C
omp
u
te
r
S
c
i
e
n
c
e
an
d
I
n
for
mati
on
T
e
c
h
n
ol
ogi
e
s
V
ol
.
6
,
N
o
.
3
,
N
ove
m
b
e
r
20
25
,
p
p.
274
~
282
IS
S
N
:
2722
-
3221
,
D
O
I
:
10.
1
1591
/
c
s
i
t
.
v
6
i
3
.
pp
27
4
-
282
274
Jou
r
n
al
h
o
m
e
pa
ge
:
ht
t
p:
/
/
i
ae
s
pr
i
m
e
.
c
om
/
i
nd
e
x
.
php
/
c
s
i
t
Op
t
i
mi
z
i
n
g
d
i
p
l
o
m
a
t
i
c
i
n
d
e
x
i
n
g
:
f
u
l
l
-
p
a
r
a
m
e
t
e
r
v
s
l
o
w
-
r
a
n
k
a
d
a
p
t
a
t
i
o
n
f
o
r
mul
t
i
-
l
a
b
e
l
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
d
i
p
l
o
m
a
t
i
c
c
a
b
l
e
s
D
e
l
a
N
u
r
l
a
i
l
a
,
A
b
b
a
S
u
gan
d
a
G
i
r
s
an
g
D
e
p
a
rt
m
e
n
t
o
f
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
,
BIN
U
S
G
ra
d
u
a
t
e
P
ro
g
ra
m
-
M
a
s
t
e
r
o
f
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
,
Bi
n
a
N
u
s
a
n
t
a
ra
U
n
i
v
e
rs
i
t
y
,
J
a
k
a
rt
a
,
In
d
o
n
e
s
i
a
A
r
ti
c
l
e
I
n
fo
A
BS
TR
A
C
T
Ar
t
i
c
l
e
h
i
s
t
or
y
:
Re
c
e
i
v
e
d
S
e
p
19
,
2
024
Re
vi
s
e
d
J
un
9
,
2025
A
c
c
e
pt
e
d
J
un
13
,
2025
A
c
c
ur
a
t
e
c
l
a
s
s
i
f
i
c
a
t
i
on
of
d
i
pl
o
m
a
t
i
c
c
a
bl
e
s
i
s
c
r
uc
i
a
l
f
o
r
M
i
s
s
i
on
’
s
e
va
l
ua
t
i
on
a
n
d
po
l
i
c
y
f
o
r
m
u
l
a
t
i
o
n.
H
ow
e
ve
r
,
t
he
s
e
d
oc
u
m
e
nt
s
of
t
e
n
c
ove
r
m
u
l
t
i
p
l
e
t
op
i
c
s
,
he
n
c
e
a
m
u
l
t
i
-
l
a
be
l
c
l
a
s
s
i
f
i
c
a
t
i
on
a
ppr
oa
c
h
i
s
n
e
c
e
s
s
a
r
y
.
T
hi
s
r
e
s
e
a
r
c
h
e
xp
l
o
r
e
s
t
he
a
p
pl
i
c
a
t
i
on
o
f
p
r
e
-
t
r
a
i
ne
d
l
a
n
gua
g
e
m
od
e
l
s
(
C
a
hy
a
B
E
R
T
,
I
ndoB
E
R
T
,
a
nd
M
B
E
R
T
)
f
o
r
m
u
l
t
i
-
l
a
b
e
l
c
l
a
s
s
i
f
i
c
a
t
i
on
o
f
di
p
l
o
m
a
t
i
c
c
a
b
l
e
e
x
e
c
u
t
i
ve
s
u
m
m
a
r
i
e
s
,
w
h
i
c
h
a
l
i
gn
w
i
t
h
t
he
d
i
pl
o
m
a
t
i
c
r
e
p
r
e
s
e
n
t
a
t
i
o
n
i
nd
e
x
.
T
he
s
t
u
dy
c
o
m
p
a
r
e
s
f
u
l
l
-
p
a
r
a
m
e
t
e
r
f
i
n
e
-
t
u
ni
ng
a
nd
l
ow
-
r
a
n
k
a
d
a
pt
a
t
i
o
n
(
L
oR
A
)
t
e
c
h
ni
que
s
us
i
ng
c
a
b
l
e
s
f
r
o
m
2
022
-
2023
.
R
e
s
u
l
t
s
de
m
ons
t
r
a
t
e
t
ha
t
I
ndo
ne
s
i
a
n
-
s
pe
c
i
f
i
c
m
ode
l
s
,
pa
r
t
i
c
u
l
a
r
l
y
t
h
e
I
ndoB
E
R
T
,
out
pe
r
f
o
r
m
m
u
l
t
i
l
i
n
gua
l
m
od
e
l
s
i
n
c
l
a
s
s
i
f
i
c
a
t
i
on
a
c
c
u
r
a
c
y.
W
hi
l
e
L
oR
A
s
h
ow
e
d
s
l
i
g
ht
l
y
l
ow
e
r
pe
r
f
or
m
a
nc
e
t
ha
n
f
ul
l
f
i
ne
-
t
un
i
ng
,
i
t
s
i
gn
i
f
i
c
a
n
t
l
y
r
e
du
c
e
d
G
P
U
m
e
m
or
y
us
a
ge
by
48
%
a
nd
t
r
a
i
n
i
ng
t
i
m
e
by
69.
7
%
.
T
h
e
s
e
f
i
nd
i
n
gs
hi
g
hl
i
gh
t
L
oR
A
’
s
po
t
e
n
t
i
a
l
f
or
r
e
s
ou
r
c
e
-
c
o
ns
t
r
a
i
ne
d
di
p
l
o
m
a
t
i
c
i
ns
t
i
t
u
t
i
on
s
,
a
d
va
n
c
i
n
g
n
a
t
ur
a
l
l
a
ngu
a
ge
p
r
oc
e
s
s
i
ng
i
n
d
i
pl
o
m
a
c
y
a
nd
o
f
f
e
r
i
ng
p
a
t
hw
a
ys
f
o
r
e
f
f
i
c
i
e
n
t
,
r
e
a
l
-
t
i
m
e
m
ul
t
i
-
l
a
b
e
l
c
l
a
s
s
i
f
i
c
a
t
i
on
t
o
e
nha
n
c
e
d
i
pl
o
m
a
t
i
c
m
i
s
s
i
on
e
va
l
ua
t
i
on
.
Ke
y
w
or
d
s
:
D
i
pl
o
m
a
t
i
c
c
a
b
l
e
s
D
i
pl
o
m
a
t
i
c
i
nde
x
F
ul
l
-
p
a
ra
m
e
t
e
r
t
uni
n
g
L
ow
-
ra
nk
a
d
a
pt
a
t
i
on
M
ul
t
i
-
l
a
be
l
c
l
a
s
s
i
fi
c
a
t
i
on
T
hi
s
i
s
an
op
e
n
ac
c
e
s
s
ar
t
i
c
l
e
u
nde
r
t
he
C
C
B
Y
-
SA
l
i
c
e
n
s
e
.
Cor
r
e
s
pon
di
n
g
Au
t
h
or
:
D
e
l
a
N
ur
l
a
i
l
a
D
e
pa
r
t
m
e
nt
of
Co
m
pu
t
e
r
S
c
i
e
n
c
e
,
BIN
U
S
G
r
a
dua
t
e
P
rogr
a
m
-
M
a
s
t
e
r
of
Co
m
pu
t
e
r
S
c
i
e
n
c
e
Bi
na
N
us
a
n
t
a
ra
U
ni
v
e
rs
i
t
y
J
a
k
a
r
t
a
11480
,
Indon
e
s
i
a
E
m
a
i
l
:
de
l
a
.
nur
l
a
i
l
a
@
b
i
nus
.
a
c
.
i
d
1.
I
N
TR
O
D
U
C
TI
O
N
F
re
e
do
m
of
c
om
m
un
i
c
a
t
i
on
i
s
g
ua
r
a
nt
e
e
d
t
o
d
i
pl
o
m
a
t
i
c
m
i
s
s
i
ons
by
A
r
t
i
c
l
e
2
7
of
t
h
e
V
i
e
n
na
Conve
n
t
i
o
n
of
1961
[
1]
.
H
ow
d
i
pl
o
m
a
c
y
w
or
ks
h
a
s
c
h
a
ng
e
d
s
i
n
c
e
t
e
c
hno
l
og
i
c
a
l
a
dva
n
c
e
m
e
nt
s
e
xi
s
t
e
d
[2]
.
N
e
ve
r
t
he
l
e
s
s
,
di
p
l
o
m
a
t
i
c
c
a
bl
e
s
re
m
a
i
n
a
n
i
m
port
a
nt
for
m
of
c
o
m
m
un
i
c
a
t
i
on
i
n
t
he
w
orl
d
of
d
i
pl
o
m
a
c
y
,
e
ve
n
t
hough
t
h
e
c
o
m
m
u
ni
c
a
t
i
on
t
e
c
hn
ol
ogy
b
e
i
ng
us
e
d
h
a
s
e
vol
v
e
d.
A
c
a
bl
e
r
e
pr
e
s
e
n
t
s
di
pl
o
m
a
t
i
c
c
orre
s
p
onde
nc
e
be
t
w
e
e
n
a
f
ore
i
gn
m
i
n
i
s
t
ry
i
n
t
h
e
ho
m
e
c
ou
nt
ry
a
n
d
i
t
s
di
p
l
o
m
a
t
i
c
m
i
s
s
i
ons
a
bro
a
d
[3
]
.
T
hro
ugh
t
he
s
e
c
a
b
l
e
s
,
s
e
ns
i
t
i
ve
i
nf
orm
a
t
i
on
i
s
e
xc
ha
n
ge
d
,
a
n
i
n
-
d
e
p
t
h
a
na
l
ys
i
s
o
f
t
h
e
a
c
c
re
di
t
e
d
c
o
unt
r
i
e
s
i
s
r
e
por
t
e
d
,
pot
e
nt
i
a
l
c
oop
e
ra
t
i
on
i
s
e
xp
l
ore
d,
a
nd
i
ns
t
ruc
t
i
ons
a
re
g
i
ve
n
from
t
he
hom
e
gov
e
rn
m
e
n
t
.
T
h
e
va
s
t
a
m
o
unt
of
d
a
t
a
e
xc
h
a
ng
e
d
prov
i
de
s
a
ri
c
h
s
ourc
e
of
i
nfor
m
a
t
i
o
n
fo
r
e
va
l
ua
t
i
ng
di
p
l
om
a
t
i
c
m
i
s
s
i
ons
.
A
c
c
ord
i
ng
t
o
B
j
ol
a
,
i
n
t
oda
y
’s
e
ra
,
d
a
t
a
i
s
“
t
h
e
ne
w
oi
l
”
[4]
,
a
nd
a
s
i
gn
i
fi
c
a
nt
a
s
s
e
t
t
o
i
t
s
ow
ne
r.
T
h
e
i
nfo
rm
a
t
i
on
gl
e
a
n
e
d
fro
m
di
pl
o
m
a
t
i
c
c
a
b
l
e
s
c
a
n
b
e
us
e
d
t
o
ga
uge
a
M
i
s
s
i
on’s
pe
rfo
rm
a
nc
e
.
T
h
e
d
i
pl
om
a
t
i
c
r
e
pr
e
s
e
n
t
a
t
i
on
i
nd
e
x
s
e
rve
s
a
s
a
nu
m
e
ri
c
a
l
a
s
s
e
s
s
m
e
nt
s
ys
t
e
m
t
h
a
t
c
ov
e
rs
a
ra
ng
e
of
pe
r
form
a
nc
e
i
n
di
c
a
t
or
s
for
a
re
a
s
i
n
c
l
u
di
ng
pol
i
t
i
c
s
,
e
c
on
om
y
,
prot
oc
o
l
a
nd
c
ons
u
l
a
r
s
e
rvi
c
e
s
,
s
o
c
i
o
-
c
ul
t
ur
a
l
a
ff
a
i
rs
,
a
nd
a
d
m
i
n
i
s
t
r
a
t
i
on.
