo
y
,
a
n
d
q
u
ick
ly
g
en
er
ated
,
an
d
h
as
b
e
co
m
e
th
e
p
r
io
r
c
h
o
ice
o
f
m
o
s
t
web
s
ite
o
wn
er
s
.
Mo
s
t
im
p
o
r
tan
tly
,
it
h
as
h
ig
h
p
o
ten
tial
co
m
m
er
cial
v
alu
e
.
Ho
wev
er
,
th
e
p
r
o
b
lem
is
th
a
t
m
o
s
t
attac
k
er
s
o
r
m
alicio
u
s
b
o
ts
u
s
e
o
p
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
(
OC
R
)
,
ch
ar
ac
ter
s
eg
m
en
tatio
n
,
an
d
en
d
-
to
-
e
n
d
d
ee
p
lear
n
in
g
tec
h
n
o
lo
g
ies
to
attac
k
tex
t
-
b
ased
C
APTCH
A
[
2
6
]
,
[
2
7
]
.
T
h
e
r
ef
o
r
e
,
r
esear
ch
er
s
in
tr
o
d
u
ce
d
v
ar
io
u
s
s
ec
u
r
ity
s
ch
e
m
es.
Ye
e
t
a
l
.
[
2
8
]
e
v
a
l
u
a
t
e
d
3
3
t
e
x
t
-
b
a
s
e
d
C
A
P
T
C
H
A
s
d
e
p
l
o
y
e
d
in
t
h
e
a
c
t
u
a
l
we
b
s
it
e
s
.
T
h
ei
r
ex
p
e
r
i
m
e
n
t
r
e
s
u
l
ts
s
h
o
w
e
d
t
h
at
C
AP
T
C
HA
p
e
r
f
o
r
m
e
d
w
e
l
l
r
e
g
a
r
d
i
n
g
s
e
c
u
r
i
t
y
a
n
d
u
s
a
b
i
l
it
y
.
S
i
m
il
a
r
ly
,
W
a
n
g
e
t
a
l
.
[
2
9
]
d
is
cu
s
s
ed
th
e
r
o
b
u
s
tn
ess
o
f
C
APTCHA
w
ith
d
if
f
er
en
t
s
ec
u
r
ity
s
ch
em
es
.
T
h
ey
also
d
is
cu
s
s
ed
th
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
o
f
u
n
ex
p
lo
r
ed
m
ac
h
in
e
attac
k
s
.
Sh
i
et
al
.
[
3
0
]
s
u
m
m
ar
ize
d
1
2
s
tan
d
a
r
d
s
ec
u
r
it
y
f
ea
tu
r
es
o
f
tex
t
C
APT
C
HA
an
d
u
s
ed
cy
cle
GAN
as
th
eir
attac
k
m
eth
o
d
to
s
y
n
th
esize
ca
p
tch
as,
ac
h
i
ev
in
g
g
o
o
d
r
esu
lts
.
Ho
wev
er
,
th
ey
d
i
d
n
o
t
in
te
g
r
ate
th
e
g
en
er
atio
n
ef
f
ec
t
o
f
ea
ch
s
ec
u
r
ity
f
ea
tu
r
e
u
n
d
er
d
if
f
er
en
t
v
alu
es.
Ma
ts
u
u
r
a
et
a
l
.
[
3
1
]
u
s
ed
s
p
atial
s
m
o
o
th
in
g
an
d
ad
v
er
s
ar
i
al
ex
am
p
les
to
g
e
n
er
ate
r
o
b
u
s
t
C
APT
C
HA,
an
d
t
h
e
i
r
e
x
p
e
r
i
m
e
n
t
r
e
s
u
lt
s
il
l
u
s
t
r
a
t
e
d
t
h
a
t
t
h
e
i
r
m
e
t
h
o
d
w
a
s
e
f
f
e
c
t
i
v
e
.
H
o
w
e
v
e
r
,
t
h
e
d
is
a
d
v
a
n
ta
g
e
o
f
t
h
e
i
r
m
e
t
h
o
d
w
a
s
t
h
at
i
t
w
a
s
a
w
h
i
te
-
b
o
x
a
t
ta
c
k
;
t
h
e
a
tt
a
c
k
e
r
s
o
r
b
o
ts
s
h
o
u
ld
k
n
o
w
t
h
e
s
t
r
u
c
t
u
r
e
o
f
t
h
e
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
.
All in
all,
th
er
e
ar
e
p
r
o
b
lem
s
with
th
e
ex
is
t
in
g
C
APTCHA
r
esear
ch
,
as f
o
llo
ws
:
−
T
h
eir
r
esear
ch
o
n
ly
p
r
o
v
id
ed
a
co
ar
s
e
-
g
r
ain
e
d
q
u
an
titativ
e
an
aly
s
is
in
s
tead
o
f
a
f
i
n
e
-
g
r
ai
n
ed
q
u
an
titativ
e
ass
es
s
m
en
t.
R
esear
c
h
er
s
o
n
ly
co
m
p
a
r
ed
t
h
e
o
v
er
all
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
,
n
o
t
t
h
e
ef
f
ec
t
o
f
s
ec
u
r
ity
m
ea
s
u
r
es
.
−
T
h
e
y
d
id
n
o
t in
v
esti
g
ate
h
o
w
t
h
e
s
ec
u
r
ity
m
ec
h
a
n
is
m
s
af
f
ec
t th
e
an
ti
-
attac
k
ab
ilit
y
.
−
T
h
e
p
r
ev
io
u
s
liter
atu
r
e
d
id
n
o
t
co
m
p
a
r
e
th
e
an
ti
-
attac
k
ab
il
ity
b
etwe
en
s
ec
u
r
ity
f
ea
tu
r
es
an
d
a
d
v
er
s
ar
ial
ex
am
p
les.
T
h
u
s
,
T
HE
co
n
t
r
ib
u
tio
n
s
ar
e
a
s
f
o
llo
ws:
−
I
m
p
lem
en
t a
f
in
e
-
g
r
ain
ed
q
u
a
n
titativ
e
ass
es
s
m
en
t
o
f
th
e
s
ec
u
r
ity
m
ec
h
a
n
is
m
s
o
f
th
e
tex
t
C
APT
C
HA
.
−
E
x
p
lo
r
e
a
n
ti
-
attac
k
s
ec
u
r
ity
m
ea
s
u
r
es
an
d
th
e
im
p
ac
t
o
f
th
e
m
.
−
C
o
n
s
tr
u
ct
a
co
n
tr
ast
ex
p
er
im
en
t
to
co
m
p
ar
e
th
e
a
n
ti
-
attac
k
ab
ilit
y
o
f
s
ec
u
r
ity
f
ea
tu
r
es
an
d
ad
v
e
r
s
ar
ial
ex
am
p
les m
eth
o
d
.
−
Pro
v
id
e
a
r
o
b
u
s
t CAP
T
C
HA
g
en
er
atio
n
m
eth
o
d
a
n
d
o
f
f
er
p
r
ac
tical
in
s
ig
h
t in
to
im
ag
e
s
ec
u
r
ity
.
L
astl
y
,
t
h
e
s
tr
u
ct
u
r
e
f
o
llo
ws:
a
b
r
ief
r
esear
ch
b
ac
k
g
r
o
u
n
d
is
p
r
o
v
id
ed
in
s
ec
tio
n
1
,
an
d
th
e
p
r
o
p
o
s
ed
m
eth
o
d
is
o
u
tlin
ed
in
s
ec
tio
n
2
.
T
h
e
la
s
t s
ec
tio
n
is
th
e
r
esu
lt
an
d
d
is
cu
s
s
io
n
.
2.
M
E
T
H
O
D
Th
e
s
ec
tio
n
d
escr
ib
es
th
e
f
r
am
ewo
r
k
o
f
th
e
ev
al
u
atio
n
m
o
d
el
f
o
r
f
in
e
-
g
r
ain
ed
s
ec
u
r
ity
f
ea
tu
r
es.
T
h
e
f
ir
s
t
p
r
o
ce
s
s
b
eg
an
b
y
ex
p
lo
r
in
g
ten
s
ec
u
r
ity
f
ea
tu
r
es
in
d
etail.
T
h
ese
s
ec
u
r
ity
f
ea
tu
r
es
ar
e
u
s
ed
in
th
e
f
o
llo
win
g
C
APTCHA
g
en
er
at
io
n
p
r
o
ce
s
s
.
Seco
n
d
ly
,
“Bas
e
C
APT
C
HA
”
was
g
en
er
ated
t
o
g
u
a
r
an
tee
all
t
h
e
C
APT
C
HA
d
atase
ts
u
n
d
er
th
e
s
am
e
b
aselin
e.
T
h
ir
d
ly
,
a
cr
u
cial
s
cr
ee
n
in
g
p
r
o
ce
s
s
ca
lled
th
e
m
u
lti
-
s
tag
e
b
est
p
ar
am
eter
s
s
elec
tio
n
(
MBP
S)
m
ec
h
an
is
m
in
clu
d
ed
t
h
e
b
est
s
in
g
le
s
ec
u
r
ity
f
ea
t
u
r
es
s
elec
tio
n
an
d
b
est
-
m
ix
ed
s
ec
u
r
ity
f
ea
tu
r
es
s
elec
tio
n
s
ta
g
es.
T
h
is
s
tag
e
i
s
o
n
e
o
f
th
e
m
o
s
t
cr
itical
to
f
ilter
r
o
b
u
s
t
s
e
cu
r
ity
f
ea
tu
r
es.
Af
ter
u
s
in
g
th
is
MBP
S
m
ec
h
an
is
m
,
r
o
b
u
s
t
m
ix
ed
s
ec
u
r
ity
f
ea
t
u
r
es
will
b
e
s
elec
ted
.
T
h
en
,
r
o
b
u
s
t
C
APTCH
A
d
atasets
wer
e
g
en
er
ated
b
y
u
s
in
g
th
ese
b
est
-
p
er
f
o
r
m
an
ce
s
e
cu
r
ity
f
ea
tu
r
es.
L
astl
y,
a
co
n
tr
ast
ex
p
er
im
en
t
was
co
n
s
tr
u
cted
.
Fig
u
r
e
1
illu
s
tr
at
es
th
e
o
v
er
all
f
r
am
ewo
r
k
f
o
r
f
in
e
-
g
r
ain
e
d
s
ec
u
r
ity
f
ea
tu
r
e
ev
alu
atio
n
.
Am
o
n
g
th
e
p
r
o
ce
s
s
es,
th
e
MBP
S
m
ec
h
an
is
m
is
a
cr
u
cial
s
tep
th
at
s
elec
ts
th
e
b
est
p
ar
am
eter
s
f
o
r
ea
ch
s
u
b
item
o
f
th
e
s
ec
u
r
ity
f
ea
tu
r
es.
C
o
n
tr
asti
v
e
ex
p
er
im
en
ts
ar
e
c
o
n
d
u
cted
to
ev
alu
ate
th
e
p
er
f
o
r
m
a
n
ce
o
f
C
APT
C
HA
u
s
in
g
a
co
m
b
in
atio
n
o
f
th
e
b
est s
ec
u
r
i
ty
f
ea
tu
r
es.
2
.
1
.
Securit
y
f
e
a
t
ures e
x
plo
r
a
t
io
n
I
n
s
p
ir
ed
b
y
Ye
et
a
l
.
[
2
8
]
,
W
an
g
et
a
l
.
,
[
2
9
]
,
we
s
u
m
m
ar
ized
an
d
b
r
o
ad
e
n
ed
ten
ty
p
ical
s
ec
u
r
ity
f
ea
tu
r
es
b
esid
es
th
eir
p
r
o
p
o
s
ed
o
n
es.