H
ow
e
ve
r
,
a
c
c
ur
a
t
e
l
y
c
l
a
s
s
i
fy
i
ng
a
nd
i
n
de
x
i
ng
t
hi
s
w
e
a
l
t
h
of
i
nfor
m
a
t
i
o
n
pos
e
s
a
s
i
gn
i
fi
c
a
nt
c
ha
l
l
e
nge
.
T
h
e
c
om
p
l
e
x
i
t
y
of
d
i
p
l
om
a
t
i
c
w
ork,
s
pa
n
ni
n
g
m
u
l
t
i
pl
e
fun
c
t
i
on
a
l
dom
a
i
ns
,
m
a
k
e
s
m
a
nu
a
l
c
l
a
s
s
i
fi
c
a
t
i
on
i
ne
ff
i
c
i
e
n
t
a
nd
pr
one
t
o
e
rror
.
A
c
c
ura
t
e
c
l
a
s
s
i
fi
c
a
t
i
on
i
s
e
s
s
e
n
t
i
a
l
not
o
nl
y
t
o
s
upp
ort
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
IS
S
N
:
2722
-
3221
O
pt
i
m
i
z
i
ng
di
p
l
om
at
i
c
i
n
de
x
i
ng:
f
ul
l
-
par
am
e
t
e
r
v
s
l
ow
-
r
a
nk
adapt
a
t
i
on
f
or
m
u
l
t
i
-
l
ab
e
l
…
(
D
e
l
a
Nur
l
ai
l
a
)
275
di
pl
o
m
a
t
i
c
r
e
pre
s
e
nt
a
t
i
on
i
nde
x
,
w
h
i
c
h
i
s
us
e
d
t
o
e
v
a
l
u
a
t
e
t
he
pe
rfo
rm
a
nc
e
of
di
p
l
om
a
t
i
c
m
i
s
s
i
ons
i
n
di
ffe
r
e
n
t
a
r
e
a
s
bu
t
a
l
s
o
t
o
e
ns
ure
t
h
a
t
for
e
i
g
n
pol
i
c
y
de
c
i
s
i
ons
a
re
b
a
s
e
d
on
t
h
e
ri
g
ht
i
nfor
m
a
t
i
on
.
O
n
t
he
ot
he
r
ha
n
d,
m
i
s
c
l
a
s
s
i
fi
c
a
t
i
on
c
ou
l
d
r
e
s
ul
t
i
n
i
n
c
orr
e
c
t
e
va
l
u
a
t
i
ons
of
a
m
i
s
s
i
on’s
s
u
c
c
e
s
s
,
m
i
s
gu
i
d
e
d
po
l
i
c
y
de
c
i
s
i
ons
,
a
nd
e
v
e
n
pos
e
ri
s
ks
t
o
na
t
i
o
na
l
s
e
c
ur
i
t
y
.
M
ul
t
i
-
l
a
be
l
c
l
a
s
s
i
f
i
c
a
t
i
on
i
s
a
t
e
c
h
ni
qu
e
i
n
m
a
c
hi
n
e
l
e
a
rni
ng
t
ha
t
e
na
bl
e
s
a
s
s
i
gn
i
ng
m
u
l
t
i
pl
e
c
a
t
e
gor
i
e
s
t
o
a
s
i
ngl
e
i
ns
t
a
nc
e
[
5]
.
U
s
i
ng
t
h
i
s
s
upe
rvi
s
e
d
l
e
a
r
ni
ng
a
ppro
a
c
h
[6
]
,
di
p
l
om
a
t
i
c
do
c
um
e
nt
s
c
a
n
b
e
c
a
t
e
gor
i
z
e
d
i
nt
o
m
ul
t
i
pl
e
t
op
i
c
s
t
ha
t
r
e
pre
s
e
nt
t
h
e
v
a
ri
o
us
n
a
t
ure
s
of
d
i
pl
o
m
a
t
i
c
w
ork
.
T
h
i
s
re
s
e
a
r
c
h
e
xpl
ore
s
t
w
o
f
i
ne
-
t
un
i
ng
a
p
proa
c
h
e
s
i
n
t
he
c
ont
e
xt
of
m
ul
t
i
-
l
a
b
e
l
c
l
a
s
s
i
fi
c
a
t
i
on
of
di
pl
o
m
a
t
i
c
c
a
bl
e
s
:
f
ul
l
p
a
ra
m
e
t
e
r
fi
ne
-
t
un
i
ng
[7],
[8]
,
a
nd
t
he
m
or
e
e
ff
i
c
i
e
n
t
l
ow
-
r
a
nk
a
d
a
pt
a
t
i
on
(
L
oRA
)
[9]
m
e
t
hod.
W
he
n
a
ppl
i
e
d
t
o
t
he
BE
R
T
m
od
e
l
[10]
,
w
hi
c
h
i
s
kn
ow
n
for
i
t
s
robus
t
pe
r
form
a
nc
e
o
n
a
v
a
ri
e
t
y
of
na
t
ur
a
l
l
a
ng
ua
g
e
pro
c
e
s
s
i
n
g
(N
L
P
)
t
a
s
ks
,
bo
t
h
m
e
t
h
ods
h
a
v
e
s
how
n
pr
om
i
s
i
ng
re
s
u
l
t
s
.
F
i
ne
-
t
uni
ng
i
s
a
proc
e
s
s
of
a
dj
us
t
i
ng
a
pr
e
-
t
ra
i
ne
d
m
ode
l
’s
p
a
r
a
m
e
t
e
rs
t
o
i
m
pr
ove
i
t
s
p
e
rfor
m
a
n
c
e
on
a
s
pe
c
i
fi
c
t
a
s
k
,
w
hi
l
e
,
L
oRA
u
pda
t
e
s
onl
y
s
pe
c
i
f
i
c
pa
ra
m
e
t
e
rs
,
a
l
l
ow
i
ng
t
he
m
od
e
l
t
o
qu
i
c
k
l
y
a
da
p
t
t
o
s
pe
c
i
f
i
c
t
a
s
ks
w
i
t
hout
re
qu
i
ri
ng
e
xt
e
ns
i
v
e
c
o
m
p
u
t
a
t
i
on
a
l
re
s
ourc
e
s
w
h
i
l
e
a
c
h
i
e
vi
ng
c
om
p
e
t
i
t
i
ve
p
e
rfor
m
a
n
c
e
r
e
s
ul
t
s
c
om
pa
re
d
t
o
fu
l
l
pa
ra
m
e
t
e
r
fi
ne
-
t
uni
ng
m
e
t
hod
.
M
ul
t
i
-
l
a
be
l
t
e
xt
c
l
a
s
s
i
f
i
c
a
t
i
on
o
f
Indo
ne
s
i
a
n
c
us
t
o
m
e
r
r
e
vi
e
w
s
us
i
ng
In
doBE
RT
a
s
a
n
e
nd
-
to
-
e
nd
m
ode
l
i
m
pro
ve
d
w
i
t
h
a
n
a
c
c
u
ra
c
y
of
up
t
o
19
.
19
%
,
c
om
p
a
r
e
d
t
o
us
i
ng
Ind
oBE
RT
w
i
t
h
CN
N
a
nd
X
G
Boos
t
c
l
a
s
s
i
fi
e
r
[11]
.
In
N
a
bi
i
l
a
h
e
t
a
l
.
[
12]
,
m
u
l
t
i
-
l
a
be
l
c
l
a
s
s
i
fi
c
a
t
i
on
of
t
o
xi
c
c
o
m
m
e
n
t
s
i
n
In
don
e
s
i
a
n
s
oc
i
a
l
m
e
di
a
by
us
i
n
g
In
doB
E
RT
for
f
e
a
t
ure
e
xt
r
a
c
t
i
on
a
nd
M
B
E
R
T
f
or
c
l
a
s
s
i
fi
c
a
t
i
on
a
c
h
i
e
ve
d
op
t
i
m
a
l
r
e
s
ul
t
s
w
i
t
h
a
n
F
1
s
c
ore
of
0
.
903
2.
In
t
he
c
on
t
e
x
t
o
f
Ind
one
s
i
a
n
l
a
ngua
ge
proc
e
s
s
i
ng,
pr
e
-
t
r
a
i
n
e
d
m
od
e
l
s
s
u
c
h
a
s
C
a
hy
a
BE
RT
ha
ve
d
e
m
o
ns
t
ra
t
e
d
p
rom
i
s
i
ng
p
e
rfor
m
a
n
c
e
a
c
ros
s
va
r
i
ous
N
L
P
t
a
s
ks
,
s
uc
h
a
s
f
or
n
a
m
e
e
n
t
i
t
y
re
c
ogn
i
t
i
on
on
ha
di
t
h
t
e
x
t
s
a
c
h
i
e
v
e
d
i
m
pre
s
s
i
v
e
F
1
-
s
c
or
e
s
of
99
.
63%
i
n
t
e
s
t
i
ng
[13]
.
W
hi
l
e
foc
us
e
d
on
N
E
R,
t
h
i
s
s
u
c
c
e
s
s
ha
s
i
ns
p
i
re
d
t
h
e
a
pp
l
i
c
a
t
i
o
n
of
C
a
hy
a
BE
RT
f
or
m
u
l
t
i
-
l
a
b
e
l
c
l
a
s
s
i
fi
c
a
t
i
on
t
a
s
ks
.
L
o
RA
i
s
us
e
d
i
n
[14]
fo
r
c
l
a
s
s
i
fyi
n
g
l
e
g
a
l
do
c
u
m
e
n
t
s
.
T
he
r
e
s
ul
t
s
s
how
t
ha
t
L
oRA
pe
rfor
m
s
be
t
t
e
r
t
ha
n
ful
l
pa
r
a
m
e
t
e
r
fi
n
e
-
t
uni
n
g
w
hi
l
e
re
qui
r
i
ng
f
e
w
e
r
c
om
pu
t
a
t
i
o
na
l
re
s
o
urc
e
s
,
w
i
t
h
a
ra
nk
of
32
.
T
he
re
s
e
a
rc
he
r
de
v
e
l
ope
d
a
da
t
a
s
e
t
c
a
l
l
e
d
T
a
i
w
a
n
L
e
g
a
l
J
udg
e
m
e
n
t
P
r
e
d
i
c
t
i
on
(
T
W
L
J
P
)
a
n
d
c
o
m
pa
re
d
t
he
pe
rfo
rm
a
nc
e
of
L
oRA
a
g
a
i
ns
t
ful
l
pa
ra
m
e
t
e
r
fi
n
e
-
t
uni
n
g
m
e
t
h
ods
a
n
d
ot
h
e
r
m
od
e
l
s
l
i
k
e
L
a
w
form
e
r
.
R
e
s
ul
t
s
s
how
t
h
a
t
L
oRA
a
c
h
i
e
v
e
d
c
om
p
a
r
a
bl
e
p
e
rfor
m
a
n
c
e
t
o
ful
l
fi
ne
-
t
uni
ng
w
h
i
l
e
s
i
gn
i
fi
c
a
n
t
l
y
re
duc
i
ng
c
o
m
put
a
t
i
on
a
l
r
e
s
our
c
e
s
a
nd
t
ra
i
ni
n
g
t
i
m
e
,
r
e
qu
i
ri
ng
on
l
y
a
bo
ut
0.
7
2%
of
t
h
e
t
ra
i
na
b
l
e
p
a
ra
m
e
t
e
rs
c
om
p
a
r
e
d
t
o
t
h
e
f
ul
l
m
ode
l
.
T
hi
s
a
r
t
i
c
l
e
w
i
l
l
c
o
m
p
a
re
c
a
hy
a
/
b
e
r
t
-
ba
s
e
/
i
n
done
s
i
a
n
-
1.
5
G
,
i
ndol
e
m
/
i
ndob
e
rt
-
b
a
s
e
-
u
nc
a
s
e
d
,
a
nd
be
rt
-
ba
s
e
-
m
u
l
t
i
l
i
ngua
l
-
c
a
s
e
d
us
i
ng
fu
l
l
pa
r
a
m
e
t
e
r
fi
ne
-
t
uni
n
g
a
nd
L
oRA
t
o
c
l
a
s
s
i
fy
t
he
di
p
l
om
a
t
i
c
c
a
bl
e
s
,
i
nt
o
m
ul
t
i
-
l
a
b
e
l
s
c
a
t
e
gor
i
e
s
,
w
h
i
c
h
i
s
p
ol
i
t
i
c
s
,
e
c
ono
m
i
c
s
,
pro
t
oc
ol
a
n
d
c
ons
u
l
a
r
s
e
r
vi
c
e
s
,
s
o
c
i
o
-
c
ul
t
ur
a
l
a
ffa
i
rs
,
a
nd
a
d
m
i
ni
s
t
r
a
t
i
ve
m
a
t
t
e
rs
.
T
he
d
a
t
a
s
e
t
us
e
d
i
s
di
p
l
o
m
a
t
i
c
c
a
b
l
e
s
s
a
m
p
l
e
d
fro
m
202
2
-
2023
,
a
n
d
onl
y
t
he
e
x
e
c
ut
i
ve
s
um
m
a
ri
e
s
s
e
c
t
i
ons
a
r
e
t
a
ke
n
.
T
h
i
s
s
t
udy
w
i
l
l
be
c
ondu
c
t
e
d
us
i
ng
a
s
i
ngl
e
N
V
ID
IA
G
e
F
o
rc
e
G
T
X
108
0
8G
B
G
P
U
t
o
e
n
a
bl
e
c
o
m
pre
he
ns
i
ve
e
v
a
l
u
a
t
i
on
t
r
a
d
e
-
offs
be
t
w
e
e
n
m
od
e
l
pe
rfor
m
a
n
c
e
a
n
d
re
s
ourc
e
u
t
i
l
i
z
a
t
i
on
,
w
hi
c
h
i
s
c
ru
c
i
a
l
for
d
i
pl
o
m
a
t
i
c
i
ns
t
i
t
ut
i
ons
t
h
a
t
ha
v
e
l
i
m
i
t
e
d
c
om
put
a
t
i
ona
l
r
e
s
ourc
e
s
.
A
c
c
ur
a
t
e
l
y
de
t
e
r
m
i
n
i
ng
a
nd
a
s
s
e
s
s
i
ng
t
he
c
a
t
e
g
ori
e
s
of
d
i
pl
om
a
t
i
c
c
a
bl
e
s
i
s
e
s
s
e
n
t
i
a
l
s
o
t
ha
t
d
i
pl
o
m
a
t
i
c
a
c
t
i
vi
t
i
e
s
c
a
n
a
l
i
gn
w
i
t
h
t
h
e
na
t
i
on
a
l
for
e
i
g
n
po
l
i
c
y
a
nd
c
ont
r
i
but
e
t
o
i
nt
e
rn
a
t
i
on
a
l
re
l
a
t
i
ons
s
t
ra
t
e
gy.
2.
M
ET
H
O
D
F
i
gure
1
s
h
ow
s
t
he
s
t
e
ps
for
t
hi
s
re
s
e
a
rc
h
i
n
ge
ne
ra
l
.
In
da
t
a
c
ol
l
e
c
t
i
on,
t
he
e
xe
c
ut
i
ve
s
um
m
a
ri
e
s
a
re
e
xt
ra
c
t
e
d
from
t
he
doc
um
e
nt
s
a
m
pl
e
s
a
nd
a
nnot
a
t
e
d.
T
he
ne
xt
s
t
e
p
i
s
da
t
a
pre
pa
ra
t
i
on,
w
hi
c
h
i
nc
l
ude
s
pre
-
pr
oc
e
s
s
i
ng,
da
t
a
s
pl
i
t
t
i
n
g,
a
nd
e
xpl
ora
t
or
y
da
t
a
a
na
l
y
s
i
s
(
E
D
A
)
.
T
w
o
di
ffe
re
nt
m
ode
l
i
ng
t
e
c
h
ni
que
s
a
re
a
ppl
i
e
d
t
o
t
he
m
ode
l
,
ful
l
-
pa
ra
m
e
t
e
r
fi
ne
-
t
uni
ng
a
nd
fi
ne
-
t
uni
ng
w
i
t
h
L
oR
A
.
T
he
fi
na
l
s
t
e
p
i
s
e
va
l
ua
t
i
ng
bot
h
m
ode
l
s
.
F
i
gure
1
.
S
t
e
ps
i
n
t
h
e
propos
e
d
m
e
t
hod
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
,
V
o
l
.
6
,
N
o
.
3
,
N
ov
e
m
be
r
20
25
:
274
-
282
276
2.
1
.
D
ata
c
o
l
l
e
c
t
i
on
T
he
da
t
a
s
e
t
ut
i
l
i
z
e
d
for
t
h
i
s
s
t
udy
w
a
s
obt
a
i
n
e
d
fr
om
t
he
M
i
n
i
s
t
ry
of
F
o
re
i
gn
A
ff
a
i
rs
of
t
h
e
Re
publ
i
c
of
Indon
e
s
i
a
,
c
o
m
pri
s
i
ng
1
,
329
di
pl
o
m
a
t
i
c
c
a
bl
e
s
c
ol
l
e
c
t
e
d
from
1
30
M
i
s
s
i
ons
.
T
h
e
e
xe
c
ut
i
ve
s
um
m
a
r
i
e
s
o
f
a
l
l
do
c
um
e
n
t
s
w
e
r
e
e
x
t
ra
c
t
e
d
a
nd
c
a
t
e
gor
i
z
e
d
us
i
ng
a
m
u
l
t
i
-
l
a
b
e
l
i
ng
s
ys
t
e
m
.
T
a
bl
e
1
d
i
s
pl
a
ys
s
a
m
p
l
e
s
of
l
a
b
e
l
e
d
da
t
a
.
E
a
c
h
s
a
m
pl
e
m
a
y
be
a
s
s
oc
i
a
t
e
d
w
i
t
h
m
ul
t
i
pl
e
c
a
t
e
go
ri
e
s
t
ha
t
r
e
fl
e
c
t
t
h
e
o
ve
r
l
a
p
pi
ng
t
he
m
e
s
pre
s
e
n
t
i
n
t
h
e
e
xe
c
ut
i
v
e
s
u
m
m
a
ri
e
s
.
T
a
b
l
e
1
.
E
xa
m
p
l
e
of
da
t
a
l
a
be
l
i
ng
No
S
u
m
m
a
ry
L
a
b
e
l
B
-
0
0
3
2
7
/
F
ra
n
k
f
u
rt
/
2
3
1
1
2
4
T
h
e
In
d
o
n
e
s
i
a
n
Co
n
s
u
l
a
t
e
G
e
n
e
ra
l
i
n
F
ra
n
k
fu
rt
h
e
l
d
a
v
i
rt
u
a
l
d
i
s
c
u
s
s
i
o
n
w
i
t
h
BN
I
L
o
n
d
o
n
e
n
t
i
t
l
e
d
“
D
e
v
e
l
o
p
i
n
g
In
d
o
n
e
s
i
a
n
D
i
a
s
p
o
ra
E
n
t
re
p
re
n
e
u
rs
h
i
p
i
n
t
h
e
Co
n
s
u
l
a
t
e
G
e
n
e
ra
l
's
A
re
a
o
f
Re
s
p
o
n
s
i
b
i
l
i
t
y
”
o
n
N
o
v
e
m
b
e
r
2
3
,
2
0
2
3
.
T
h
e
e
v
e
n
t
w
a
s
o
p
e
n
e
d
b
y
t
h
e
In
d
o
n
e
s
i
a
n
Co
n
s
u
l
G
e
n
e
ra
l
i
n
F
ra
n
k
fu
rt
a
n
d
fe
a
t
u
re
d
s
p
e
a
k
e
rs
fro
m
BN
I
L
o
n
d
o
n
a
n
d
BN
I
H
e
a
d
q
u
a
rt
e
rs
.
....
e
c
o
n
o
m
i
c
,
s
o
c
i
o
-
c
u
l
t
u
ra
l
B
-
0
0
2
6
8
/
Bra
t
i
s
l
a
v
a
/
2
3
1
2
2
7
T
h
e
In
d
o
n
e
s
i
a
n
E
m
b
a
s
s
y
i
n
Bra
t
i
s
l
a
v
a
h
e
l
d
a
c
o
m
m
u
n
i
t
y
o
u
t
re
a
c
h
e
v
e
n
t
i
n
t
h
e
fo
rm
o
f
a
Ch
ri
s
t
m
a
s
s
e
rv
i
c
e
a
n
d
c
e
l
e
b
ra
t
i
o
n
o
n
D
e
c
e
m
b
e
r
1
9
,
2
0
2
3
.
T
h
e
e
v
e
n
t
w
a
s
o
rg
a
n
i
z
e
d
i
n
c
o
l
l
a
b
o
ra
t
i
o
n
w
i
t
h
t
h
e
In
d
o
n
e
s
i
a
n
Ch
ri
s
t
i
a
n
F
a
m
i
l
y
i
n
S
l
o
v
a
k
i
a
(G
a
ri
s
i
n
d
o
).
T
h
e
e
v
e
n
t
w
a
s
a
t
t
e
n
d
e
d
b
y
a
ro
u
n
d
1
0
0
In
d
o
n
e
s
i
a
n
s
a
n
d
m
e
m
b
e
rs
o
f
t
h
e
In
d
o
n
e
s
i
a
n
d
i
a
s
p
o
ra
i
n
....
s
o
c
i
o
-
c
u
l
t
u
ra
l
T
he
d
a
t
a
s
e
t
e
n
c
om
pa
s
s
e
s
fi
ve
c
a
t
e
gor
i
e
s
of
d
i
pl
om
a
t
i
c
fun
c
t
i
ons
:
pol
i
t
i
c
a
l
,
e
c
ono
m
i
c
,
s
o
c
i
o
-
c
ul
t
ur
a
l
,
prot
o
c
ol
a
nd
c
ons
u
l
a
r
,
a
nd
a
d
m
i
n
i
s
t
r
a
t
i
on.
E
a
c
h
of
t
he
s
e
fu
nc
t
i
ons
i
s
s
p
e
c
i
fi
c
a
l
l
y
de
t
a
i
l
e
d
i
n
t
he
r
e
gu
l
a
t
i
on
[15]
.
F
i
gur
e
2
i
l
l
us
t
ra
t
e
s
t
he
di
s
t
ri
b
ut
i
on
of
do
c
um
e
n
t
s
ba
s
e
d
on
how
m
a
ny
l
a
b
e
l
s
e
a
c
h
do
c
um
e
nt
h
a
s
,
r
a
ng
i
ng
from
on
e
t
o
fi
v
e
l
a
b
e
l
s
pe
r
d
oc
u
m
e
nt
.
T
he
d
a
t
a
r
e
v
e
a
l
s
t
ha
t
s
i
ng
l
e
-
l
a
b
e
l
do
c
u
m
e
n
t
s
a
re
t
he
m
os
t
p
re
v
a
l
e
nt
,
t
ot
a
l
i
n
g
842
doc
um
e
nt
s
.
D
o
c
u
m
e
n
t
s
w
i
t
h
du
a
l
l
a
be
l
s
f
ol
l
ow
,
a
c
c
oun
t
i
ng
fo
r
3
23
doc
u
m
e
nt
s
.
T
he
nu
m
b
e
r
of
doc
u
m
e
n
t
s
w
i
t
h
t
h
re
e
l
a
b
e
l
s
a
m
oun
t
e
d
t
o
61
doc
u
m
e
nt
s
,
w
i
t
h
four
l
a
b
e
l
s
re
a
c
h
i
ng
78
,
a
nd
t
he
l
e
a
s
t
c
om
m
on
a
re
do
c
um
e
nt
s
w
i
t
h
f
i
v
e
l
a
b
e
l
s
,
on
l
y
25
doc
um
e
nt
s
.
F
i
gure
2
.
D
i
s
t
ri
b
ut
i
on
of
t
he
nu
m
b
e
r
o
f
l
a
b
e
l
s
i
n
t
h
e
da
t
a
s
e
t
2.
2
.
D
ata
p
r
e
p
a
r
at
i
on
In
ord
e
r
t
o
pr
e
s
e
rv
e
i
t
s
c
ont
e
xt
[16
]
,
t
e
xt
pr
e
-
pr
oc
e
s
s
i
ng
i
s
m
i
n
i
m
i
z
e
d,
w
i
t
h
onl
y
r
e
m
ova
l
of
U
RL
s
from
t
h
e
e
xe
c
ut
i
ve
s
um
m
a
r
y.