T
h
e
s
ec
u
r
ity
f
ea
tu
r
es
wer
e
d
is
co
v
er
e
d
f
r
o
m
C
APTCHA
's
f
am
o
u
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
xp
erimen
ta
l res
ea
r
ch
o
n
text
C
A
P
TC
HA
o
f fin
e
-
g
r
a
in
ed
…
(
Qia
n
Wa
n
g
)
537
web
s
ites
(
Mic
r
o
s
o
f
t,
ed
u
ca
tio
n
Ma
lay
s
ia
g
lo
b
al
s
er
v
ices,
a
n
d
L
esh
an
L
ib
r
ar
y
)
o
r
Kag
g
le
(
a
n
o
tab
le
s
c
ien
ce
d
ata
co
m
m
u
n
ity
)
.
T
en
ty
p
ica
l
s
ec
u
r
ity
f
ea
tu
r
es
wer
e
ex
p
lo
r
ed
:
co
lo
r
s
,
d
o
ts
,
Gau
s
s
ian
n
o
is
e,
lin
es,
r
o
tat
e
,
f
o
n
ts
,
s
ca
le,
tr
an
s
f
o
r
m
e
r
,
b
ac
k
g
r
o
u
n
d
(
b
g
)
,
an
d
o
v
er
lap
.
Fig
u
r
e
1
.
T
h
e
f
r
a
m
ewo
r
k
o
f
f
i
n
e
-
g
r
ain
e
d
s
ec
u
r
ity
f
ea
t
u
r
es
e
v
alu
atio
n
2
.
2
.
B
a
s
e
CAP
T
CH
A
"
g
ener
a
t
io
n
Th
e
"
B
ase
C
APTC
HA
"
was
u
s
ed
to
test
th
e
an
ti
-
attac
k
a
b
ilit
ies
o
f
tex
t
C
APTCHA
w
ith
v
ar
io
u
s
s
ec
u
r
ity
f
ea
tu
r
es,
wh
ich
wer
e
ex
p
lo
r
ed
i
n
s
ec
tio
n
2
.
1
.
I
t
was
r
an
d
o
m
l
y
ass
em
b
led
b
u
t
m
eti
cu
lo
u
s
ly
d
esig
n
ed
with
s
p
ec
if
ic
p
ar
am
eter
s
,
in
c
l
u
d
in
g
C
alib
r
i
f
o
n
t,
a
wh
ite
b
ac
k
g
r
o
u
n
d
,
a
wid
th
o
f
1
9
2
p
ix
els,
a
h
eig
h
t
o
f
6
4
p
ix
els,
an
d
b
lu
e
ch
a
r
ac
ter
s
co
lo
r
.
Usi
n
g
"
B
ase
C
A
PTC
HA
"
en
s
u
r
e
s
th
e
in
itializa
tio
n
p
r
o
ce
s
s
is
o
n
th
e
s
am
e
b
aselin
e.
2
.
3
.
M
ulti
-
s
t
a
g
e
bes
t
pa
ra
met
er
s
s
elec
t
io
n
m
ec
ha
nis
m
I
n
f
lu
e
n
ce
d
b
y
th
e
f
ea
tu
r
e
s
elec
tio
n
ap
p
r
o
ac
h
in
[
3
2
]
,
we
co
m
b
in
ed
th
e
ch
a
r
ac
ter
is
tics
o
f
th
e
co
m
b
in
ed
C
APTCHA
s
ec
u
r
ity
f
ea
tu
r
es
ar
ea
a
n
d
p
r
o
p
o
s
ed
a
MBP
S
m
ec
h
an
is
m
.
T
h
is
m
ec
h
an
is
m
is
im
p
lem
en
ted
s
tep
b
y
s
tep
an
d
ca
n
b
e
tar
g
eted
an
d
ef
f
ec
tiv
e
wh
en
ev
al
u
atin
g
t
h
e
p
e
r
f
o
r
m
a
n
ce
o
f
ea
ch
s
ec
u
r
ity
f
ea
tu
r
e.
Nec
ess
ar
y
p
r
e
p
ar
atio
n
s
s
h
o
u
ld
b
e
m
a
d
e
b
ef
o
r
e
MBP
S m
ec
h
an
is
m
s
ar
e
im
p
lem
en
te
d
.
2
.
3
.
1
.
P
re
pa
ra
t
i
o
ns
o
f
t
he
M
B
P
S
m
ec
ha
nis
m
A
.
H
a
rdwa
re
re
qu
irem
ent
s
T
h
e
ex
p
er
im
en
t
u
s
es
s
p
ec
if
ic
h
ar
d
war
e
c
o
m
p
o
n
en
ts
to
en
s
u
r
e
o
p
tim
al
ex
p
e
r
im
en
tal
p
e
r
f
o
r
m
an
ce
,
as
s
h
o
wn
in
T
a
b
le
1
.
C
o
m
p
ar
e
d
to
ce
n
tr
al
p
r
o
ce
s
s
in
g
u
n
it
(
C
PU
)
,
th
e
h
ar
d
war
e
en
v
ir
o
n
m
en
t
o
f
GPU
ca
n
p
r
o
v
id
e
f
aster
p
r
o
ce
s
s
in
g
s
p
ee
d
,
r
ed
u
ce
p
r
o
g
r
am
ex
ec
u
ti
o
n
tim
e,
an
d
en
h
an
ce
o
v
er
a
ll
ef
f
icien
cy
.
T
h
e
alg
o
r
ith
m
u
s
ed
in
th
is
r
esear
ch
p
r
o
g
r
am
is
im
p
lem
en
ted
with
Py
th
o
n
,
VSC
o
d
e,
an
d
An
ac
o
n
d
a.
T
h
ese
in
teg
r
ated
d
ev
el
o
p
m
en
t e
n
v
ir
o
n
m
en
t
(
I
DE
)
to
o
ls
h
elp
s
tr
ea
m
li
n
e
s
o
f
twar
e
in
ter
ac
tio
n
s
an
d
d
ep
lo
y
m
e
n
t.
T
h
u
s
,
th
ese
h
ar
d
war
e
r
e
q
u
ir
em
e
n
ts
p
r
o
v
id
e
th
e
f
o
u
n
d
atio
n
f
o
r
th
e
s
m
o
o
th
e
x
ec
u
tio
n
o
f
th
e
e
x
p
er
i
m
en
t.
T
ab
le
1
.
Har
d
war
e
co
n
f
ig
u
r
ati
o
n
o
f
ex
p
er
i
m
en
ts
H
a
r
d
w
a
r
e
c
o
n
f
i
g
u
r
a
t
i
o
n
S
p
e
c
i
f
i
c
i
n
f
o
r
m
a
t
i
o
n
H
a
r
d
w
a
r
e
f
o
r
c
o
m
p
u
t
a
t
i
o
n
G
P
U
(
N
V
I
D
I
A
3
0
6
0
)
O
S
(
o
p
e
r
a
t
i
n
g
sy
s
t
e
m
)
W
i
n
d
o
w
s 1
1
C
P
U
I
n
t
e
l
(
R
)
C
o
r
e
(
T
M
)
i
5
R
A
M
(
r
a
n
d
o
m a
c
c
e
ss m
e
mo
r
y
)
3
2
7
6
8
M
B
P
r
o
g
r
a
m
mi
n
g
e
n
v
i
r
o
n
m
e
n
t
V
S
C
o
d
e
,
A
n
a
c
o
n
d
a
,
P
y
t
h
o
n
B
.
M
a
chine r
ec
o
g
nitio
n m
o
del
T
h
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
(
d
e
ep
-
C
APTCHA)
f
r
o
m
Z
ah
r
a
No
u
r
y
an
d
Ma
h
d
i Reza
ei
[
2
8
]
w
as u
s
ed
as
an
attac
k
m
eth
o
d
t
o
test
th
e
s
ec
u
r
ity
o
f
C
APTCHA.
T
h
e
attac
k
alg
o
r
ith
m
was
ch
o
s
en
b
e
ca
u
s
e
it
r
an
k
s
f
ir
s
t
am
o
n
g
te
x
t
-
b
ased
C
APTCH
A
r
ec
o
g
n
itio
n
alg
o
r
ith
m
s
o
n
th
e
web
s
ite
[
33
]
.
Mo
r
e
o
v
er
,
th
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
p
er
f
o
r
m
ed
well
an
d
was
ap
p
r
o
p
r
iately
u
s
ed
f
o
r
t
h
e
ex
p
er
im
e
n
tal
v
er
if
icatio
n
.
T
h
e
s
im
p
lifie
d
d
ee
p
-
C
APT
C
HA
a
r
ch
itectu
r
e
i
s
d
ep
icted
in
Fig
u
r
e
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
1
,
Ap
r
il
20
2
5
:
5
3
5
-
5
4
5
538
Fig
u
r
e
2
.
T
h
e
ar
ch
itectu
r
e
o
f
d
ee
p
-
C
APTCHA
2.
3
.
2.
B
est
s
ing
le
s
ec
urit
y
f
ea
t
ure
pa
ra
m
et
er
s
s
elec
t
io
n sta
g
e
R
ec
o
g
n
itio
n
alg
o
r
ith
m
s
wer
e
u
s
ed
to
ev
alu
ate
C
APTCH
As
u
s
in
g
o
n
ly
a
s
in
g
le
s
ec
u
r
ity
f
e
atu
r
e
.
T
h
e
p
ar
am
eter
s
elec
tio
n
cr
iter
ia
ar
e
b
ased
o
n
th
e
c
h
ar
ac
ter
is
tics
o
f
s
ec
u
r
ity
f
ea
tu
r
es.
A
p
ar
am
e
ter
was
ad
d
e
d
ea
c
h
tim
e
wh
en
g
en
er
atin
g
tex
t
C
APTCHA
d
ata
s
ets
f
r
o
m
“b
ase
C
APT
C
HA”
.
E
ac
h
n
ewly
g
en
er
ated
d
ataset
is
k
ep
t
th
e
s
am
e
s
ize
to
m
ain
t
ain
co
n
s
is
ten
cy
.
Me
an
wh
ile,
s
u
itab
le
n
u
m
b
er
s
will
b
e
co
v
er
ed
,
a
n
d
s
ec
u
r
ity
f
ea
tu
r
e
p
a
r
am
eter
s
will
g
r
a
d
u
ally
b
e
a
d
ju
s
ted
.
T
y
p
ical
s
ec
u
r
ity
f
ea
tu
r
es
an
d
d
etailed
r
e
aso
n
s
f
o
r
p
ar
am
eter
s
elec
tio
n
is
s
h
o
wn
in
T
ab
le
2
.
T
ab
le
2
.
T
e
n
ty
p
ical
s
ec
u
r
ity
f
ea
tu
r
es p
ar
am
eter
s
an
d
s
elec
tio
n
r
ea
s
o
n
s
N
o
.
S
e
c
u
r
i
t
y
f
e
a
t
u
r
e
s (sf
)
P
a
r
a
me
t
e
r
’
s
se
t
t
i
n
g
s
l
i
st
S
e
l
e
c
t
i
o
n
r
e
a
s
o
n
s
/
B
a
se
C
A
P
T
C
H
A
{
C
a
l
i
b
r
i
f
o
n
t
,
w
h
i
t
e
b
a
c
k
g
r
o
u
n
d
,
w
i
d
t
h
1
9
2
p
i
x
e
l
s,
h
e
i
g
h
t
6
4
p
i
x
e
l
s
,
b
l
u
e
c
h
a
r
a
c
t
e
r
s c
o
l
o
r
}
/
sf
_
1
C
o
l
o
r
s
{
b
l
u
e
,
g
r
a
y
b
l
u
e
;
g
r
a
y
-
r
e
d
,
g
r
a
y
-
r
e
d
-
b
l
u
e
,
g
r
a
y
-
y
e
l
l
o
w
-
b
l
u
e
,
l
i
g
h
t
g
r
a
y
-
y
e
l
l
o
w
-
b
l
u
e
,
l
i
g
h
t
-
y
e
l
l
o
w
-
b
l
u
e
-
g
r
e
e
n
,
l
i
g
h
t
-
y
e
l
l
o
w
-
b
l
u
e
-
g
r
e
e
n
-
g
r
a
y
}
1
.
v
a
r
i
o
u
s
c
o
l
o
r
s wh
i
c
h
c
a
n
r
e
f
l
e
c
t
t
h
e
i
mp
a
c
t
o
f
c
o
l
o
u
r
s
2
.