D
a
t
a
s
p
l
i
t
t
i
ng
i
s
a
pro
c
e
s
s
of
di
vi
d
i
ng
a
da
t
a
s
e
t
i
nt
o
s
e
p
a
ra
t
e
s
u
bs
e
t
s
t
o
re
d
uc
e
bi
a
s
,
pre
v
e
nt
o
ve
rf
i
t
t
i
ng
,
a
nd
a
c
c
ur
a
t
e
l
y
e
v
a
l
u
a
t
e
t
h
e
m
o
de
l
p
e
rfor
m
a
n
c
e
on
uns
e
e
n
da
t
a
[
17]
.
F
or
t
hi
s
re
s
e
a
rc
h
,
a
s
t
r
a
t
i
f
i
e
d
s
a
m
p
l
i
ng
m
e
t
hod
i
s
a
p
pl
i
e
d
,
w
h
e
r
e
t
he
popu
l
a
t
i
on
of
t
he
d
a
t
a
i
s
d
i
vi
d
e
d
i
n
t
o
di
ff
e
r
e
nt
s
ubgroups
a
n
d
e
ns
ur
e
s
t
h
a
t
t
h
e
fi
n
a
l
s
a
m
p
l
e
s
r
e
fl
e
c
t
t
h
e
pro
p
ort
i
ons
of
t
h
e
s
e
s
ubg
roups
[18
]
.
T
hi
s
a
ppro
a
c
h
i
s
c
ruc
i
a
l
for
m
ul
t
i
-
l
a
b
e
l
d
a
t
a
s
e
t
s
,
a
s
i
t
h
e
l
ps
m
a
i
nt
a
i
n
t
he
b
a
l
a
n
c
e
of
l
a
b
e
l
c
om
b
i
n
a
t
i
ons
a
c
ros
s
a
l
l
s
ubs
e
t
s
,
e
ns
uri
n
g
t
h
a
t
e
a
c
h
s
ubs
e
t
i
s
re
pr
e
s
e
n
t
a
t
i
v
e
of
t
he
ov
e
r
a
l
l
d
a
t
a
di
s
t
r
i
bu
t
i
on
.
By
a
ppl
yi
ng
t
hi
s
a
ppro
a
c
h
,
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
IS
S
N
:
2722
-
3221
O
pt
i
m
i
z
i
ng
di
p
l
om
at
i
c
i
n
de
x
i
ng:
f
ul
l
-
par
am
e
t
e
r
v
s
l
ow
-
r
a
nk
adapt
a
t
i
on
f
or
m
u
l
t
i
-
l
ab
e
l
…
(
D
e
l
a
Nur
l
ai
l
a
)
277
da
t
a
s
e
t
i
s
s
p
l
i
t
i
n
t
o
t
hr
e
e
s
ubs
e
t
s
:
8
0%
f
or
t
he
t
r
a
i
n
i
ng
s
e
t
a
nd
t
h
e
r
e
m
a
i
ni
ng
20%
for
t
h
e
t
e
s
t
i
ng
s
e
t
,
w
h
i
l
s
t
one
-
t
e
n
t
h
of
t
he
t
ra
i
ni
ng
s
e
t
i
s
ra
nd
om
l
y
s
e
l
e
c
t
e
d
for
v
a
l
i
da
t
i
on
da
t
a
[19
]
.
H
ow
e
ve
r
,
du
e
t
o
t
h
e
c
ons
t
ra
i
nt
s
of
m
a
i
nt
a
i
n
i
ng
l
a
b
e
l
ba
l
a
n
c
e
a
c
ros
s
s
ubs
e
t
s
i
n
m
u
l
t
i
-
l
a
be
l
d
a
t
a
s
e
t
s
l
e
d
t
o
a
m
o
di
f
i
e
d
di
s
t
ri
bu
t
i
on
,
a
s
s
how
n
i
n
T
a
b
l
e
2
.
F
ur
t
he
r
m
or
e
,
a
dd
i
t
i
on
a
l
E
D
A
w
a
s
p
e
rfor
m
e
d
t
o
a
n
a
l
y
z
e
t
he
t
r
a
i
n
i
ng
d
a
t
a
,
i
n
ord
e
r
t
o
g
e
t
m
or
e
i
nf
o
a
bout
t
he
da
t
a
p
a
t
t
e
rn
.
T
a
b
l
e
2
.
D
a
t
a
s
e
t
s
t
a
t
i
s
t
i
c
M
u
l
t
i
-
l
a
b
e
l
D
a
t
a
s
e
t
D
a
t
a
t
ra
i
n
D
a
t
a
t
e
s
t
D
a
t
a
v
a
l
i
d
a
t
i
o
n
1
l
a
b
e
l
(s
i
n
g
l
e
)
842
605
169
68
2
l
a
b
e
l
s
323
232
65
26
3
l
a
b
e
l
s
61
44
12
5
4
l
a
b
e
l
s
78
56
16
6
5
l
a
b
e
l
s
25
18
5
2
2.
3
.
M
od
e
l
s
e
tt
i
n
gs
In
t
hi
s
r
e
s
e
a
rc
h,
t
hre
e
d
i
ffe
re
n
t
N
L
P
m
od
e
l
s
w
e
r
e
s
e
l
e
c
t
e
d:
t
w
o
B
E
R
T
-
b
a
s
e
m
ode
l
s
t
r
a
i
n
e
d
i
n
m
ono
l
i
ngu
a
l
,
na
m
e
l
y
Ind
oBE
RT
[20
]
a
nd
C
a
hy
a
BE
RT
[13]
,
a
s
w
e
l
l
a
s
on
e
m
ul
t
i
l
i
ngu
a
l
m
ode
l
,
M
BE
RT
[2
1]
.
T
he
s
e
m
od
e
l
s
s
ha
r
e
t
h
e
s
a
m
e
c
onf
i
gur
a
t
i
on
f
or
e
m
b
e
dd
i
ng
s
i
z
e
,
nu
m
b
e
r
of
hi
d
de
n
l
a
y
e
rs
,
a
n
d
a
t
t
e
nt
i
on
h
e
a
ds
w
h
i
c
h
a
r
e
768
,
12
,
a
nd
1
2
,
r
e
s
pe
c
t
i
ve
l
y.
T
a
bl
e
3
pr
ovi
d
e
s
a
s
u
m
m
a
ry
of
t
he
hype
r
pa
r
a
m
e
t
e
r
c
onfi
gura
t
i
o
ns
us
e
d
i
n
pri
or
r
e
s
e
a
rc
h
t
o
d
e
ve
l
op
t
h
e
pre
-
t
ra
i
n
e
d
B
E
R
T
t
ha
t
a
r
e
u
t
i
l
i
z
e
d
i
n
t
hi
s
re
s
e
a
r
c
h.
T
a
b
l
e
3
.
H
yp
e
rp
a
ra
m
e
t
e
r
s
u
m
m
a
ry
of
BE
R
T
m
o
de
l
s
M
o
d
e
l
T
y
p
e
P
a
ra
m
e
t
e
rs
Co
rp
u
s
s
i
z
e
V
o
c
a
b
s
i
z
e
c
a
h
y
a
/
b
e
rt
-
b
a
s
e
/
i
n
d
o
n
e
s
i
a
n
-
1
.
5
G
M
o
n
o
l
i
n
g
u
a
l
(In
d
o
n
e
s
i
a
n
)
110
M
1
.
5
G
32
,
000
i
n
d
o
l
e
m
/
i
n
d
o
b
e
rt
-
b
a
s
e
-
u
n
c
a
s
e
d
M
o
n
o
l
i
n
g
u
a
l
(In
d
o
n
e
s
i
a
n
)
110
M
220
M
31
,
923
b
e
rt
-
b
a
s
e
-
m
u
l
t
i
l
i
n
g
u
a
l
-
c
a
s
e
d
M
u
l
t
i
l
i
n
g
u
a
l
172
M
W
i
k
i
p
e
d
i
a
(1
0
4
l
a
n
g
u
a
g
e
s
)
119
,
5
4
7
T
w
o
prom
i
n
e
nt
a
ppro
a
c
h
e
s
w
e
re
us
e
d
for
t
h
i
s
s
t
udy
,
w
h
i
c
h
a
re
fu
l
l
-
p
a
ra
m
e
t
e
r
f
i
n
e
-
t
un
i
ng
a
nd
pa
ra
m
e
t
e
r
-
e
ff
i
c
i
e
n
t
t
un
i
ng
,
w
i
t
h
a
pa
r
t
i
c
ul
a
r
fo
c
us
o
n
L
oRA
.
F
ul
l
-
pa
ra
m
e
t
e
r
f
i
ne
-
t
un
i
ng
i
nv
ol
v
e
s
u
pda
t
i
n
g
a
l
l
pa
ra
m
e
t
e
rs
of
a
pre
-
t
ra
i
ne
d
m
od
e
l
duri
n
g
t
he
t
r
a
i
n
i
ng
proc
e
s
s
,
a
l
l
ow
i
ng
t
he
m
ode
l
t
o
a
da
p
t
f
ul
l
y
t
o
t
he
t
a
s
k
[22],
[23]
.
O
n
t
h
e
ot
h
e
r
h
a
nd
,
L
oRA
a
i
m
s
t
o
a
d
a
pt
m
o
de
l
s
w
i
t
h
m
i
ni
m
a
l
c
ha
n
ge
s
t
o
t
he
or
i
gi
n
a
l
pa
ra
m
e
t
e
rs
,
w
hi
c
h
r
e
s
ul
t
s
i
n
hi
gh
p
a
r
a
m
e
t
e
r
e
ff
i
c
i
e
n
c
y
[24
]
.
I
t
i
nt
rodu
c
e
s
t
ra
i
na
b
l
e
r
a
nk
d
e
c
o
m
pos
i
t
i
on
m
a
t
r
i
c
e
s
t
o
t
h
e
w
e
i
gh
t
s
,
a
l
l
ow
i
ng
a
d
a
pt
a
t
i
ons
w
h
i
l
e
k
e
e
p
i
ng
m
os
t
of
t
he
ori
gi
n
a
l
m
o
de
l
fro
z
e
n
[9]
.
T
a
b
l
e
4
out
l
i
n
e
s
t
he
hype
rp
a
ra
m
e
t
e
rs
us
e
d
dur
i
ng
t
h
e
fi
n
e
-
t
uni
n
g
p
roc
e
s
s
,
a
n
d
T
a
bl
e
5
de
t
a
i
l
i
n
g
t
h
e
hype
rpa
r
a
m
e
t
e
rs
s
pe
c
i
f
i
c
t
o
L
oRA
-
b
a
s
e
d
t
un
i
ng
.
A
l
l
e
x
pe
r
i
m
e
nt
s
w
e
r
e
c
ond
uc
t
e
d
us
i
ng
a
n
N
V
ID
IA
G
e
F
orc
e
G
T
X
1080
8G
B
G
P
U
.
T
a
b
l
e
4
.
H
yp
e
rp
a
ra
m
e
t
e
r
s
e
t
t
i
ngs
f
or
f
ul
l
-
pa
r
a
m
e
t
e
r
fi
n
e
-
t
uni
ng
H
y
p
e
rp
a
ra
m
e
t
e
r
V
a
l
u
e
L
e
a
rn
i
n
g
ra
t
e
1e
-
5
E
p
o
c
h
30
Ba
t
c
h
s
i
z
e
16
W
a
rm
u
p
ra
t
i
o
0
.
1
L
R
s
c
h
e
d
u
l
e
r
t
y
p
e
L
i
n
e
a
r
D
ro
p
o
u
t
ra
t
e
0
.
3
W
e
i
g
h
t
d
e
c
a
y
0
.
0
1
P
a
t
i
e
n
c
e
3
T
a
b
l
e
5
.
H
yp
e
rp
a
ra
m
e
t
e
r
s
e
t
t
i
ngs
f
or
L
oRA
H
y
p
e
rp
a
ra
m
e
t
e
r
V
a
l
u
e
L
e
a
rn
i
n
g
ra
t
e
3e
-
5
E
p
o
c
h
30
Ba
t
c
h
s
i
z
e
4
W
a
rm
u
p
ra
t
i
o
0
.
1
L
R
s
c
h
e
d
u
l
e
r
t
y
p
e
Co
s
i
n
e
D
ro
p
o
u
t
ra
t
e
0
.
3
W
e
i
g
h
t
d
e
c
a
y
0
.
0
1
P
a
t
i
e
n
c
e
3
Ra
n
k
16
A
l
p
h
a
32
T
a
rg
e
t
m
o
d
u
l
e
s
Q,
K,
V
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
,
V
o
l
.
6
,
N
o
.
3
,
N
ov
e
m
be
r
20
25
:
274
-
282
278
2.
4
.
Eva
l
u
ati
on
m
e
th
od
In
t
h
i
s
s
t
udy
,
t
he
e
v
a
l
u
a
t
i
on
m
e
t
ri
c
s
us
e
d
t
o
m
e
a
s
ur
e
t
he
pe
r
form
a
n
c
e
of
t
he
m
u
l
t
i
-
l
a
be
l
c
l
a
s
s
i
fi
c
a
t
i
on
t
a
s
k
a
re
h
a
m
m
i
n
g
l
os
s
a
n
d
s
ubs
e
t
a
c
c
ur
a
c
y
[6
]
.
H
a
m
m
i
n
g
l
os
s
e
v
a
l
ua
t
e
s
w
h
e
t
he
r
e
a
c
h
l
a
b
e
l
of
e
a
c
h
s
a
m
pl
e
i
s
p
re
d
i
c
t
e
d
c
orr
e
c
t
l
y
.
A
l
ow
e
r
h
a
m
m
i
ng
l
os
s
v
a
l
u
e
i
nd
i
c
a
t
e
s
t
ha
t
t
h
e
m
od
e
l
i
s
m
ore
a
c
c
ur
a
t
e
i
n
pre
di
c
t
i
ng
e
a
c
h
l
a
b
e
l
.
S
ubs
e
t
a
c
c
ur
a
c
y
e
v
a
l
u
a
t
e
s
w
he
t
he
r
t
he
e
nt
i
re
s
e
t
of
l
a
b
e
l
s
f
or
a
gi
v
e
n
s
a
m
p
l
e
i
s
pre
di
c
t
e
d
e
x
a
c
t
l
y
c
orre
c
t
l
y.
T
he
hi
ghe
r
t
he
s
ubs
e
t
a
c
c
ura
c
y
v
a
l
u
e
,
t
h
e
be
t
t
e
r
t
he
p
e
rfor
m
a
n
c
e
of
t
h
e
m
o
de
l
.
3.
R
ES
U
LTS
A
N
D
D
I
S
C
U
S
S
I
O
N
T
o
de
t
e
r
m
i
ne
t
he
o
pt
i
m
a
l
t
oke
n
l
e
n
gt
h
,
a
t
ok
e
ni
z
e
r
f
rom
e
a
c
h
pr
e
-
t
ra
i
ne
d
m
od
e
l
i
s
us
e
d
t
o
c
a
l
c
ul
a
t
e
t
he
di
s
t
r
i
bu
t
i
on
of
t
ok
e
n
l
e
ng
t
hs
i
n
t
h
e
d
a
t
a
s
e
t
.
T
a
bl
e
6
s
how
s
t
he
di
s
t
r
i
bu
t
i
on
of
t
ok
e
n
c
oun
t
s
for
e
a
c
h
m
od
e
l
,
w
he
re
t
h
e
m
a
j
ori
t
y
o
f
s
a
m
pl
e
s
h
a
ve
t
ok
e
n
l
e
ngt
hs
und
e
r
51
2,
e
xc
e
pt
for
M
BE
R
T
,
w
h
e
re
onl
y
0
.
6%
of
t
he
da
t
a
a
re
out
l
i
e
rs
.
T
h
e
re
f
ore
,
5
12
i
s
us
e
d
a
s
t
he
m
a
x
i
m
u
m
t
ok
e
n
l
e
ng
t
h
.
By
m
a
xi
m
i
z
i
n
g
,
t
ok
e
n
l
e
n
gt
hs
he
l
ps
t
o
c
a
p
t
ure
t
he
e
s
s
e
n
t
i
a
l
c
ont
e
xt
of
t
h
e
da
t
a
,
e
na
bl
i
ng
a
ri
c
he
r
un
de
r
s
t
a
ndi
n
g
o
f
t
he
t
e
xt
.
T
a
b
l
e
6
.
D
i
s
t
ri
b
ut
i
on
o
f
t
oke
n
c
oun
t
s
M
o
d
e
l
M
i
n
T
o
k
e
n
M
a
x
T
o
k
e
n
%
o
f
e
n
t
ri
e
s
<
=
5
1
2
c
a
h
y
a
/
b
e
rt
-
b
a
s
e
/
i
n
d
o
n
e
s
i
a
n
-
1
.
5
G
13
437
100
i
n
d
o
l
e
m
/
i
n
d
o
b
e
rt
-
b
a
s
e
-
u
n
c
a
s
e
d
14
442
100
b
e
rt
-
b
a
s
e
-
m
u
l
t
i
l
i
n
g
u
a
l
-
c
a
s
e
d
16
568
9
9
.
4
A
ddi
t
i
ona
l
E
D
A
w
a
s
p
e
rfor
m
e
d
t
o
a
na
l
yz
e
t
h
e
t
r
a
i
n
i
ng
da
t
a
,
r
e
ve
a
l
i
ng
t
h
a
t
i
t
r
e
m
a
i
ns
i
m
b
a
l
a
nc
e
d
.
T
o
a
ddr
e
s
s
t
hi
s
une
ve
n
d
i
s
t
ri
but
i
on,
t
w
o
c
o
m
pl
e
m
e
nt
a
ry
a
pproa
c
he
s
w
e
r
e
a
pp
l
i
e
d:
d
a
t
a
a
ug
m
e
n
t
a
t
i
o
n
a
nd
ra
ndo
m
un
de
rs
a
m
p
l
i
ng
(RU
S
)
.
D
a
t
a
a
u
gm
e
nt
a
t
i
on
i
s
a
t
e
c
h
ni
que
t
h
a
t
i
n
c
re
a
s
e
s
t
he
a
m
oun
t
a
nd
di
v
e
rs
i
t
y
of
da
t
a
by
m
od
i
fy
i
ng
e
x
i
s
t
i
ng
e
x
a
m
p
l
e
s
or
c
re
a
t
i
ng
a
n
e
w
o
ne
[25]
.
O
n
e
c
o
m
m
on
d
a
t
a
a
ug
m
e
nt
a
t
i
on
m
e
t
h
od
i
s
ba
c
k
t
r
a
ns
l
a
t
i
on
[26
]
–
[28]
.
In
t
h
i
s
s
t
udy
,
b
a
c
k
t
r
a
ns
l
a
t
i
on
i
s
us
e
d
w
he
r
e
t
he
ori
g
i
na
l
Ind
one
s
i
a
n
t
e
xt
i
s
t
ra
ns
l
a
t
e
d
t
o
E
ngl
i
s
h,
a
nd
t
he
n
c
o
nve
r
t
e
d
ba
c
k
t
o
Indo
ne
s
i
a
n
.
T
h
i
s
pro
c
e
s
s
i
s
c
a
r
ri
e
d
ou
t
us
i
ng
t
h
e
T
r
a
ns
l
a
t
o
r
c
l
a
s
s
fro
m
t
h
e
goo
gl
e
t
r
a
ns
l
i
bra
ry
.
A
ra
t
i
o
t
hr
e
s
hol
d
i
s
i
n
t
rodu
c
e
d
t
o
e
ns
ur
e
t
h
e
a
ugm
e
nt
a
t
i
on
do
e
s
not
di
s
propor
t
i
o
na
t
e
l
y
i
nc
r
e
a
s
e
s
y
nt
h
e
t
i
c
da
t
a
,
he
l
pi
ng
t
o
m
a
i
n
t
a
i
n
a
ba
l
a
n
c
e
d
d
a
t
a
s
e
t
.
T
o
fu
rt
h
e
r
ba
l
a
n
c
e
t
he
da
t
a
,
RU
S
w
e
re
i
m
pl
e
m
e
n
t
e
d
.
RU
S
i
s
a
m
e
t
hod
for
h
a
ndl
i
ng
i
m
b
a
l
a
nc
e
d
da
t
a
by
ra
nd
om
l
y
r
e
m
ovi
n
g
s
a
m
p
l
e
s
from
t
h
e
m
a
j
or
i
t
y
c
l
a
s
s
[
29]
–
[3
1]
.
A
s
s
how
n
i
n
F
i
gur
e
3
,
t
he
pr
oc
e
s
s
of
b
a
c
k
t
ra
ns
l
a
t
i
on
a
nd
RU
S
h
a
s
s
i
gni
fi
c
a
n
t
l
y
a
l
t
e
r
e
d
t
h
e
di
s
t
ri
but
i
on
of
l
a
b
e
l
c
om
b
i
na
t
i
o
ns
i
n
t
h
e
t
r
a
i
n
i
ng
d
a
t
a
.
T
he
m
os
t
no
t
a
b
l
e
c
ha
nge
s
a
r
e
obs
e
rve
d
i
n
s
i
ng
l
e
a
nd
doubl
e
l
a
be
l
c
o
m
bi
n
a
t
i
ons
.
T
h
e
ori
g
i
na
l
da
t
a
of
9
55
s
a
m
pl
e
s
ha
s
b
e
e
n
e
xp
a
nd
e
d
t
o
1
,
655
s
a
m
p
l
e
s
,
a
n
i
nc
r
e
a
s
e
of
73
.
3%
.
T
h
e
l
e
a
s
t
fre
que
n
t
da
t
a
c
a
t
e
gory
c
o
nt
a
i
n
i
n
g
5
l
a
be
l
s
a
l
s
o
doub
l
e
d
i
n
qu
a
nt
i
t
y
.
F
i
gure
3
.
D
a
t
a
t
ra
i
ni
n
g
w
i
t
h
m
u
l
t
i
pl
e
l
a
b
e
l
s
:
or
i
g
i
na
l
vs
ba
l
a
n
c
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
IS
S
N
:
2722
-
3221
O
pt
i
m
i
z
i
ng
di
p
l
om
at
i
c
i
n
de
x
i
ng:
f
ul
l
-
par
am
e
t
e
r
v
s
l
ow
-
r
a
nk
adapt
a
t
i
on
f
or
m
u
l
t
i
-
l
ab
e
l
…
(
D
e
l
a
Nur
l
ai
l
a
)
279
F
i
gure
4
s
how
s
t
he
h
a
m
m
i
ng
l
os
s
m
e
t
r
i
c
s
a
c
r
os
s
va
r
i
ous
e
p
oc
hs
.
In
t
he
f
ul
l
-
pa
r
a
m
e
t
e
r
f
i
n
e
-
t
un
i
ng
,
a
s
s
how
n
i
n
F
i
g
ure
4(
a
),
In
doBE
RT
a
c
hi
e
v
e
s
t
he
l
ow
e
s
t
ha
m
m
i
ng
l
os
s
va
l
ue
c
om
p
a
re
d
t
o
Ca
h
ya
B
E
R
T
a
nd
M
BE
RT
.
W
hi
l
e
i
n
F
i
gure
4(b)
,
t
h
e
h
a
m
m
i
ng
l
os
s
v
a
l
u
e
fo
r
L
oRA
-
b
a
s
e
d
t
un
i
ng
d
e
m
ons
t
r
a
t
e
s
t
h
a
t
L
oRA
-
IndoBE
RT
c
o
ns
i
s
t
e
nt
l
y
m
a
i
nt
a
i
ns
t
he
l
ow
e
s
t
l
os
s
.
F
i
gur
e
5
c
o
m
p
a
re
s
s
ubs
e
t
a
c
c
ura
c
y
.
T
he
fu
l
l
-
p
a
r
a
m
e
t
e
r
fi
ne
-
t
un
i
ng
i
n
F
i
g
ure
5(
a
)
Indo
BE
R
T
s
t
a
r
t
s
w
i
t
h
t
he
l
ow
e
s
t
a
c
c
ura
c
y
bu
t
s
ur
pa
s
s
e
s
ot
h
e
r
m
ode
l
s
b
y
e
po
c
h
8.