C
l
o
se
t
o
t
h
e
b
g
c
o
l
o
r
,
n
o
t
e
a
s
y
t
o
r
e
c
o
g
n
i
z
e
.
sf
_
2
B
a
c
k
g
r
o
u
n
d
(
b
g
)
{P
a
i
n
t
i
n
g
s
o
f
V
a
n
G
o
g
h
,
M
o
n
e
t
,
C
h
i
n
e
se
s
t
y
l
e
i
m
a
g
e
s}
1
.
D
i
v
e
r
se
b
g
s
t
y
l
e
s.
2
.
U
s
e
r
s c
a
n
a
l
so
r
e
c
o
g
n
i
z
e
.
sf
_
3
F
o
n
t
s
{
C
a
l
i
b
r
i
,
S
TX
I
N
W
E
I
,
F
ZSTK
,
M
e
x
c
e
l
l
e
n
t
3
D
,
M
TF
To
a
st
,
B
r
o
a
d
c
a
st
t
i
t
l
i
n
g
,
C
a
r
t
o
o
n
,
I
n
so
m
n
i
a
}
1
.
v
a
r
i
o
u
s
g
e
o
m
e
t
r
i
c
s
h
a
p
e
s
f
o
n
t
s
:
so
l
i
d
,
h
o
l
l
o
w
,
a
n
d
sh
a
d
o
w
.
2
.
d
i
f
f
e
r
e
n
t
t
y
p
e
s
o
f
f
o
n
t
s
:
r
e
g
u
l
a
r
,
a
r
t
a
n
d
c
a
r
t
o
o
n
sf
_
4
G
u
ssi
e
n
o
i
se
[
1
.
2
k
,
4
k
,
6
k
,
8
k
,
1
0
k
]
1
.
G
r
a
d
u
a
l
l
y
a
d
j
u
s
t
e
d
t
h
e
s
e
c
u
r
i
t
y
f
e
a
t
u
r
e
s
p
a
r
a
me
t
e
r
s.
2
.
U
s
e
r
s c
a
n
a
l
so
r
e
c
o
g
n
i
z
e
.
sf
_
5
D
o
t
s
[
2
0
,
4
0
,
8
0
,
1
6
0
,
4
0
0
,
6
0
0
]
sf
_
6
Li
n
e
s
[
1
0
,
3
0
,
7
0
,
1
0
0
]
sf
_
7
R
o
t
a
t
e
[
2
0
,
4
0
,
6
0
,
8
0
]
sf
_
8
S
c
a
l
e
[5
,
7
,
1
0
,
1
2
,
1
5
,
1
7
,
2
0
]
sf
_
9
Tr
a
n
sf
o
r
mer
[
h
0
1
,
h
0
2
,
h
0
3
,
v
0
1
,
v
0
1
-
h
0
1
,
v
0
2
h
0
1
]
sf
_
1
0
O
v
e
r
l
a
p
[
v
1
0
h
4
,
v
1
0
h
5
,
v
1
0
h
6
,
v
1
0
h
7
,
v
1
0
h
8
,
v
1
0
h
9
]
A.
B
est
s
ing
le
s
ec
urit
y
f
ea
t
ure
s
elec
t
io
n ste
ps
I
n
th
is
s
tag
e,
we
g
en
er
ated
m
an
y
C
APTCHA
d
atase
ts
ac
co
r
d
in
g
t
o
th
e
p
ar
a
m
eter
s
ettin
g
s
lis
t
in
T
ab
le
2
.
T
o
q
u
ick
ly
co
m
p
ar
e
th
e
im
p
ac
t
o
f
ea
ch
s
ec
u
r
ity
f
e
atu
r
e
'
s
p
ar
am
eter
s
,
th
e
g
en
er
ated
d
ataset
s
ize
was
s
et
to
2
,
0
0
0
.
T
h
is
s
ize
n
u
m
b
er
is
s
u
itab
le
f
o
r
th
e
i
n
itial
s
cr
ee
n
in
g
p
h
ase,
esp
ec
iall
y
w
h
en
co
m
p
ar
in
g
th
e
ef
f
ec
ts
o
f
d
if
f
er
en
t
p
ar
am
eter
s
u
n
d
er
th
e
s
am
e
s
ec
u
r
ity
f
ea
tu
r
e.
Step
s
1
to
4
illu
s
tr
ate
h
o
w
th
e
b
est
s
in
g
le
s
ec
u
r
ity
f
ea
tu
r
e
is
s
elec
ted
.
Step
1
)
g
en
e
r
ated
d
atasets
with
s
in
g
le
s
ec
u
r
ity
f
ea
tu
r
e
p
ar
a
m
eter
s
.
Step
2
)
t
ested
C
APTCHA
an
ti
-
attac
k
ab
ilit
y
u
s
in
g
th
e
d
ee
p
-
C
APT
C
HA
r
ec
o
g
n
itio
n
m
o
d
el
.
Step
3
)
s
cr
ee
n
ed
t
h
e
b
est
-
p
er
f
o
r
m
in
g
p
ar
am
eter
s
u
n
d
er
t
h
e
s
am
e
s
ec
u
r
ity
f
ea
tu
r
e
.
Step
4
)
c
o
m
p
ar
e
th
e
p
er
f
o
r
m
a
n
ce
o
f
ten
s
ec
u
r
ity
f
ea
tu
r
es a
n
d
s
elec
t th
e
to
p
o
n
es.
B.
Resul
t
s
I
n
th
is
p
ar
t,
av
er
a
g
e
ac
cu
r
ac
y
s
u
cc
ess
r
ate
(
AASR
)
is
u
s
ed
to
e
v
alu
ate
th
e
a
v
er
ag
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
,
wh
ic
h
ca
n
in
d
ir
ec
t
ly
r
ef
lect
C
APTCHA’
s
r
esis
t
an
ce
to
m
ac
h
in
e
r
ec
o
g
n
itio
n
.
T
h
e
v
er
tical
ax
is
r
ep
r
esen
ts
th
e
m
ea
n
test
ac
cu
r
ac
y
r
ate,
w
h
ile
th
e
h
o
r
izo
n
tal
ax
is
r
ep
r
es
en
ts
th
e
n
u
m
b
e
r
o
f
ep
o
ch
s
.
T
h
e
r
esu
lts
o
f
th
e
s
in
g
le
s
ec
u
r
ity
f
ea
t
u
r
e
p
ar
am
eter
s
elec
tio
n
ar
e
s
h
o
wn
in
Fig
u
r
e
3
.
T
o
clar
if
y
th
e
r
esu
lts
,
we
co
n
tr
ast
all
th
e
s
ec
u
r
ity
f
ea
tu
r
e
cu
r
v
es f
r
o
m
Fig
u
r
e
s
3
(
a)
t
o
3
(
h
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
xp
erimen
ta
l res
ea
r
ch
o
n
text
C
A
P
TC
HA
o
f fin
e
-
g
r
a
in
ed
…
(
Qia
n
Wa
n
g
)
539
Fig
u
r
e
3
(
a)
illu
s
tr
ates
th
e
p
er
f
o
r
m
an
ce
o
f
c
o
lo
r
s
ec
u
r
ity
f
ea
t
u
r
e
p
ar
am
eter
s
.
T
h
e
b
est
p
ar
a
m
eter
s
ar
e
lig
h
t
-
y
ello
w
-
b
lu
e
-
g
r
ee
n
-
g
r
e
y
in
th
e
wh
o
le,
an
d
af
ter
f
iv
e
e
p
o
ch
s
,
th
e
r
ec
o
g
n
itio
n
r
ate
r
e
ac
h
ed
1
0
0
%.
T
h
is
m
ea
n
s
th
at
th
is
s
in
g
le
s
ec
u
r
ity
f
ea
tu
r
e
was
ea
s
y
to
b
r
ea
k
.
Fig
u
r
e
s
3
(
b
)
,
3
(
c)
,
an
d
3
(
f
)
r
ep
r
es
en
t
d
o
ts
,
lin
es,
an
d
g
u
s
s
ie
n
o
is
e
s
ec
u
r
ity
f
ea
tu
r
es
wh
o
s
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
r
e
ac
h
es
1
0
0
%
with
in
te
n
ep
o
c
h
s
.
T
h
e
f
o
u
r
s
ec
u
r
ity
f
ea
tu
r
es we
r
e
also
wea
k
in
ter
m
s
o
f
an
ti
-
attac
k
ab
ilit
y
.
Similar
ly
,
th
e
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
o
f
Fig
u
r
e
s
3
(
d
)
,
3
(
c)
,
a
n
d
3
(
g
)
m
ea
n
s
th
at
tr
an
s
f
o
r
m
er
s
,
f
o
n
ts
,
an
d
r
o
tate
s
ec
u
r
ity
f
ea
tu
r
es
r
e
ac
h
1
0
0
%
with
in
2
0
ep
o
ch
s
.
T
h
e
th
r
ee
s
ec
u
r
ity
f
ea
tu
r
es
ar
e
in
th
e
m
id
d
le
o
f
an
ti
-
attac
k
ab
ilit
y
p
er
f
o
r
m
a
n
c
e.
Scale
Fig
u
r
e
3
(
i)
,
o
v
er
lap
Fig
u
r
e
3
(
h
)
,
an
d
b
ac
k
g
r
o
u
n
d
Fi
g
u
r
e
3
(
j)
p
e
r
f
o
r
m
ed
well;
t
h
ei
r
b
est p
ar
am
eter
p
r
ec
is
io
n
d
id
n
o
t
r
ea
ch
1
0
0
% with
in
5
0
e
p
o
ch
s
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
(
g
)
(
h
)
(
i)
(
j)
Fig
u
r
e
3.
Per
f
o
r
m
an
c
e
o
f
s
in
g
l
e
s
ec
u
r
ity
f
ea
tu
r
e
p
ar
am
eter
s
:
(
a)
co
lo
r
s
,
(
b
)
d
o
ts
,
(
c)
lin
es,
(
d
)
tr
an
s
f
o
r
m
er
,
(
e)
r
o
tate,
(
f
)
g
u
s
s
ie
n
o
is
e,
(
g
)
f
o
n
ts
,
(
h
)
o
v
e
r
lap
,
(
h
)
o
v
er
lap
,
(
i)
s
ca
le,
an
d
(
j)
b
ac
k
g
r
o
u
n
d
(
b
g
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
1
,
Ap
r
il
20
2
5
:
5
3
5
-
5
4
5
540
Fig
u
r
e
4
r
ev
ea
ls
t
h
e
b
est
p
er
f
o
r
m
in
g
c
u
r
v
es
o
f
s
in
g
le
s
ec
u
r
it
y
f
ea
tu
r
e
,
p
r
o
v
id
in
g
clea
r
c
o
m
p
ar
is
o
n
o
f
d
if
f
e
r
en
t
p
ar
am
eter
s
.
T
ab
le
3
illu
s
tr
ates
th
e
m
ac
h
in
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
w
h
en
th
e
n
u
m
b
er
o
f
e
p
o
ch
s
is
s
et
to
5
0
,
s
h
o
win
g
h
o
w
th
e
ac
cu
r
ac
y
ch
an
g
es
as
th
e
s
ec
u
r
ity
f
ea
tu
r
es
ch
an
g
es
.
Si
n
c
e
t
h
e
s
ix
‘
w
ea
k
’
s
e
cu
r
i
ty
f
e
at
u
r
es
(
co
lo
r
s
a
n
d
t
r
a
n
s
f
o
r
m
er
)
we
r
e
q
u
i
ck
ly
r
e
c
o
g
n
i
ze
d
a
n
d
d
i
d
n
o
t
s
i
g
n
i
f
ic
an
tl
y
c
o
n
t
r
i
b
u
te
t
o
t
h
e
o
v
e
r
all
s
ec
u
r
it
y
,
t
h
ey
wer
e
ex
clu
d
ed
f
r
o
m
f
u
r
th
er
ev
al
u
atio
n
.
T
h
is
ex
clu
s
io
n
allo
wed
u
s
to
f
o
cu
s
o
n
m
o
r
e
ef
f
ec
tiv
e
s
ec
u
r
ity
f
ea
tu
r
es.
T
h
e
f
o
u
r
s
t
r
o
n
g
es
t
o
n
es
(
bg
,
o
v
e
r
la
p
,
s
c
ale
,
an
d
r
o
tat
e)
w
er
e
s
cr
ee
n
e
d
to
g
e
n
e
r
a
te
m
ix
e
d
s
ec
u
r
it
y
f
ea
tu
r
es
C
APTCH
A
i
n
t
h
e
n
e
x
t st
ag
e
.
T
h
is
a
p
p
r
o
a
c
h
p
r
io
r
i
tiz
es t
h
e
m
o
s
t
c
h
al
le
n
g
i
n
g
s
e
cu
r
i
ty
f
ea
t
u
r
es
,
wh
i
ch
c
a
n
s
i
m
p
li
f
y
th
e
ev
al
u
a
t
io
n
p
r
o
ce
s
s
.
T
a
b
le
3
.
T
h
e
r
e
co
g
n
it
io
n
a
cc
u
r
ac
y
o
f
t
h
e
b
es
t si
n
g
l
e
s
ec
u
r
it
y
m
ec
h
an
is
m
s
(
e
p
o
c
h
s
=
5
0
)
N
o
.
S
e
c
u
r
i
t
y
f
e
a
t
u
r
e
s
B
e
st
si
n
g
l
e
se
c
u
r
i
t
y
f
e
a
t
u
r
e
s
p
a
r
a
me
t
e
r
s
M
a
c
h
i
n
e
r
e
c
o
g
n
i
t
i
o
n
a
c
c
u
r
a
c
y
(
%)
1
O
v
e
r
l
a
p
v
1
0
h
8
4
5
.
3
2
B
a
c
k
g
r
o
u
n
d
C
h
i
n
e
se
-
s
t
y
l
e
0
7
7
9
.
7
3
S
c
a
l
e
2
0
8
3
.
2
4
R
o
t
a
t
e
8
0
9
8
.
7
5
Li
n
e
s
1
0
0
9
9
.
3
6
D
o
t
s
8
0
0
9
9
.
8
7
G
u
ssi
e
n
o
i
se
1
.
2
k
9
9
.
9
8
F
o
n
t
s
M
TF
To
a
st
9
9
.
9
9
C
o
l
o
u
r
s
l
i
g
h
t
-
y
e
l
l
o
w
-
b
l
u
e
-
g
r
e
e
n
-
g
r
e
y
1
0
0
10
Tr
a
n
sf
o
r
mer
h
0
2
1
0
0
Fig
u
r
e
4.
Acc
u
r
ac
y
o
f
th
e
b
est s
in
g
le
s
ec
u
r
ity
f
ea
tu
r
es p
a
r
am
eter
co
m
p
ar
is
o
n
cu
r
v
es
2
.
3.
3
.
B
est
m
ix
ed
s
ec
urit
y
f
e
a
t
ures
pa
ra
m
et
er
s
s
elec
t
io
n
s
t
a
g
e
A.
M
ix
ed
s
ec
urit
y
f
ea
t
ures C
AP
T
CH
A
T
h
is
s
ec
t
io
n
e
m
p
l
o
y
e
d
th
e
t
o
p
f
o
u
r
r
o
b
u
s
t
s
ec
u
r
it
y
f
ea
tu
r
es
(
b
g
,
o
v
e
r
la
p
,
s
ca
le
,
a
n
d
r
o
ta
te
)
to
g
e
n
e
r
a
te
m
i
x
e
d
s
e
c
u
r
i
ty
f
ea
t
u
r
es
.
H
er
e,
we
c
o
v
e
r
all
th
e
c
o
m
b
i
n
at
io
n
s
o
f
t
h
es
e
f
o
u
r
s
t
r
o
n
g
o
n
es
.
M
ix
tw
o
,
t
h
r
ee
,
a
n
d
f
o
u
r
d
i
f
f
e
r
e
n
t
co
m
b
i
n
at
io
n
s
o
f
s
e
c
u
r
i
ty
f
ea
t
u
r
es
s
im
u
l
ta
n
e
o
u
s
l
y
.
Si
m
ila
r
l
y
,
t
h
e
d
ee
p
-
C
APTCHA
r
ec
o
g
n
itio
n
alg
o
r
ith
m
also
m
ea
s
u
r
e
d
th
e
a
n
ti
-
r
ec
o
g
n
itio
n
ab
ilit
y
o
f
m
ix
e
d
s
ec
u
r
ity
f
ea
tu
r
es
.
B.
Resul
t
s
a
nd
a
na
ly
s
is
T
h
e
co
n
tr
ast
cu
r
v
es
ar
e
s
h
o
wn
in
Fig
u
r
e
5
.
Me
a
n
wh
ile,
th
e
r
e
co
g
n
itio
n
ac
c
u
r
ac
y
o
f
th
e
m
i
x
ed
s
ec
u
r
ity
f
ea
tu
r
es
wh
en
th
e
e
p
o
ch
s
ar
e
eq
u
al
to
1
0
0
is
d
is
p
lay
ed
in
T
ab
le
4
.
T
h
e
m
o
s
t
ef
f
ec
tiv
e
an
ti
-
r
ec
o
g
n
itio
n
ca
p
ab
ilit
y
,
am
o
n
g
all
p
er
m
u
tatio
n
s
o
f
m
ix
e
d
s
ec
u
r
ity
m
ec
h
an
is
m
s
,
is
ac
h
iev
ed
b
y
co
m
b
in
in
g
o
v
er
lap
,
b
ac
k
g
r
o
u
n
d
(
b
g
)
,
s
ca
le
,
an
d
r
o
tate
,
b
o
asti
n
g
an
e
f
f
i
ca
cy
r
ate
o
f
4
.
8
1
%.
T
h
e
o
p
ti
m
al
an
ti
-
r
ec
o
g
n
itio
n
ca
p
ab
ilit
y
am
o
n
g
th
r
ee
co
m
b
i
n
atio
n
s
o
f
m
ix
ed
s
ec
u
r
ity
m
ec
h
an
is
m
s
,
in
co
r
p
o
r
atin
g
o
v
e
r
lap
,
s
ca
l
e
,
an
d
r
o
tat
e,
is
at
an
im
p
r
ess
iv
e
7
.
6
9
%
ef
f
icac
y
r
ate.
T
h
e
s
u
p
er
io
r
a
n
ti
-
r
ec
o
g
n
itio
n
ca
p
ab
ilit
y
o
f
c
o
m
b
in
in
g
two
m
i
x
ed
s
ec
u
r
ity
m
ec
h
an
is
m
s
,
s
p
ec
if
ic
ally
o
v
er
lap
a
n
d
s
ca
lin
g
,
y
ield
s
an
im
p
r
ess
iv
e
ef
f
icac
y
r
ate
o
f
1
2
.
5
6
%.
Fro
m
th
e
s
tatis
tical
d
ata,
we
f
in
d
th
e
f
o
llo
win
g
:
−
C
o
m
p
ar
ed
t
o
C
APTCHA
wit
h
a
s
in
g
le
s
ec
u
r
ity
f
ea
tu
r
e
,
m
ix
ed
o
n
es
d
e
m
o
n
s
tr
ate
s
u
p
er
i
o
r
r
esis
tan
ce
to
m
ac
h
in
e
attac
k
s
.
Ov
er
la
p
p
in
g
ch
ar
ac
ter
s
an
d
co
m
p
lex
b
ac
k
g
r
o
u
n
d
s
wer
e
two
e
f
f
ec
tiv
e
s
ec
u
r
ity
f
ea
tu
r
es
ag
ain
s
t m
ac
h
in
e
r
ec
o
g
n
itio
n
.
−
A
d
d
in
g
c
o
m
p
lex
b
ac
k
g
r
o
u
n
d
i
n
ter
f
e
r
en
ce
ca
n
im
p
r
o
v
e
th
e
an
ti
-
attac
k
ab
ilit
y
.
C
o
n
s
id
er
in
g
th
e
p
er
f
o
r
m
an
ce
o
f
eig
h
t
ty
p
ical
b
ac
k
g
r
o
u
n
d
s
,
th
o
s
e
b
g
im
ag
es
h
ad
h
ig
h
er
s
atu
r
atio
n
an
d
we
r
e
m
o
r
e
s
ec
u
r
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
xp
erimen
ta
l res
ea
r
ch
o
n
text
C
A
P
TC
HA
o
f fin
e
-
g
r
a
in
ed
…
(
Qia
n
Wa
n
g
)
541
T
h
is
is
b
ec
au
s
e
th
e
h
ig
h
er
t
h
e
co
n
tr
ast
b
etwe
en
th
e
b
ac
k
g
r
o
u
n
d
an
d
th
e
c
h
ar
ac
ter
s
,
t
h
e
m
o
re
d
i
f
f
icu
lt
it
is
f
o
r
th
e
m
ac
h
in
e
a
n
d
th
e
h
u
m
a
n
to
r
ec
o
g
n
ize
co
r
r
ec
tly
.
−
A
b
alan
ce
b
etwe
en
an
ti
-
attac
k
ab
ilit
y
an
d
u
s
er
-
f
r
ien
d
lin
es
s
s
h
o
u
ld
b
e
c
o
n
s
id
er
ed
ca
r
e
f
u
lly
.
Fig
u
r
e
6
illu
s
tr
ates
s
am
p
les
o
f
th
e
b
est
-
m
ix
ed
s
ec
u
r
ity
f
ea
tu
r
es
o
f
C
APTCHA.
Alth
o
u
g
h
Fig
u
r
e
6
(
a
)
co
m
b
in
atio
n
s
o
f
“o
v
e
r
lap
+b
g
+scale
+r
o
tate
”
h
av
e
th
e
b
est
p
e
r
f
o
r
m
an
ce
,
Fig
u
r
e
6
(
b
)
co
m
b
in
atio
n
s
o
f
“s
ca
le
+
o
v
er
lap
+r
o
tate”
m
ay
b
e
th
e
m
o
s
t
ap
p
r
o
p
r
iate
o
n
es
in
ap
p
licatio
n
s
b
ec
au
s
e
co
m
p
lex
b
ac
k
g
r
o
u
n
d
s
in
f
lu
en
ce
b
o
t
h
h
u
m
an
an
d
m
alicio
u
s
bo
ts
.
T
h
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
r
ate
i
s
n
o
t
th
e
o
n
ly
co
n
s
id
er
atio
n
;
u
s
er
-
f
r
ien
d
lin
ess
is
an
o
th
er
co
n
s
id
er
atio
n
,
a
n
d
t
h
e
b
alan
ce
b
etwe
en
th
em
s
h
o
u
ld
b
e
co
n
s
id
er
ed
ca
r
e
f
u
lly
.
−
No
tab
ly
,
th
e
r
ec
o
g
n
itio
n
m
o
d
el
d
id
n
o
t
co
n
v
er
g
e
with
in
1
0
0
ep
o
ch
s
.
I
n
s
u
f
f
icien
t
s
am
p
le
s
ize
an
d
in
ad
eq
u
ate
tr
ain
in
g
ep
o
c
h
s
w
er
e
r
ea
s
o
n
s
f
o
r
th
e
m
o
d
el'
s
f
ailu
r
e
to
c
o
n
v
e
r
g
e.
I
n
th
eo
r
y
,
in
cr
ea
s
in
g
t
h
e
s
am
p
le
s
ize
o
r
th
e
n
u
m
b
er
o
f
ep
o
ch
s
co
u
ld
im
p
r
o
v
e
th
e
m
o
d
el'
s
co
n
v
er
g
en
ce
s
p
ee
d
,
w
h
ich
s
h
o
u
l
d
b
e
th
e
f
o
llo
win
g
wo
r
k
.
Fig
u
r
e
5
.