T
he
L
oRA
-
b
a
s
e
d
t
u
ni
ng
gr
a
ph
i
n
F
i
gure
5(
b)
re
v
e
a
l
s
a
l
onge
r
t
ra
i
ni
ng
pe
r
i
od
,
w
i
t
h
L
oRA
-
Ind
oBE
RT
de
m
o
ns
t
ra
t
i
n
g
t
he
h
i
ghe
s
t
f
i
na
l
a
c
c
ur
a
c
y
.
(a
)
(b)
F
i
gure
4
.
H
a
m
m
i
n
g
l
os
s
for
(
a
)
f
ul
l
-
pa
r
a
m
e
t
e
r
fi
n
e
-
t
uni
ng
a
n
d
(b)
L
o
RA
-
ba
s
e
d
t
un
i
ng
(a
)
(b)
F
i
gure
5
.
S
ubs
e
t
a
c
c
ur
a
c
y
f
or
(
a
)
ful
l
-
p
a
ra
m
e
t
e
r
fi
ne
-
t
uni
ng
a
nd
(b)
L
oRA
-
b
a
s
e
d
t
un
i
ng
T
he
e
xp
e
r
i
m
e
n
t
a
s
i
l
l
u
s
t
r
a
t
e
d
i
n
T
a
bl
e
7
,
s
h
ow
s
t
h
a
t
a
m
o
ng
t
h
e
f
u
l
l
f
i
n
e
-
t
u
n
e
d
m
o
d
e
l
s
,
I
nd
o
BE
R
T
a
c
h
i
e
v
e
s
t
h
e
l
o
w
e
s
t
h
a
m
m
i
n
g
l
o
s
s
of
0
.
08
3
1
a
nd
s
h
a
r
e
s
t
h
e
s
a
m
e
h
i
g
h
e
s
t
v
a
l
u
e
f
or
s
u
bs
e
t
a
c
c
ur
a
c
y
o
f
0
.
67
79
w
i
t
h
C
a
hy
a
B
E
R
T
.
T
h
i
s
s
ug
g
e
s
t
s
t
h
a
t
I
n
do
n
e
s
i
a
n
-
s
p
e
c
i
f
i
c
p
r
e
-
t
r
a
i
n
e
d
m
o
d
e
l
s
a
r
e
p
a
r
t
i
c
u
l
a
rl
y
e
f
f
e
c
t
i
v
e
f
o
r
t
h
i
s
t
a
s
k
.
I
n
t
e
r
e
s
t
i
ng
l
y
,
t
h
e
i
m
p
l
e
m
e
n
t
a
t
i
on
o
f
L
oR
A
m
e
t
h
o
d
s
h
ow
s
a
s
l
i
g
h
t
d
e
c
r
e
a
s
e
i
n
t
h
e
p
e
r
fo
r
m
a
n
c
e
o
f
a
l
l
m
od
e
l
s
.
F
or
i
n
s
t
a
n
c
e
,
L
o
RA
-
I
nd
o
B
E
R
T
h
a
s
a
h
i
g
h
e
r
h
a
m
m
i
n
g
l
os
s
(0
.
0
91
4)
a
nd
l
o
w
e
r
s
ubs
e
t
a
c
c
u
r
a
c
y
(0
.
65
17
)
c
o
m
p
a
r
e
d
t
o
i
t
s
fu
l
l
f
i
n
e
-
t
u
n
e
d
m
od
e
l
.
T
h
e
M
B
E
R
T
,
w
hi
l
e
c
o
m
p
e
t
i
t
i
v
e
,
c
on
s
i
s
t
e
n
t
l
y
un
d
e
r
p
e
r
fo
r
m
s
c
o
m
p
a
r
e
d
t
o
I
nd
on
e
s
i
a
n
-
s
p
e
c
i
f
i
c
m
od
e
l
s
,
h
i
g
h
l
i
g
h
t
i
n
g
t
h
e
i
m
p
or
t
a
n
c
e
of
a
l
a
ng
u
a
g
e
-
s
p
e
c
i
f
i
c
m
o
d
e
l
f
or
t
h
i
s
t
a
s
k
.
A
s
s
how
n
i
n
T
a
b
l
e
7
,
t
h
e
re
i
s
a
s
l
i
gh
t
pe
rf
orm
a
n
c
e
ga
p
be
t
w
e
e
n
L
oRA
a
n
d
t
he
f
ul
l
f
i
ne
-
t
un
e
d
m
ode
l
.
T
h
i
s
d
i
ff
e
re
n
c
e
c
a
n
be
a
t
t
r
i
bu
t
e
d
t
o
a
r
c
hi
t
e
c
t
u
ra
l
di
f
fe
re
n
c
e
s
i
n
pa
r
a
m
e
t
e
r
up
da
t
i
ng
,
w
he
r
e
fu
l
l
fi
ne
-
t
uni
n
g
m
odi
f
i
e
s
a
l
l
c
o
m
po
ne
n
t
s
of
t
he
m
od
e
l
s
,
w
hi
l
e
L
oR
A
onl
y
u
pda
t
e
s
p
a
r
a
m
e
t
e
rs
i
n
t
h
e
s
e
l
f
-
a
t
t
e
nt
i
on
l
a
y
e
r
t
hrough
L
oRA
.
A
s
r
e
por
t
e
d
i
n
T
a
bl
e
8,
L
o
RA
on
l
y
t
r
a
i
ns
a
bou
t
0
.
78%
of
t
ot
a
l
pa
r
a
m
e
t
e
rs
i
n
Ca
hy
a
BE
RT
a
n
d
i
ndoB
E
R
T
,
a
nd
0.
4
9%
i
n
m
B
E
R
T
.
In
t
he
m
ul
t
i
-
l
a
b
e
l
c
ont
e
xt
,
w
he
r
e
t
he
d
a
t
a
o
ft
e
n
c
on
t
a
i
n
ove
rl
a
pp
i
ng
t
opi
c
s
,
L
oRA
’s
l
i
m
i
t
e
d
pa
r
a
m
e
t
e
r
s
p
a
c
e
pa
rt
i
c
u
l
a
rl
y
a
f
fe
c
t
s
i
t
s
a
b
i
l
i
t
y
t
o
c
a
p
t
ur
e
c
om
p
l
e
x
t
op
i
c
i
nt
e
rde
pe
nd
e
n
c
i
e
s
b
e
t
w
e
e
n
5
di
f
fe
r
e
nt
l
a
b
e
l
s
.
T
hi
s
a
r
c
h
i
t
e
c
t
ura
l
c
ons
t
ra
i
nt
be
c
o
m
e
s
e
v
i
d
e
nt
i
n
ou
r
r
e
s
ul
t
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
,
V
o
l
.
6
,
N
o
.
3
,
N
ov
e
m
be
r
20
25
:
274
-
282
280
w
he
re
L
oRA
re
q
ui
r
e
s
m
o
re
e
po
c
hs
t
o
a
c
hi
e
ve
opt
i
m
a
l
pe
r
fo
rm
a
n
c
e
c
o
m
p
a
re
d
t
o
ful
l
fi
ne
-
t
uni
ng,
a
s
a
t
r
a
d
e
-
off
b
e
t
w
e
e
n
pa
ra
m
e
t
e
r
e
ffi
c
i
e
nc
y
a
nd
m
od
e
l
a
da
p
t
a
b
i
l
i
t
y
.
T
a
b
l
e
7
.
E
va
l
ua
t
i
o
n
r
e
s
ul
t
s
on
t
e
s
t
i
ng
da
t
a
M
o
d
e
l
H
a
m
m
i
n
g
l
o
s
s
S
u
b
s
e
t
a
c
c
u
ra
c
y
c
a
h
y
a
/
b
e
rt
-
b
a
s
e
/
i
n
d
o
n
e
s
i
a
n
-
1
.
5
G
0
.
0
8
6
9
0
.
6
7
7
9
i
n
d
o
l
e
m
/
i
n
d
o
b
e
rt
-
b
a
s
e
-
u
n
c
a
s
e
d
0
.
0
8
3
1
0
.
6
7
7
9
b
e
rt
-
b
a
s
e
-
m
u
l
t
i
l
i
n
g
u
a
l
-
c
a
s
e
d
0
.
0
9
7
4
0
.
6
5
1
7
c
a
h
y
a
/
b
e
rt
-
b
a
s
e
/
i
n
d
o
n
e
s
i
a
n
-
1
.
5
G
+
L
o
RA
0
.
1
0
1
9
0
.
6
2
5
5
i
n
d
o
l
e
m
/
i
n
d
o
b
e
rt
-
b
a
s
e
-
u
n
c
a
s
e
d
+
L
o
RA
0
.
0
9
1
4
0
.
6
5
1
7
b
e
rt
-
b
a
s
e
-
m
u
l
t
i
l
i
n
g
u
a
l
-
c
a
s
e
d
+
L
o
RA
0
.
1
0
0
4
0
.
6
2
1
7
T
a
b
l
e
8
.
T
ra
i
na
bl
e
p
a
ra
m
e
t
e
rs
M
o
d
e
l
T
o
t
a
l
p
a
ra
m
e
t
e
r
L
o
RA
t
ra
i
n
a
b
l
e
p
a
ra
m
e
t
e
r
T
ra
i
n
a
b
l
e
(%
)
c
a
h
y
a
/
b
e
rt
-
b
a
s
e
/
i
n
d
o
n
e
s
i
a
n
-
1
.
5
G
1
1
3
.
8
7
3
.
6
7
4
8
8
8
.
5
8
1
0
.
7
8
0
3
i
n
d
o
l
e
m
/
i
n
d
o
b
e
rt
-
b
a
s
e
-
u
n
c
a
s
e
d
1
1
3
.
8
1
4
.
5
3
8
8
8
8
.
5
8
1
0
.
7
8
0
7
b
e
rt
-
b
a
s
e
-
m
u
l
t
i
l
i
n
g
u
a
l
-
c
a
s
e
d
1
8
1
.
1
0
9
.
7
7
0
8
8
8
.
5
8
1
0
.
4
9
0
6
W
hi
l
e
L
o
RA
a
c
h
i
e
v
e
s
s
l
i
ght
l
y
l
ow
e
r
p
e
rfor
m
a
nc
e
m
e
t
ri
c
s
,
F
i
gure
6
s
how
s
t
h
a
t
L
o
RA
-
IndoB
E
R
T
onl
y
us
e
s
52
%
of
G
P
U
m
e
m
ory
a
nd
t
r
a
i
n
i
ng
d
ura
t
i
on
c
o
m
pl
e
t
e
d
i
n
u
nde
r
45
m
i
nut
e
s
,
w
hi
c
h
i
s
a
69.
7%
re
duc
t
i
on
i
n
t
i
m
e
c
o
m
pa
re
d
t
o
ful
l
f
i
ne
-
t
un
i
ng
.
T
hi
s
e
ff
i
c
i
e
n
c
y
m
a
k
e
s
L
oRA
a
n
a
t
t
ra
c
t
i
ve
op
t
i
on
for
i
ns
t
i
t
ut
i
ons
w
i
t
h
l
i
m
i
t
e
d
c
o
m
pu
t
a
t
i
on
a
l
r
e
s
our
c
e
s
,
a
s
i
t
c
a
n
e
ffe
c
t
i
v
e
l
y
run
on
c
ons
u
m
e
r
-
g
ra
d
e
G
P
U
s
.
F
ut
u
re
i
m
pro
ve
m
e
n
t
s
c
o
ul
d
f
oc
us
o
n
op
t
i
m
i
z
i
ng
L
oRA
’s
a
r
c
hi
t
e
c
t
ur
e
s
pe
c
i
f
i
c
a
l
l
y
for
m
u
l
t
i
-
l
a
be
l
c
l
a
s
s
i
f
i
c
a
t
i
on
t
a
s
ks
,
s
uc
h
a
s
de
v
e
l
opi
ng
m
e
c
h
a
ni
s
m
s
t
o
be
t
t
e
r
c
a
pt
ur
e
l
a
be
l
re
l
a
t
i
ons
hi
ps
or
e
xpl
o
ri
ng
s
e
l
e
c
t
i
ve
p
a
r
a
m
e
t
e
r
a
da
p
t
a
t
i
on
i
n
o
t
h
e
r
l
a
y
e
rs
w
h
i
l
e
pre
s
e
rvi
n
g
i
t
s
c
or
e
e
ffi
c
i
e
n
c
y
be
n
e
fi
t
s
.