C
o
m
p
a
r
is
o
n
cu
r
v
es
o
f
m
ix
ed
s
ec
u
r
ity
m
ec
h
a
n
is
m
s
C
APTC
HA
T
a
b
le
4
.
T
h
e
r
e
co
g
n
it
io
n
a
cc
u
r
ac
y
o
f
m
i
x
ed
s
ec
u
r
it
y
f
ea
tu
r
e
s
(
e
p
o
c
h
s
=
1
0
0
)
M
i
x
e
d
s
e
c
u
r
i
t
y
f
e
a
t
u
r
e
s
R
e
c
o
g
n
i
t
i
o
n
a
c
c
u
r
a
c
y
(
%)
M
i
x
e
d
s
e
c
u
r
i
t
y
f
e
a
t
u
r
e
s
R
e
c
o
g
n
i
t
i
o
n
a
c
c
u
r
a
c
y
(
%)
O
v
e
r
l
a
p
+
b
g
+
s
c
a
l
e
+
r
o
t
a
t
e
4
.
8
1
B
g
+
r
o
t
a
t
e
+
s
c
a
l
e
2
1
.
9
4
O
v
e
r
l
a
p
+
sc
a
l
e
+
r
o
t
a
t
e
7
.
6
9
S
c
a
l
e
+
r
o
t
a
t
e
3
3
.
5
0
O
v
e
r
l
a
p
+
sc
a
l
e
+
b
g
8
.
6
2
O
v
e
r
l
a
p
+
b
g
4
2
.
9
4
O
v
e
r
l
a
p
+
b
g
+
r
o
t
a
t
e
9
.
6
9
B
g
+
r
o
t
a
t
e
4
4
.
6
9
O
v
e
r
l
a
p
+
sc
a
l
e
1
2
.
5
6
S
c
a
l
e
+
b
g
5
3
.
8
1
O
v
e
r
l
a
p
+
r
o
t
a
t
e
2
1
.
3
1
(
a)
(
b)
Fig
u
r
e
6.
Sam
p
les o
f
th
e
b
est
-
m
ix
ed
s
ec
u
r
ity
f
ea
t
u
r
es o
f
C
A
PTCHA
(
a)
o
v
er
lap
+b
g
+scale
+r
o
tate
o
v
er
lap
+scale
+r
o
tate
an
d
(
b
)
o
v
er
lap
+scale
+r
o
tate
C.
Ro
bu
s
t
CAP
T
CH
A
da
t
a
s
et
s
Usi
n
g
th
e
s
ec
u
r
ity
f
ea
tu
r
e
c
o
m
b
in
atio
n
s
m
o
tio
n
e
d
in
s
ec
tio
n
2
.
3
.
3
.
,
th
e
n
ewly
g
en
er
a
ted
s
ec
u
r
e
C
APT
C
HA
p
er
f
o
r
m
s
well
r
e
g
ar
d
in
g
an
ti
-
attac
k
ab
ilit
y
an
d
u
s
er
-
f
r
ien
d
lin
ess
.
T
h
e
in
te
g
r
atio
n
o
f
m
u
ltip
le
s
ec
u
r
ity
f
ea
tu
r
es
en
h
an
ce
s
its
r
esis
tan
ce
to
r
ec
o
g
n
itio
n
attac
k
s
,
m
ak
in
g
it
m
o
r
e
s
ec
u
r
e
th
a
n
C
APTCHA
s
with
s
im
p
le
s
ec
u
r
ity
f
ea
tu
r
es.
I
n
ad
d
itio
n
to
s
ec
u
r
ity
c
o
n
s
id
er
atio
n
s
,
h
u
m
an
u
s
er
s
ca
n
ea
s
ily
r
ec
o
g
n
ize
th
e
C
APT
C
HA
with
o
u
t
d
if
f
icu
lty
.
Valid
atio
n
test
s
will
b
e
co
n
d
u
cted
to
co
n
f
ir
m
th
e
s
ec
u
r
ity
p
er
f
o
r
m
an
ce
o
f
th
e
C
APT
C
HA
d
ataset
s
with
m
ix
ed
s
ec
u
r
ity
f
ea
tu
r
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
1
,
Ap
r
il
20
2
5
:
5
3
5
-
5
4
5
542
2
.
4
.
Co
ntr
a
s
t
e
x
perim
ent
a
nd
re
s
ult
A
co
m
p
a
r
ativ
e
e
x
p
er
im
e
n
t
wa
s
im
p
lem
en
ted
to
co
n
tr
ast
t
h
e
an
ti
-
attac
k
ab
ilit
y
o
f
C
APTC
HA
u
s
in
g
d
if
f
er
en
t
n
o
is
e
-
ad
d
in
g
m
eth
o
d
s
.
Ad
v
er
s
ar
ial
ex
am
p
les
an
d
s
ec
u
r
ity
f
ea
tu
r
es
wer
e
b
o
t
h
ad
d
ed
n
o
is
e
to
C
APT
C
HA.
T
h
e
d
if
f
er
en
ce
is
wh
er
e
th
e
n
o
is
e
was
ap
p
lie
d
.
Ad
v
er
s
ar
ial
ex
am
p
les
in
tr
o
d
u
ce
n
o
is
e
to
d
ee
p
lear
n
in
g
n
etwo
r
k
s
,
wh
ile
s
ec
u
r
ity
f
ea
tu
r
es in
tr
o
d
u
ce
n
o
is
e
to
th
e
C
APTCH
A
g
en
er
atio
n
p
r
o
ce
s
s
.
W
e
f
ir
s
t
g
en
er
ated
b
aselin
e
C
APTCHA
u
s
in
g
Py
th
o
n
'
s
ca
p
tch
a
lib
r
ar
y
(
d
ef
au
lt
f
o
n
ts
,
v
a
r
y
in
g
co
lo
r
s
,
dot
s
,
lin
es,
an
d
s
o
lid
b
ac
k
g
r
o
u
n
d
s
)
.
Fo
llo
win
g
th
e
m
eth
o
d
in
[
3
1
]
,
we
g
en
er
ate
d
th
e
AE
C
APTCH
A
u
s
in
g
g
au
s
s
ian
s
m
o
o
th
in
g
a
n
d
th
e
f
a
s
t
g
r
ad
ien
t
s
ig
n
m
eth
o
d
(
FGSM)
.
Nex
t,
we
g
en
er
ate
d
o
u
r
d
ataset
b
y
co
m
b
in
in
g
two
s
ec
u
r
ity
f
ea
tu
r
es
(
b
g
+scale
)
.
T
h
e
ex
p
e
r
im
en
tal
p
a
r
am
et
er
s
s
ettin
g
s
ar
e
s
h
o
wn
i
n
T
ab
l
e
5
.
I
n
th
is
p
ar
t,
we
ca
r
ef
u
lly
b
alan
ce
d
u
s
er
-
f
r
ien
d
lin
ess
an
d
attac
k
r
esi
s
tan
ce
,
s
e
ttin
g
th
e
b
ac
k
g
r
o
u
n
d
im
a
g
e
tr
an
s
p
ar
en
cy
to
0
.
6
5
,
a
v
alu
e
o
b
tai
n
ed
f
r
o
m
th
e
ex
p
er
im
en
t,
to
en
s
u
r
e
p
r
ac
tical
an
d
ap
p
licab
le
r
esu
lts
.
Sam
p
le
s
izes
(
2
.
0
0
0
;
10
,
0
0
0
)
wer
e
u
s
ed
to
ev
al
u
ate
th
e
s
ec
u
r
ity
im
p
ac
t.
T
a
b
le
5
.
E
x
p
e
r
i
m
e
n
t
al
C
APTC
HA
p
a
r
a
m
e
te
r
s
s
etti
n
g
s
C
A
P
TC
H
A
t
y
p
e
P
a
r
a
me
t
e
r
s/
o
p
t
i
o
n
s
S
e
t
t
i
n
g
s
B
a
se
l
i
n
e
C
A
P
T
C
H
A
C
h
a
r
a
c
t
e
r
En
g
l
i
sh
u
p
p
e
r
c
a
se
l
e
t
t
e
r
s
N
u
mb
e
r
o
f
l
e
t
t
e
r
s
4
W
i
d
t
h
,
h
e
i
g
h
t
1
9
2
,
6
4
F
o
n
t
,
f
o
n
t
s
i
z
e
,
d
o
t
s
,
l
i
n
e
s,
b
g
c
o
l
o
r
D
e
f
a
u
l
t
A
E
C
A
P
T
C
H
A
[
3
1
]
S
p
a
t
i
a
l
sm
o
o
t
h
i
n
g
σ
=
1
.
2
5
F
G
S
M
ε=0
.
2
O
u
r
C
A
P
T
C
H
A
(
b
g
+
s
c
a
l
e
)
C
A
P
TC
H
A
t
r
a
n
sp
a
r
e
n
c
y
0
.
6
5
B
a
c
k
g
r
o
u
n
d
C
h
i
n
e
se
-
s
t
y
l
e
0
7
O
f
f
set
R
a
n
d
o
m (
5
,
1
0
)
S
c
a
l
e
R
a
n
d
o
m
(
2
0
,
6
0
)
T
h
e
an
ti
-
attac
k
ab
ilit
y
p
er
f
o
r
m
an
ce
was
ev
alu
ated
b
y
co
n
tr
ast
ex
p
er
im
en
ts
,
an
d
th
e
r
esu
lts
ar
e
s
h
o
wn
in
Fig
u
r
e
7
.
W
e
f
o
u
n
d
th
at
th
e
AASR
f
o
r
th
e
b
aselin
e
C
A
PTCHA
in
cr
ea
s
ed
s
ig
n
if
ican
tly
f
r
o
m
2
8
.
6
3
%
with
2
,
0
0
0
s
am
p
les
to
7
2
.
5
6
%
with
1
0
,
0
0
0
s
am
p
le
s
,
s
h
o
win
g
h
ig
h
er
v
u
ln
er
ab
ilit
y
with
lar
g
er
d
atasets
.
AE
C
APT
C
HA
[
3
1
]
h
ad
a
s
im
ilar
tr
en
d
,
with
a
s
lig
h
tly
h
ig
h
er
AASR
o
f
0
.
2
9
1
3
f
o
r
2
,
0
0
0
s
am
p
les
an
d
a
r
e
d
u
ctio
n
t
o
0
.
6
4
6
4
with
1
0
,
0
0
0
s
am
p
les,
in
d
icat
in
g
b
etter
r
esis
tan
ce
th
an
t
h
e
b
aselin
e
b
u
t
s
till
v
u
ln
e
r
ab
le.
Ou
r
C
APTCHA
d
em
o
n
s
tr
ates
th
e
b
est,
with
th
e
lo
west
AASR
o
f
0
.
2
5
6
9
f
o
r
2
,
0
0
0
s
am
p
les
an
d
0
.
6
2
4
f
o
r
1
0
,
0
0
0
s
am
p
les,
m
ain
tain
in
g
s
u
p
e
r
io
r
r
esis
tan
ce
to
attac
k
s
.
Fig
u
r
e
7
.
T
h
e
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
o
f
th
e
c
o
n
tr
ast ex
p
er
im
e
n
t
B
ased
o
n
th
e
r
esu
lts
o
f
Fig
u
r
e
7
,
we
d
is
co
v
er
e
d
th
at:
−
B
aselin
e
C
A
PTC
HA
b
ec
o
m
es
h
ig
h
ly
v
u
ln
e
r
ab
le
as
th
e
n
u
m
b
er
o
f
s
am
p
les
in
cr
ea
s
es.
AE
C
APT
C
HA
[
3
1
]
s
h
o
ws b
etter
r
esis
tan
ce
th
an
th
e
b
aselin
e
with
1
0
,
0
0
0
s
am
p
le
s
b
u
t is st
ill r
elativ
ely
v
u
ln
er
a
b
le.
−
Ou
r
C
APTCH
A
co
n
s
is
ten
tly
d
em
o
n
s
tr
ates
th
e
b
est
p
er
f
o
r
m
an
ce
,
wi
th
th
e
lo
west
attac
k
s
u
cc
ess
r
ate
ac
r
o
s
s
b
o
th
s
am
p
le
s
izes,
m
ak
in
g
it
th
e
m
o
s
t
s
ec
u
r
e
o
p
tio
n
i
n
th
is
co
m
p
a
r
is
o
n
.
T
h
ese
r
esu
lts
s
u
g
g
est
th
at
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
xp
erimen
ta
l res
ea
r
ch
o
n
text
C
A
P
TC
HA
o
f fin
e
-
g
r
a
in
ed
…
(
Qia
n
Wa
n
g
)
543
o
u
r
C
APTCHA
d
esig
n
ef
f
ec
tiv
ely
b
alan
ce
s
u
s
ab
ilit
y
an
d
s
ec
u
r
ity
b
y
r
ed
u
ci
n
g
ad
v
er
s
ar
ial
attac
k
s
u
cc
ess
r
ates.
−
Su
f
f
ici
en
t
s
am
p
les
ca
n
en
a
b
l
e
th
e
r
ec
o
g
n
itio
n
m
o
d
el
to
o
b
tain
p
atter
n
in
f
o
r
m
atio
n
a
n
d
ac
h
iev
e
h
i
g
h
er
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
tu
d
y
in
v
esti
g
ated
th
e
ef
f
ec
ts
o
f
f
in
e
-
g
r
ai
n
ed
s
ec
u
r
ity
f
ea
tu
r
es
o
n
th
e
an
ti
-
attac
k
a
b
ilit
y
o
f
tex
t
-
b
ased
C
APTCHA
s
.
W
h
ile
ea
r
lier
s
tu
d
ies,
s
u
ch
as
W
an
g
et
a
l.
[
2
9
]
,
h
a
v
e
e
x
p
lo
r
e
d
t
h
e
v
u
ln
er
ab
ilit
ies
o
f
b
o
th
C
h
in
ese
an
d
E
n
g
lis
h
tex
t
C
A
PTCHA,
th
ey
h
av
e
n
o
t
ex
p
lic
itly
ad
d
r
ess
ed
th
e
in
f
lu
en
ce
o
f
ad
v
er
s
ar
ial
n
o
is
e
an
d
ad
v
a
n
ce
d
s
ec
u
r
ity
f
ea
tu
r
e
s
o
n
th
e
an
ti
-
attac
k
ca
p
ab
ilit
y
o
f
E
n
g
lis
h
-
lan
g
u
a
g
e
C
APTC
HA.
W
e
f
o
u
n
d
th
at
tex
t
C
APTCH
A
with
m
in
im
al
s
ec
u
r
ity
f
ea
tu
r
es
d
em
o
n
s
tr
ate
d
lo
w
r
esis
tan
ce
to
attac
k
s
,
co
n
s
is
te
n
t
with
r
esu
lt
o
f
[
2
9
]
.
Ho
wev
e
r
,
o
u
r
r
esu
l
ts
also
s
h
o
w
th
at
d
ee
p
lear
n
in
g
-
b
ased
attac
k
alg
o
r
ith
m
s
ca
n
ac
h
iev
e
h
ig
h
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
ev
en
a
f
ter
ad
d
in
g
s
ec
u
r
ity
f
ea
tu
r
es
s
u
ch
as
co
m
p
lex
b
ac
k
g
r
o
u
n
d
s
an
d
r
o
tatio
n
.
Ad
d
itio
n
ally
,
i
n
cr
ea
s
in
g
t
h
e
s
am
p
le
s
ize
s
ig
n
if
ican
tly
im
p
r
o
v
ed
m
ac
h
in
e
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
b
y
m
o
r
e
th
an
3
5
%,
as illu
s
tr
ated
in
Fig
u
r
e
7
.
Ou
r
s
tu
d
y
s
u
g
g
ests
th
at
ad
d
in
g
ad
v
e
r
s
ar
ial
n
o
is
e
an
d
s
ec
u
r
i
ty
f
ea
tu
r
es
r
ed
u
ce
d
m
ac
h
i
n
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
b
y
2
%,
co
m
p
a
r
ed
to
th
e
f
in
d
in
g
s
in
[
3
1
]
.
Ho
wev
e
r
,
t
h
ese
m
ea
s
u
r
es
alo
n
e
ar
e
in
s
u
f
f
icien
t
to
p
r
e
v
en
t
attac
k
s
en
tire
ly
.
Un
lik
e
W
an
g
et
a
l.
r
esear
ch
[
2
9
]
,
wh
ich
in
v
esti
g
ated
b
o
th
C
h
in
ese
an
d
E
n
g
lis
h
C
APT
C
HA,
th
is
s
tu
d
y
f
o
cu
s
es
ex
clu
s
iv
ely
o
n
E
n
g
lis
h
-
lan
g
u
a
g
e
C
APTC
HA,
p
r
o
v
id
in
g
a
m
o
r
e
d
etaile
d
in
v
esti
g
atio
n
in
t
o
h
o
w
s
ec
u
r
ity
f
ea
tu
r
es im
p
ac
t r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
.
T
h
is
s
tu
d
y
ev
alu
ated
2
,
0
0
0
C
APT
C
HA
s
am
p
les
an
d
co
m
p
ar
ed
t
h
e
ef
f
ec
ts
o
f
s
in
g
le
an
d
m
i
x
ed
s
ec
u
r
ity
f
ea
tu
r
es.
Ho
wev
e
r
,
th
is
s
am
p
le
s
ize
m
ay
n
o
t
b
e
s
u
f
f
icien
t
to
ca
p
tu
r
e
th
e
f
u
ll
im
p
a
ct
o
f
d
ee
p
lear
n
in
g
mode
ls
.
Fu
r
th
er
s
tu
d
ies
with
lar
g
er
d
atasets
an
d
a
g
r
ea
ter
v
ar
iety
o
f
s
ec
u
r
ity
f
ea
tu
r
es
ar
e
n
ee
d
ed
to
v
alid
ate
th
ese
f
in
d
in
g
s
.
Ou
r
s
tu
d
y
d
em
o
n
s
tr
ates
th
at
wh
ile
ad
d
in
g
s
ec
u
r
ity
f
ea
t
u
r
es
lik
e
ad
v
er
s
ar
ial
n
o
is
e
ca
n
r
ed
u
ce
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
,
in
cr
ea
s
in
g
th
e
s
am
p
le
s
ize
s
ig
n
if
ican
tly
im
p
r
o
v
es
m
ac
h
in
e
r
ec
o
g
n
it
io
n
.
Fu
tu
r
e
s
tu
d
ies
m
ay
ex
p
lo
r
e
o
th
er
C
APTCHA
ty
p
es,
s
u
ch
as
im
ag
e
-
tex
t
C
APTC
HA,
wh
ile
also
c
o
n
s
id
er
in
g
u
s
ab
ilit
y
m
etr
ics.
Fu
r
th
er
m
o
r
e,
ad
v
an
ce
d
im
ag
e
g
en
er
atio
n
tech
n
i
q
u
es
lik
e
s
ty
le
tr
an
s
f
er
alg
o
r
ith
m
s
an
d
h
u
m
an
co
g
n
itiv
e
f
ac
to
r
s
co
u
ld
g
u
id
e
m
o
r
e
s
ec
u
r
e
C
APTCHA
d
ev
elo
p
m
en
t.
R
ec
en
t
o
b
s
er
v
atio
n
s
s
u
g
g
est
th
at
tex
t
C
APTC
HA
with
m
in
i
m
al
s
ec
u
r
ity
m
ea
s
u
r
es
r
em
ain
v
u
ln
er
ab
le
to
d
ee
p
lear
n
i
n
g
attac
k
s
,
ev
e
n
af
ter
im
p
lem
e
n
tin
g
ad
d
itio
n
al
f
ea
tu
r
es.
Ou
r
f
i
n
d
in
g
s
,
in
lin
e
with
p
r
ev
io
u
s
s
tu
d
ies
[
2
9
]
,
[
3
1
]
,
h
ig
h
lig
h
t
th
e
lim
itatio
n
s
o
f
th
e
cu
r
r
en
t
C
APTC
HA
d
esig
n
an
d
p
r
o
p
o
s
e
th
e
MBP
S
m
ec
h
an
is
m
f
o
r
o
p
tim
izin
g
s
ec
u
r
ity
f
ea
t
u
r
es.
T
h
ese
in
s
ig
h
ts
co
u
ld
b
e
ap
p
lied
to
en
h
an
ce
C
APTCHA
r
o
b
u
s
tn
ess
an
d
im
ag
e
g
en
e
r
atio
n
tech
n
i
q
u
es.
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
in
v
esti
g
ates
th
e
an
ti
-
attac
k
p
er
f
o
r
m
an
ce
o
f
s
ec
u
r
ity
f
ea
tu
r
es,
with
ex
p
er
im
e
n
tal
r
esu
lts
id
en
tify
in
g
t
h
e
m
o
s
t
ef
f
e
ctiv
e
o
n
es.
T
h
e
s
tu
d
y
p
r
o
p
o
s
ed
th
e
MBP
S
m
ec
h
an
is
m
to
ass
es
s
t
h
e
p
er
f
o
r
m
an
ce
o
f
s
ec
u
r
ity
f
ea
tu
r
es,
wh
ich
ca
n
b
e
u
tili
ze
d
to
o
p
tim
ize
C
APTCHA
g
en
er
atio
n
.
T
h
e
r
esu
lt
s
p
r
o
v
id
ed
a
s
o
lid
f
o
u
n
d
atio
n
f
o
r
th
e
d
ev
elo
p
m
e
n
t
o
f
m
o
r
e
r
o
b
u
s
t
C
APTCHA
s
y
s
tem
s
.
Ho
wev
er
,
o
u
r
f
in
d
in
g
s
d
em
o
n
s
tr
ate
th
at
tex
t
-
b
ased
C
APTCHA,
wh
ic
h
h
as
lim
ited
s
ec
u
r
ity
f
ea
tu
r
es,
r
em
ain
s
v
u
ln
er
a
b
le
to
r
ec
o
g
n
itio
n
al
g
o
r
ith
m
s
b
ased
o
n
d
ee
p
lear
n
in
g
.
Fu
tu
r
e
wo
r
k
m
a
y
ex
p
lo
r
e
h
y
b
r
id
C
APT
C
HAs
th
at
in
teg
r
ate
b
o
th
ad
v
an
ce
d
im
ag
e
g
en
er
atio
n
tech
n
iq
u
es
an
d
h
u
m
an
s
em
an
tic
u
n
d
er
s
tan
d
in
g
to
e
n
h
an
ce
th
e
s
ec
u
r
ity
.
B
y
co
m
b
in
in
g
th
e
co
n
s
id
er
atio
n
o
f
s
ec
u
r
ity
f
ea
t
u
r
es
with
h
u
m
an
c
o
g
n
i
tiv
e
ca
p
a
b
ilit
ies,
f
u
tu
r
e
C
APTCHA
r
es
ea
r
ch
co
u
l
d
d
esig
n
m
o
r
e
s
ec
u
r
e
im
a
g
e
-
tex
t CAP
T
C
HA.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
e
a
u
th
o
r
s
th
an
k
th
e
C
o
lleg
e
o
f
C
o
m
p
u
tin
g
,
I
n
f
o
r
m
atics
an
d
Ma
th
em
atics
(
KPPIM
)
an
d
Un
iv
er
s
iti
T
ek
n
o
lo
g
i M
AR
A
(
UiT
M)
f
o
r
all
th
eir
h
elp
an
d
s
u
p
p
o
r
t in
t
h
is
r
esear
ch
.
RE
F
E
R
E
NC
E
S
[
1
]
J
.
M
.
T
o
r
r
e
s,
C
.
I
.
C
o
m
e
sañ
a
,
a
n
d
P
.
J.
G
a
r
c
í
a
-
N
i
e
t
o
,
“
R
e
v
i
e
w
:
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
a
p
p
l
i
e
d
t
o
c
y
b
e
r
se
c
u
r
i
t
y
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Ma
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
C
y
b
e
rn
e
t
i
c
s
,
v
o
l
.