F
i
gure
6
.
GPU
m
e
m
or
y
u
t
i
l
i
z
a
t
i
on
4.
C
O
N
C
LU
S
I
O
N
T
hi
s
s
t
udy
ha
s
s
uc
c
e
s
s
ful
l
y
de
m
ons
t
r
a
t
e
d
t
he
e
ffe
c
t
i
ve
n
e
s
s
o
f
ut
i
l
i
z
i
ng
pr
e
-
t
ra
i
ne
d
l
a
ngu
a
g
e
m
ode
l
s
,
e
s
pe
c
i
a
l
l
y
C
a
hya
BE
R
T
,
Indo
BE
R
T
,
a
nd
M
B
E
R
T
,
w
i
t
h
fu
l
l
-
pa
ra
m
e
t
e
r
f
i
ne
-
t
un
i
ng
a
nd
L
oRA
-
ba
s
e
d
t
u
ni
ng
t
e
c
hni
q
ue
s
f
or
m
ul
t
i
-
l
a
be
l
c
l
a
s
s
i
fi
c
a
t
i
on
of
di
p
l
om
a
t
i
c
c
a
bl
e
s
.
T
hi
s
re
s
e
a
r
c
h
f
i
nds
t
h
a
t
t
h
e
Ind
one
s
i
a
n
-
s
p
e
c
i
fi
c
m
ode
l
,
pa
r
t
i
c
ul
a
rl
y
I
ndoB
E
RT
,
ou
t
pe
rfor
m
s
t
he
m
ul
t
i
l
i
ng
u
a
l
m
ode
l
i
n
a
c
c
ura
t
e
l
y
c
a
t
e
gor
i
z
i
ng
d
i
pl
o
m
a
t
i
c
c
a
b
l
e
s
i
n
t
o
po
l
i
t
i
c
s
,
e
c
ono
m
i
c
s
,
pro
t
oc
o
l
a
n
d
c
ons
u
l
a
r
s
e
r
vi
c
e
s
,
s
oc
i
o
-
c
ul
t
ura
l
a
ff
a
i
rs
,
a
nd
a
d
m
i
n
i
s
t
ra
t
i
v
e
m
a
t
t
e
rs
.
L
oRA
de
m
ons
t
ra
t
e
d
s
i
gni
f
i
c
a
nt
a
dv
a
nt
a
g
e
s
i
n
G
P
U
m
e
m
ory
c
ons
u
m
pt
i
on
by
4
8%
a
n
d
t
r
a
i
n
i
ng
t
i
m
e
by
69
.
7%
c
o
m
pa
r
e
d
t
o
fu
l
l
f
i
ne
-
t
un
i
ng
.
H
ow
e
ve
r,
by
fo
c
us
i
ng
L
oRA
a
d
a
pt
a
t
i
on
on
t
he
s
e
l
f
-
a
t
t
e
nt
i
on
l
a
y
e
r,
t
he
s
t
udy
r
e
ve
a
l
e
d
s
e
ve
ra
l
c
ha
l
l
e
nge
s
,
w
h
e
r
e
t
he
m
ode
l
’s
a
bi
l
i
t
y
t
o
c
a
pt
ure
c
o
m
p
l
e
x
r
e
l
a
t
i
o
ns
hi
ps
be
t
w
e
e
n
l
a
b
e
l
s
w
a
s
s
l
i
ght
l
y
c
o
m
pro
m
i
s
e
d
.
L
oRA
s
how
e
d
a
s
m
a
l
l
dr
op
i
n
pe
rfor
m
a
n
c
e
m
e
t
r
i
c
s
c
o
m
pa
re
d
t
o
t
h
e
fu
l
l
fi
ne
-
t
un
i
ng
m
e
t
hod
,
by
3
.
87%
l
ow
e
r
i
n
s
ubs
e
t
a
c
c
ura
c
y
a
nd
9
.
99%
hi
gh
e
r
i
n
ha
m
m
i
ng
l
os
s
.
F
u
t
ure
r
e
s
e
a
rc
h
c
oul
d
e
xpl
ore
op
t
i
m
i
z
i
ng
L
oRA
’s
a
pp
l
i
c
a
t
i
o
n
a
c
r
os
s
di
ffe
r
e
nt
m
od
e
l
l
a
ye
rs
or
e
xpl
ori
ng
h
ybri
d
f
i
nd
-
t
uni
n
g
a
ppro
a
c
h
e
s
t
ha
t
c
a
n
b
e
t
t
e
r
pr
e
s
e
rv
e
s
e
m
a
nt
i
c
c
om
p
l
e
x
i
t
y
.
U
l
t
i
m
a
t
e
l
y,
t
hi
s
e
xp
e
ri
m
e
nt
s
how
s
t
ha
t
L
oRA
i
s
a
prom
i
s
i
ng
a
pp
roa
c
h
for
di
p
l
om
a
t
i
c
i
ns
t
i
t
ut
i
ons
w
i
t
h
l
i
m
i
t
e
d
c
o
m
put
a
t
i
on
a
l
re
s
our
c
e
s
,
bri
dg
i
ng
t
h
e
ga
p
be
t
w
e
e
n
m
ode
l
p
e
rfor
m
a
nc
e
a
nd
c
o
m
pu
t
a
t
i
o
na
l
e
ffi
c
i
e
nc
y.
Evaluation Warning : The document was created with Spire.PDF for Python.
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
IS
S
N
:
2722
-
3221
O
pt
i
m
i
z
i
ng
di
p
l
om
at
i
c
i
n
de
x
i
ng:
f
ul
l
-
par
am
e
t
e
r
v
s
l
ow
-
r
a
nk
adapt
a
t
i
on
f
or
m
u
l
t
i
-
l
ab
e
l
…
(
D
e
l
a
Nur
l
ai
l
a
)
281
F
U
N
D
I
N
G
I
N
F
O
R
M
A
TI
O
N
N
o
fund
i
ng
i
nv
ol
v
e
d.
A
U
TH
O
R
C
O
N
TR
I
BU
TI
O
N
S
S
TA
T
EM
EN
T
T
hi
s
j
our
na
l
us
e
s
t
h
e
Cont
ri
bu
t
or
Rol
e
s
T
a
x
ono
m
y
(
C
Re
di
T
)
t
o
re
c
ogn
i
z
e
i
nd
i
vi
du
a
l
a
ut
hor
c
ont
r
i
bu
t
i
ons
,
r
e
du
c
e
a
ut
hors
h
i
p
di
s
pu
t
e
s
,
a
nd
fa
c
i
l
i
t
a
t
e
c
o
l
l
a
bora
t
i
on
.
N
ame
of
A
u
th
o
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
D
e
l
a
N
ur
l
a
i
l
a
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
A
bba
S
u
ga
nd
a
G
i
rs
a
n
g
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
ft
w
a
re
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
rm
a
l
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
s
o
u
rc
e
s
D
:
D
a
t
a
Cu
ra
t
i
o
n
O
:
W
ri
t
i
n
g
-
O
ri
g
i
n
a
l
D
ra
ft
E
:
W
ri
t
i
n
g
-
Re
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
s
u
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
rv
i
s
i
o
n
P
:
P
ro
j
e
c
t
a
d
m
i
n
i
s
t
ra
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
s
i
t
i
o
n
C
O
N
F
LI
C
T
O
F
I
N
T
ER
ES
T
S
TA
T
EM
EN
T
A
ut
hors
s
t
a
t
e
no
c
onf
l
i
c
t
of
i
n
t
e
r
e
s
t
.
D
A
TA
A
V
A
I
LA
BI
LI
TY
T
he
d
a
t
a
t
ha
t
s
uppor
t
t
he
f
i
ndi
ngs
o
f
t
hi
s
s
t
ud
y
a
re
r
e
s
t
ri
c
t
e
d
a
nd
not
a
va
i
l
a
bl
e
t
o
t
h
e
publ
i
c
du
e
t
o
c
onfi
de
n
t
i
a
l
i
t
y
re
q
ui
r
e
m
e
nt
s
fro
m
t
he
d
a
t
a
p
rovi
d
e
r
i
ns
t
i
t
ut
i
o
n.
R
EF
ER
EN
C
ES
[1
]
U
n
i
t
e
d
N
a
t
i
o
n
s
,
“
V
i
e
n
n
a
c
o
n
v
e
n
t
i
o
n
o
n
d
i
p
l
o
m
a
t
i
c
r
e
l
a
t
i
o
n
s
(
1
9
6
1
),
”
M
a
x
P
l
a
n
c
k
E
n
c
y
c
l
o
p
e
d
i
a
o
f
P
u
b
l
i
c
I
n
t
e
r
n
a
t
i
o
n
a
l
L
a
w
,
2
0
0
9
.
[O
n
l
i
n
e
].
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
t
re
a
t
i
e
s
.
u
n
.
o
rg
/
d
o
c
/
T
re
a
t
i
e
s
/
1
9
6
4
/
0
6
/
1
9
6
4
0
6
2
4
0
2
-
1
0
A
M
/
Ch
_
III
_
3
p
.
p
d
f
.
[A
c
c
e
s
s
e
d
A
u
g
.
1
0
,
2
0
2
4
]
.
[2
]
S
.
Ba
rm
a
n
,
“
D
i
g
i
t
a
l
d
i
p
l
o
m
a
c
y
:
t
h
e
i
n
fl
u
e
n
c
e
o
f
d
i
g
i
t
a
l
p
l
a
t
fo
rm
s
o
n
g
l
o
b
a
l
d
i
p
l
o
m
a
c
y
a
n
d
fo
re
i
g
n
p
o
l
i
c
y
,
”
V
i
d
y
a
-
a
J
o
u
r
n
a
l
o
f
G
u
j
a
r
a
t
U
n
i
v
e
r
s
i
t
y
,
v
o
l
.
3
,
n
o
.
1
,
p
p
.
6
1
–
7
5
,
2
0
2
4
,
d
o
i
:
1
0
.
4
7
4
1
3
/
v
i
d
y
a
.
v
3
i
1
.
3
0
4
.
[3
]
C.
F
e
rn
a
n
d
e
s
,
“
T
h
e
c
a
b
l
e
s
a
n
d
t
h
e
i
r
re
c
e
p
t
i
o
n
,
”
i
n
W
h
a
t
u
n
c
l
e
S
a
m
W
a
n
t
s
,
P
a
l
g
ra
v
e
P
i
v
o
t
S
i
n
g
a
p
o
re
,
2
0
1
9
.
[4
]
C.
Bj
o
l
a
,
J
.
Ca
s
s
i
d
y
,
a
n
d
I.
M
a
n
o
rc
,
“
P
u
b
l
i
c
d
i
p
l
o
m
a
c
y
i
n
t
h
e
d
i
g
i
t
a
l
a
g
e
,
”
T
h
e
H
a
g
u
e
J
o
u
r
n
a
l
o
f
D
i
p
l
o
m
a
c
y
,
v
o
l
.
1
4
,
n
o
.
1
–
2
,
p
p
.
8
3
–
1
0
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
6
3
/
1
8
7
1
1
9
1
X
-
1
4
0
1
1
0
3
2
.
[5
]
A
.
N
.
T
a
re
k
e
g
n
,
M
.
G
i
a
c
o
b
i
n
i
,
a
n
d
K
.
M
i
c
h
a
l
a
k
,
“
A
re
v
i
e
w
o
f
m
e
t
h
o
d
s
fo
r
i
m
b
a
l
a
n
c
e
d
m
u
l
t
i
-
l
a
b
e
l
c
l
a
s
s
i
fi
c
a
t
i
o
n
,
”
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
1
1
8
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
c
o
g
.
2
0
2
1
.
1
0
7
9
6
5
.
[6
]
V
.
S
.
T
i
d
a
k
e
a
n
d
S
.
S
.
S
a
n
e
,
“
E
v
a
l
u
a
t
i
o
n
o
f
m
u
l
t
i
-
l
a
b
e
l
c
l
a
s
s
i
fi
e
rs
i
n
v
a
ri
o
u
s
d
o
m
a
i
n
s
u
s
i
n
g
d
e
c
i
s
i
o
n
t
re
e
,
”
A
d
v
a
n
c
e
s
i
n
In
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
a
n
d
Co
m
p
u
t
i
n
g
,
v
o
l
.