1
0
,
n
o
.
1
0
,
p
p
.
2
8
2
3
–
2
8
3
6
,
O
c
t
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
3
0
4
2
-
018
-
0
0
9
0
6
-
1.
[
2
]
O
.
E.
I
g
b
e
k
e
l
e
,
A
.
A
.
A
d
e
b
i
y
i
,
A
.
I
.
F
r
a
n
c
i
s
,
A
.
O.
M
a
r
i
o
n
,
a
n
d
O
.
O
.
O
l
u
d
a
y
o
,
“
R
e
s
e
a
r
c
h
t
r
e
n
d
s
o
n
C
A
P
TC
H
A
:
a
s
y
s
t
e
m
a
t
i
c
l
i
t
e
r
a
t
u
r
e
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
(
I
J
E
C
E)
,
v
o
l
.
1
1
,
n
o
.
5
,
p
.
4
3
0
0
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
1
i
5
.
p
p
4
3
0
0
-
4
3
1
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
1
,
Ap
r
il
20
2
5
:
5
3
5
-
5
4
5
544
[
3
]
L.
V
.
A
h
n
,
M
.
B
l
u
m,
a
n
d
J.
La
n
g
f
o
r
d
,
“
Te
l
l
i
n
g
h
u
m
a
n
s
a
n
d
c
o
mp
u
t
e
r
s
a
p
a
r
t
a
u
t
o
mat
i
c
a
l
l
y
,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
f
t
h
e
A
C
M
,
v
o
l
.
4
7
,
n
o
.
2
,
p
p
.
5
6
–
6
0
,
F
e
b
.
2
0
0
4
,
d
o
i
:
1
0
.
1
1
4
5
/
9
6
6
3
8
9
.
9
6
6
3
9
0
.
[
4
]
C
.
J.
H
e
r
n
a
n
d
e
z
-
C
a
s
t
r
o
a
n
d
A
.
R
i
b
a
g
o
r
d
a
,
“
P
i
t
f
a
l
l
s
i
n
C
A
P
TC
H
A
d
e
si
g
n
a
n
d
i
m
p
l
e
me
n
t
a
t
i
o
n
:
t
h
e
m
a
t
h
C
A
P
TC
H
A
,
a
c
a
se
st
u
d
y
,
”
C
o
m
p
u
t
e
rs
a
n
d
S
e
c
u
ri
t
y
,
v
o
l
.
2
9
,
n
o
.
1
,
p
p
.
1
4
1
–
1
5
7
,
F
e
b
.
2
0
1
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
se.
2
0
0
9
.
0
6
.
0
0
6
.
[
5
]
M
.
K
u
ma
r
,
M
.
K
.
Ji
n
d
a
l
,
a
n
d
M
.
K
u
ma
r
,
“
A
sy
st
e
m
a
t
i
c
su
r
v
e
y
o
n
C
A
P
T
C
H
A
r
e
c
o
g
n
i
t
i
o
n
:
t
y
p
e
s,
c
r
e
a
t
i
o
n
a
n
d
b
r
e
a
k
i
n
g
t
e
c
h
n
i
q
u
e
s
,
”
A
rch
i
v
e
s
o
f
C
o
m
p
u
t
a
t
i
o
n
a
l
Me
t
h
o
d
s
i
n
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
2
9
,
n
o
.
2
,
p
p
.
1
1
0
7
–
1
1
3
6
,
M
a
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
8
3
1
-
021
-
0
9
6
0
8
-
4.
[
6
]
H
.
S
h
a
o
,
Y
.
X
i
a
,
K
.
M
e
n
g
,
a
n
d
C
.
P
i
a
o
,
“
S
t
u
d
y
o
f
h
o
l
l
o
w
l
e
t
t
e
r
C
A
P
T
C
H
A
s
r
e
c
o
g
n
i
t
i
o
n
t
e
c
h
n
o
l
o
g
y
b
a
se
d
o
n
c
o
l
o
r
f
i
l
l
i
n
g
a
l
g
o
r
i
t
h
m,
”
J
o
u
rn
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
S
y
st
e
m
s
,
v
o
l
.
1
9
,
n
o
.
4
,
p
p
.
5
4
0
–
5
5
3
,
2
0
2
3
,
d
o
i
:
1
0
.
3
7
4
5
/
JI
P
S
.
0
2
.
0
2
0
2
.
[
7
]
S
.
K
i
m
a
n
d
S
.
C
h
o
i
,
“
D
o
t
C
H
A
:
a
n
i
n
t
e
r
a
c
t
i
v
e
3
D
t
e
x
t
-
b
a
s
e
d
C
A
P
T
C
H
A
,
”
J
o
u
r
n
a
l
o
f
We
b
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
8
,
n
o
.
8
,
p
p
.
8
3
7
–
8
6
4
,
J
a
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
3
0
5
2
/
j
w
e
1
5
4
0
-
9
5
8
9
.
1
8
8
4
.
[
8
]
H
.
G
a
o
,
M
.
Ta
n
g
,
Y
.
Li
u
,
P
.
Z
h
a
n
g
,
a
n
d
X
.
L
i
u
,
“
R
e
sea
r
c
h
o
n
t
h
e
s
e
c
u
r
i
t
y
o
f
m
i
c
r
o
so
f
t
’
s
t
w
o
-
l
a
y
e
r
c
ap
t
c
h
a
,
”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
I
n
f
o
rm
a
t
i
o
n
Fo
r
e
n
si
c
s a
n
d
S
e
c
u
r
i
t
y
,
v
o
l
.
1
2
,
n
o
.
7
,
p
p
.
1
6
7
1
–
1
6
8
5
,
Ju
l
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
F
S
.
2
0
1
7
.
2
6
8
2
7
0
4
.
[
9
]
J.
Z
h
a
n
g
e
t
a
l
.
,
“
A
s
e
c
u
r
e
a
n
n
u
l
i
C
A
P
TC
H
A
s
y
s
t
e
m,
”
C
o
m
p
u
t
e
rs
a
n
d
S
e
c
u
r
i
t
y
,
v
o
l
.
1
2
5
,
p
.
1
0
3
0
2
5
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
se.
2
0
2
2
.
1
0
3
0
2
5
.
[
1
0
]
G
.
C
h
a
n
g
e
t
a
l
.
,
“
T
h
e
r
o
b
u
st
n
e
s
s
o
f
b
e
h
a
v
i
o
r
-
v
e
r
i
f
i
c
a
t
i
o
n
-
b
a
se
d
sl
i
d
e
r
C
A
P
TC
H
A
s,
”
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
S
e
c
u
ri
t
y
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
1
,
p
.
1
0
3
7
1
1
,
M
a
r
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
i
sa
.
2
0
2
4
.
1
0
3
7
1
1
.
[
1
1
]
M
.
M
o
r
a
d
i
a
n
d
M
.
R
.
K
e
y
v
a
n
p
o
u
r
,
“
A
n
o
v
e
l
C
A
P
TC
H
A
s
c
h
e
me
b
a
s
e
d
o
n
f
a
c
i
a
l
e
x
p
r
e
ss
i
o
n
r
e
c
o
n
st
r
u
c
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
t
r
o
n
i
c
B
u
s
i
n
e
ss
,
v
o
l
.
1
5
,
n
o
.
4
,
p
p
.
3
6
8
–
3
8
8
,
2
0
2
0
,
d
o
i
:
1
0
.
1
5
0
4
/
I
JEB
.
2
0
2
0
.
1
1
1
0
6
1
.
[
1
2
]
Y
.
C
.
S
.
R
e
d
d
y
,
M
.
V
.
R
a
o
,
M
.
K
.
R
a
o
,
C
.
V
.
P
.
K
u
mar,
a
n
d
A
.
A
.
S
a
i
,
“
G
r
a
p
h
i
c
a
l
p
a
ssw
o
r
d
u
si
n
g
C
A
P
T
C
H
A
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Ad
v
a
n
c
e
s
i
n
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
5
,
n
o
.
2
,
p
.
9
4
,
J
u
n
.
2
0
1
6
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
a
a
s.
v
5
.
i
2
.
p
p
9
4
-
1
0
0
.
[
1
3
]
M
.
T
a
n
g
,
H
.
G
a
o
,
Y
.
Z
h
a
n
g
,
Y
.
L
i
u
,
P
.
Zh
a
n
g
,
a
n
d
P
.
W
a
n
g
,
“
R
e
se
a
r
c
h
o
n
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
i
n
b
r
e
a
k
i
n
g
t
e
x
t
-
b
a
se
d
c
a
p
t
c
h
a
s
a
n
d
d
e
si
g
n
i
n
g
i
ma
g
e
-
b
a
se
d
c
a
p
t
c
h
a
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
f
o
rm
a
t
i
o
n
Fo
r
e
n
s
i
c
s
a
n
d
S
e
c
u
ri
t
y
,
v
o
l
.
1
3
,
n
o
.
1
0
,
p
p
.
2
5
2
2
–
2
5
3
7
,
O
c
t
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
F
S
.
2
0
1
8
.
2
8
2
1
0
9
6
.
[
1
4
]
Y
.
W
e
n
,
“
R
o
b
u
st
i
m
a
g
e
-
b
a
se
d
C
A
P
T
C
H
A
g
e
n
e
r
a
t
i
o
n
u
si
n
g
a
d
v
e
r
sar
i
a
l
a
t
t
a
c
k
,
”
i
n
T
h
i
rd
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
C
o
m
p
u
t
i
n
g
a
n
d
H
u
m
a
n
-
C
o
m
p
u
t
e
r
I
n
t
e
r
a
c
t
i
o
n
(
I
C
H
C
I
2
0
2
2
)
,
K
.
S
u
b
r
a
ma
n
i
a
n
,
Ed
.
,
S
P
I
E,
Jan
.
2
0
2
3
,
p
.
7
1
.
d
o
i
:
1
0
.
1
1
1
7
/
1
2
.
2
6
5
5
9
3
4
.
[
1
5
]
H
.
K
w
o
n
,
H
.
Y
o
o
n
,
a
n
d
K
.
W
.
P
a
r
k
,
“
C
A
P
TC
H
A
i
ma
g
e
g
e
n
e
r
a
t
i
o
n
:
t
w
o
-
st
e
p
st
y
l
e
-
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
i
n
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s
,
”
S
e
n
so
rs (S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
2
0
,
n
o
.
5
,
p
.
1
4
9
5
,
M
a
r
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
0
0
5
1
4
9
5
.
[
1
6
]
V
.
S
.
R
a
t
h
o
r
,
B
.
G
a
r
g
,
M
.
P
a
t
i
l
,
a
n
d
G
.
K
.
S
h
a
r
ma
,
“
S
e
c
u
r
i
t
y
a
n
a
l
y
s
i
s
o
f
i
ma
g
e
C
A
P
T
C
H
A
u
s
i
n
g
a
m
a
sk
R
-
C
N
N
-
b
a
se
d
a
t
t
a
c
k
mo
d
e
l
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ad
H
o
c
a
n
d
U
b
i
q
u
i
t
o
u
s
C
o
m
p
u
t
i
n
g
,
v
o
l
.
3
6
,
n
o
.
4
,
p
p
.
2
3
8
–
2
4
7
,
2
0
2
1
,
d
o
i
:
1
0
.
1
5
0
4
/
I
JA
H
U
C
.
2
0
2
1
.
1
1
4
1
0
8
.
[
1
7
]
Z.
C
h
e
n
g
,
H
.
G
a
o
,
Z.
Li
u
,
H
.
W
u
,
Y
.
Zi
,
a
n
d
G
.
P
e
i
,
“
I
mag
e
-
b
a
se
d
C
A
P
T
C
H
A
s
b
a
se
d
o
n
n
e
u
r
a
l
s
t
y
l
e
t
r
a
n
sf
e
r
,
”
I
ET
I
n
f
o
rm
a
t
i
o
n
S
e
c
u
r
i
t
y
,
v
o
l
.
1
3
,
n
o
.
6
,
p
p
.
5
1
9
–
5
2
9
,
N
o
v
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
4
9
/
i
e
t
-
i
f
s.
2
0
1
8
.
5
0
3
6
.
[
1
8
]
R
.
S
h
a
o
,
Z
.
S
h
i
,
J
.
Y
i
,
P
.
Y
.
C
h
e
n
,
a
n
d
C
.
J.
H
si
e
h
,
“
R
o
b
u
s
t
t
e
x
t
C
A
P
T
C
H
A
s
u
s
i
n
g
a
d
v
e
r
sari
a
l
e
x
a
m
p
l
e
s,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
-
2
0
2
2
I
EEE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Bi
g
D
a
t
a
,
B
i
g
D
a
t
a
2
0
2
2
,
I
EEE,
D
e
c
.
2
0
2
2
,
p
p
.
1
4
9
5
–
1
5
0
4
.
d
o
i
:
1
0
.
1
1
0
9
/
B
i
g
D
a
t
a
5
5
6
6
0
.
2
0
2
2
.
1
0
0
2
1
1
0
0
.
[
1
9
]
D
.
H
i
t
a
j
,
B
.
H
i
t
a
j
,
S
.
J
a
j
o
d
i
a
,
a
n
d
L.
V
.
M
a
n
c
i
n
i
,
“
C
a
p
t
u
r
e
t
h
e
b
o
t
:
u
s
i
n
g
a
d
v
e
r
sari
a
l
e
x
a
mp
l
e
s
t
o
i
m
p
r
o
v
e
C
A
P
TC
H
A
r
o
b
u
s
t
n
e
ss
t
o
b
o
t
a
t
t
a
c
k
s,”
I
EEE
I
n
t
e
l
l
i
g
e
n
t
S
y
st
e
m
s
,
v
o
l
.
3
6
,
n
o
.
5
,
p
p
.
1
0
4
–
1
1
2
,
S
e
p
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
M
I
S
.
2
0
2
0
.
3
0
3
6
1
5
6
.
[
2
0
]
N
.
B
.
A
r
d
h
i
t
a
a
n
d
N
.
U
.
M
a
u
l
i
d
e
v
i
,
“
R
o
b
u
st
a
d
v
e
r
sari
a
l
e
x
a
mp
l
e
a
s
c
a
p
t
c
h
a
g
e
n
e
r
a
t
o
r
,
”
i
n
2
0
2
0
7
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ad
v
a
n
c
e
d
I
n
f
o
rm
a
t
i
c
s:
C
o
n
c
e
p
t
s
,
T
h
e
o
r
y
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s,
I
C
A
I
C
T
A
2
0
2
0
,
I
EEE,
S
e
p
.
2
0
2
0
,
p
p
.
1
–
4
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
C
TA
4
9
8
6
1
.
2
0
2
0
.
9
4
2
9
0
4
8
.
[
2
1
]
N
.
D
i
n
h
,
K
.
Tr
a
n
-
Tr
u
n
g
,
a
n
d
V
.
T
.
H
o
a
n
g
,
“
A
u
g
me
n
t
C
A
P
TC
H
A
s
e
c
u
r
i
t
y
u
si
n
g
a
d
v
e
r
sar
i
a
l
e
x
a
mp
l
e
s
w
i
t
h
n
e
u
r
a
l
s
t
y
l
e
t
r
a
n
sf
e
r
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
8
3
5
5
3
–
8
3
5
6
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
9
8
4
4
2
.
[
2
2
]
M
.
O
sad
c
h
y
,
J.
H
e
r
n
a
n
d
e
z
-
C
a
st
r
o
,
S
.
G
i
b
so
n
,
O
.
D
u
n
k
e
l
m
a
n
,
a
n
d
D
.
P
e
r
e
z
-
C
a
b
o
,
“
N
o
b
o
t
e
x
p
e
c
t
s
t
h
e
d
e
e
p
C
A
P
TC
H
A
!
I
n
t
r
o
d
u
c
i
n
g
i
mm
u
t
a
b
l
e
a
d
v
e
r
sari
a
l
e
x
a
m
p
l
e
s,
w
i
t
h
a
p
p
l
i
c
a
t
i
o
n
s
t
o
C
A
P
T
C
H
A
ge
n
e
r
a
t
i
o
n
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
f
o
rm
a
t
i
o
n
Fo
re
n
s
i
c
s
a
n
d
S
e
c
u
ri
t
y
,
v
o
l
.
1
2
,
n
o
.
1
1
,
p
p
.
2
6
4
0
–
2
6
5
3
,
N
o
v
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
F
S
.
2
0
1
7
.
2
7
1
8
4
7
9
.
[
2
3
]
P
.
W
a
n
g
,
H
.
G
a
o
,
C
.
X
i
a
o
,
X
.
G
u
o
,
Y
.
G
a
o
,
a
n
d
Y
.
Zi
,
“
Ex
t
e
n
d
e
d
r
e
se
a
r
c
h
o
n
t
h
e
s
e
c
u
r
i
t
y
o
f
v
i
s
u
a
l
r
e
a
s
o
n
i
n
g
C
A
P
TC
H
A
,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
D
e
p
e
n
d
a
b
l
e
a
n
d
S
e
c
u
re
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
0
,
n
o
.
6
,
p
p
.
4
9
7
6
–
4
9
9
2
,
N
o
v
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TD
S
C
.
2
0
2
3
.
3
2
3
8
4
0
8
.
[
2
4
]
N
.
D
.
Tr
o
n
g
,
T.
H
.
H
u
o
n
g
,
a
n
d
V
.
T
.
H
o
a
n
g
,
“
N
e
w
c
o
g
n
i
t
i
v
e
d
e
e
p
-
l
e
a
r
n
i
n
g
C
A
P
TC
H
A
,
”
S
e
n
so
rs
,
v
o
l
.
2
3
,
n
o
.
4
,
p
.
2
3
3
8
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s2
3
0
4
2
3
3
8
.
[
2
5
]
X
.
Ji
a
,
J
.
X
i
a
o
,
a
n
d
C
.
W
u
,
“
TI
C
S
:
t
e
x
t
–
i
ma
g
e
-
b
a
s
e
d
s
e
ma
n
t
i
c
C
A
P
TC
H
A
sy
n
t
h
e
si
s
v
i
a
m
u
l
t
i
-
c
o
n
d
i
t
i
o
n
a
d
v
e
r
sari
a
l
l
e
a
r
n
i
n
g
,
”
Vi
su
a
l
C
o
m
p
u
t
e
r
,
v
o
l
.
3
8
,
n
o
.
3
,
p
p
.
9
6
3
–
9
7
5
,
M
a
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
3
7
1
-
021
-
0
2
0
6
1
-
1.
[
2
6
]
M
.
G
u
e
r
a
r
,
L
.
V
e
r
d
e
r
a
me,
M
.
M
i
g
l
i
a
r
d
i
,
F
.
P
a
l
mi
e
r
i
,
a
n
d
A
.
M
e
r
l
o
,
“
G
o
t
t
a
C
A
P
T
C
H
A
’
Em
a
l
l
:
a
s
u
r
v
e
y
o
f
2
0
y
e
a
r
s
o
f
t
h
e
h
u
m
a
n
-
or
-
c
o
mp
u
t
e
r
d
i
l
e
mm
a
,
”
A
C
M
C
o
m
p
u
t
i
n
g
S
u
r
v
e
y
s
,
v
o
l
.
5
4
,
n
o
.
9
,
p
p
.
1
–
3
3
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
4
5
/
3
4
7
7
1
4
2
.
[
2
7
]
Y
.
Zi
,
H
.
G
a
o
,
Z
.
C
h
e
n
g
,
a
n
d
Y
.
Li
u
,
“
A
n
e
n
d
-
to
-
e
n
d
a
t
t
a
c
k
o
n
t
e
x
t
C
A
P
T
C
H
A
s,”
I
EEE
T
r
a
n
sa
c
t
i
o
n
s
o
n
I
n
f
o
r
m
a
t
i
o
n
F
o
re
n
si
c
s
a
n
d
S
e
c
u
ri
t
y
,
v
o
l
.
1
5
,
p
p
.
7
5
3
–
7
6
6
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
F
S
.
2
0
1
9
.
2
9
2
8
6
2
2
.
[
2
8
]
G
.
Y
e
e
t
a
l
.
,
“
U
si
n
g
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
sari
a
l
n
e
t
w
o
r
k
s
t
o
b
r
e
a
k
a
n
d
p
r
o
t
e
c
t
t
e
x
t
c
a
p
t
c
h
a
s
,
”
AC
M
T
ra
n
s
a
c
t
i
o
n
s
o
n
Pri
v
a
c
y
a
n
d
S
e
c
u
r
i
t
y
,
v
o
l
.
2
3
,
n
o
.
2
,
p
p
.
1
–
2
9
,
M
a
y
2
0
2
0
,
d
o
i
:
1
0
.
1
1
4
5
/
3
3
7
8
4
4
6
.
[
2
9
]
P
.
W
a
n
g
,
H
.
G
a
o
,
X
.
G
u
o
,
C
.
X
i
a
o
,
F
.
Q
i
,
a
n
d
Z.
Y
a
n
,
“
A
n
e
x
p
e
r
i
me
n
t
a
l
i
n
v
e
s
t
i
g
a
t
i
o
n
o
f
t
e
x
t
-
b
a
s
e
d
C
A
P
T
C
H
A
a
t
t
a
c
k
s
a
n
d
t
h
e
i
r
r
o
b
u
st
n
e
s
s,”
AC
M
C
o
m
p
u
t
i
n
g
S
u
rve
y
s
,
v
o
l
.
5
5
,
n
o
.
9
,
p
p
.
1
–
3
8
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
4
5
/
3
5
5
9
7
5
4
.
[
3
0
]
C
.
S
h
i
e
t
a
l
.
,
“
A
d
v
e
r
sar
i
a
l
C
A
P
T
C
H
A
s,”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
C
y
b
e
r
n
e
t
i
c
s
,
v
o
l
.
5
2
,
n
o
.
7
,
p
p
.
6
0
9
5
–
6
1
0
8
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
T
C
Y
B
.
2
0
2
1
.
3
0
7
1
3
9
5
.
[
3
1
]
Y
.
M
a
t
s
u
u
r
a
,
H
.
K
a
t
o
,
a
n
d
I
.
S
a
sas
e
,
“
A
d
v
e
r
sar
i
a
l
t
e
x
t
-
b
a
s
e
d
C
A
P
T
C
H
A
g
e
n
e
r
a
t
i
o
n
me
t
h
o
d
u
t
i
l
i
z
i
n
g
s
p
a
t
i
a
l
sm
o
o
t
h
i
n
g
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
I
EEE
G
l
o
b
a
l
C
o
m
m
u
n
i
c
a
t
i
o
n
s
C
o
n
f
e
r
e
n
c
e
,
G
L
O
BEC
O
M
,
I
EEE,
D
e
c
.
2
0
2
1
,
p
p
.
1
–
6
.
d
o
i
:
1
0
.
1
1
0
9
/
G
LO
B
EC
O
M
4
6
5
1
0
.
2
0
2
1
.
9
6
8
5
0
4
6
.
[
3
2
]
I
.
K
e
s
h
t
a
,
P
.
S
.
D
e
s
h
p
a
n
d
e
,
M
.
S
h
a
b
a
z
,
M
.
S
o
n
i
,
M
.
K.
B
h
a
d
l
a
,
a
n
d
Y
.
M
u
h
a
mm
e
d
,
“
M
u
l
t
i
-
st
a
g
e
b
i
o
m
e
d
i
c
a
l
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
e
x
t
r
a
c
t
i
o
n
a
l
g
o
r
i
t
h
m
f
o
r
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
,
”
S
N
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
5
,
n
o
.
5
,
p
.
1
3
1
,
M
a
y
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
4
2
4
5
2
-
023
-
0
5
3
3
9
-
2.
Evaluation Warning : The document was created with Spire.PDF for Python.