6
7
3
,
p
p
.
1
1
7
–
1
2
7
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
9
8
1
-
10
-
7
2
4
5
-
1
_
1
3
.
[7
]
C.
S
u
n
,
X
.
Q
i
u
,
Y
.
X
u
,
a
n
d
X
.
H
u
a
n
g
,
“
H
o
w
t
o
fi
n
e
-
t
u
n
e
BE
RT
fo
r
t
e
x
t
c
l
a
s
s
i
fi
c
a
t
i
o
n
?
,
”
L
e
c
t
u
r
e
No
t
e
s
i
n
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
(i
n
c
l
u
d
i
n
g
s
u
b
s
e
r
i
e
s
L
e
c
t
u
r
e
No
t
e
s
i
n
A
r
t
i
f
i
c
i
a
l
In
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
r
e
No
t
e
s
i
n
B
i
o
i
n
f
o
r
m
a
t
i
c
s
)
,
p
p
.
1
9
4
–
2
0
6
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
030
-
3
2
3
8
1
-
3
_
1
6
.
[8
]
J
.
D
o
d
g
e
,
G
.
Il
h
a
rc
o
,
R.
S
c
h
w
a
rt
z
,
A
.
F
a
rh
a
d
i
,
H
.
H
a
j
i
s
h
i
rz
i
,
a
n
d
N
.
S
m
i
t
h
,
“
F
i
n
e
-
t
u
n
i
n
g
p
re
t
ra
i
n
e
d
l
a
n
g
u
a
g
e
m
o
d
e
l
s
:
w
e
i
g
h
t
i
n
i
t
i
a
l
i
z
a
t
i
o
n
s
,
d
a
t
a
o
rd
e
r
s
,
a
n
d
e
a
rl
y
s
t
o
p
p
i
n
g
,
”
a
r
X
i
v
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
,
2
0
2
0
.
[9
]
E
.
H
u
e
t
a
l
.
,
“
L
o
Ra
:
L
o
w
-
ra
n
k
a
d
a
p
t
a
t
i
o
n
o
f
l
a
rg
e
l
a
n
g
u
a
g
e
m
o
d
e
l
s
,
”
ICL
R
2
0
2
2
-
1
0
t
h
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
e
r
e
n
c
e
o
n
L
e
a
r
n
i
n
g
R
e
p
r
e
s
e
n
t
a
t
i
o
n
s
,
2
0
2
2
.
[1
0
]
J
.
D
e
v
l
i
n
,
M
.
-
W
.
Ch
a
n
g
,
K
.
L
e
e
,
a
n
d
K
.
T
o
u
t
a
n
o
v
a
,
“
BE
RT
:
p
re
-
t
ra
i
n
i
n
g
o
f
d
e
e
p
b
i
d
i
re
c
t
i
o
n
a
l
t
ra
n
s
fo
rm
e
rs
fo
r
l
a
n
g
u
a
g
e
u
n
d
e
rs
t
a
n
d
i
n
g
,
”
E
a
r
l
y
H
u
m
a
n
D
e
v
e
l
o
p
m
e
n
t
,
v
o
l
.
8
3
,
n
o
.
1
,
p
p
.
1
–
1
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
a
rl
h
u
m
d
e
v
.
2
0
0
6
.
0
5
.
0
2
2
.
[1
1
]
N
.
K
.
N
i
s
s
a
a
n
d
E
.
Y
u
l
i
a
n
t
i
,
“
M
u
l
t
i
-
l
a
b
e
l
t
e
x
t
c
l
a
s
s
i
fi
c
a
t
i
o
n
o
f
In
d
o
n
e
s
i
a
n
c
u
s
t
o
m
e
r
re
v
i
e
w
s
u
s
i
n
g
b
i
d
i
re
c
t
i
o
n
a
l
e
n
c
o
d
e
r
re
p
re
s
e
n
t
a
t
i
o
n
s
fro
m
t
ra
n
s
fo
rm
e
rs
l
a
n
g
u
a
g
e
m
o
d
e
l
,
”
In
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
Co
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
3
,
n
o
.
5
,
p
p
.
5
6
4
1
–
5
6
5
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
3
i
5
.
p
p
5
6
4
1
-
5
6
5
2
.
[1
2
]
G
.
Z
.
N
a
b
i
i
l
a
h
,
I.
N
.
A
l
a
m
,
E
.
S
.
P
u
rw
a
n
t
o
,
a
n
d
M
.
F
.
H
i
d
a
y
a
t
,
“
In
d
o
n
e
s
i
a
n
m
u
l
t
i
l
a
b
e
l
c
l
a
s
s
i
fi
c
a
t
i
o
n
u
s
i
n
g
In
d
o
BE
RT
e
m
b
e
d
d
i
n
g
a
n
d
M
BE
RT
c
l
a
s
s
i
fi
c
a
t
i
o
n
,
”
In
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
Co
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
4
,
n
o
.
1
,
p
p
.
1
0
7
1
–
1
0
7
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
4
i
1
.
p
p
1
0
7
1
-
1
0
7
8
.
[1
3
]
E
.
T
.
L
u
t
h
fi
,
Z
.
I.
M
.
Y
u
s
o
h
,
a
n
d
B.
M
.
A
b
o
o
b
a
i
d
e
r,
“
BE
RT
b
a
s
e
d
n
a
m
e
d
e
n
t
i
t
y
re
c
o
g
n
i
t
i
o
n
fo
r
a
u
t
o
m
a
t
e
d
h
a
d
i
t
h
n
a
rra
t
o
r
i
d
e
n
t
i
fi
c
a
t
i
o
n
,
”
In
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
3
,
n
o
.
1
,
p
p
.
6
0
4
–
6
1
1
,
2
0
2
2
,
d
o
i
:
1
0
.
1
4
5
6
9
/
I
J
A
CS
A
.
2
0
2
2
.
0
1
3
0
1
7
3
.
[1
4
]
K
.
C.
Ch
i
e
n
,
C
.
H
.
Ch
a
n
g
,
a
n
d
R.
D
e
r
S
u
n
,
“
U
s
i
n
g
p
a
ra
m
e
t
e
r
e
ffi
c
i
e
n
t
f
i
n
e
-
t
u
n
i
n
g
o
n
l
e
g
a
l
a
rt
i
fi
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
”
i
n
CE
U
R
W
o
r
k
s
h
o
p
P
r
o
c
e
e
d
i
n
g
s
,
v
o
l
.
3
6
3
7
,
2
0
2
3
.
[1
5
]
M
i
n
i
s
t
ry
o
f
F
o
re
i
g
n
A
ffa
i
rs
,
“
D
e
c
re
e
o
f
t
h
e
M
i
n
i
s
t
e
r
o
f
F
o
re
i
g
n
A
f
fa
i
rs
N
u
m
b
e
r
S
K
.
0
6
/
A
/
O
T
/
V
I/
2
0
0
4
/
0
1
y
e
a
r
2
0
0
4
c
o
n
c
e
rn
i
n
g
O
rg
a
n
i
z
a
t
i
o
n
a
n
d
W
o
rk
i
n
g
P
ro
c
e
d
u
re
s
o
f
t
h
e
Re
p
u
b
l
i
c
o
f
In
d
o
n
e
s
i
a
's
re
p
re
s
e
n
t
a
t
i
v
e
s
a
b
ro
a
d
”
,
(i
n
Ba
h
a
s
a
:
K
e
p
u
t
u
s
a
n
M
e
n
t
e
ri
L
u
a
r
N
e
g
e
ri
N
o
m
o
r
S
K
.
0
6
/
A
/
O
T
/
V
I/
2
0
0
4
/
0
1
T
a
h
u
n
2
0
0
4
t
e
n
t
a
n
g
O
rg
a
n
i
s
a
s
i
d
a
n
T
a
t
a
K
e
rj
a
P
e
rw
a
k
i
l
a
n
Re
p
u
b
l
i
k
In
d
o
n
e
s
i
a
d
i
L
u
a
r
N
e
g
e
ri
),
2
0
0
4
.
[1
6
]
A
.
K
u
rn
i
a
s
i
h
a
n
d
L
.
P
.
M
a
n
i
k
,
“
O
n
t
h
e
ro
l
e
o
f
t
e
x
t
p
re
p
ro
c
e
s
s
i
n
g
i
n
BE
RT
e
m
b
e
d
d
i
n
g
-
b
a
s
e
d
D
N
N
s
fo
r
c
l
a
s
s
i
fy
i
n
g
i
n
fo
rm
a
l
t
e
x
t
s
,
”
In
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
3
,
n
o
.
6
,
p
p
.
9
2
7
–
9
3
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
4
5
6
9
/
IJ
A
CS
A
.
2
0
2
2
.
0
1
3
0
6
1
0
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2722
-
3221
Com
pu
t
S
c
i
Inf
T
e
c
h
nol
,
V
o
l
.
6
,
N
o
.
3
,
N
ov
e
m
be
r
20
25
:
274
-
282
282
[1
7
]
I.
O
.
M
u
ra
i
n
a
,
“
Id
e
a
l
d
a
t
a
s
e
t
s
p
l
i
t
t
i
n
g
ra
t
i
o
s
i
n
m
a
c
h
i
n
e
l
e
a
rn
i
n
g
a
l
g
o
ri
t
h
m
s
:
g
e
n
e
ra
l
c
o
n
c
e
rn
s
fo
r
d
a
t
a
s
c
i
e
n
t
i
s
t
s
a
n
d
d
a
t
a
a
n
a
l
y
s
t
s
,
”
i
n
7
t
h
In
t
e
r
n
a
t
i
o
n
a
l
M
a
r
d
i
n
A
r
t
u
k
l
u
S
c
i
e
n
t
i
f
i
c
R
e
s
e
a
r
c
h
e
s
Co
n
f
e
r
e
n
c
e
,
p
p
.
4
9
6
–
5
0
4
,
2
0
2
2
.
[1
8
]
K
.
S
e
c
h
i
d
i
s
,
G
.
T
s
o
u
m
a
k
a
s
,
a
n
d
I.
V
l
a
h
a
v
a
s
,
“
O
n
t
h
e
s
t
ra
t
i
fi
c
a
t
i
o
n
o
f
m
u
l
t
i
-
l
a
b
e
l
d
a
t
a
,
”
L
e
c
t
u
r
e
No
t
e
s
i
n
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
(i
n
c
l
u
d
i
n
g
s
u
b
s
e
r
i
e
s
L
e
c
t
u
r
e
No
t
e
s
i
n
A
r
t
i
f
i
c
i
a
l
In
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
r
e
No
t
e
s
i
n
B
i
o
i
n
f
o
r
m
a
t
i
c
s
)
,
p
p
.
1
4
5
–
1
5
8
,
2
0
1
1
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
642
-
2
3
8
0
8
-
6
_
1
0
.
[1
9
]
J
.
L
e
e
e
t
a
l
.
,
“
K
-
M
H
a
S
:
a
m
u
l
t
i
-
l
a
b
e
l
h
a
t
e
s
p
e
e
c
h
d
e
t
e
c
t
i
o
n
d
a
t
a
s
e
t
i
n
K
o
re
a
n
o
n
l
i
n
e
n
e
w
s
c
o
m
m
e
n
t
,
”
F
i
n
d
i
n
g
s
o
f
t
h
e
A
s
s
o
c
i
a
t
i
o
n
f
o
r
Co
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
:
E
M
NL
P
2
0
2
3
,
p
p
.
1
4
2
6
4
–
1
4
2
7
8
,
2
0
2
2
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
3
.
fi
n
d
i
n
g
s
-
e
m
n
l
p
.
9
5
2
.
[2
0
]
F
.
K
o
t
o
,
A
.
Ra
h
i
m
i
,
J
.
H
.
L
a
u
,
a
n
d
T
.
Ba
l
d
w
i
n
,
“
In
d
o
L
E
M
a
n
d
In
d
o
BE
RT
:
a
b
e
n
c
h
m
a
rk
d
a
t
a
s
e
t
a
n
d
p
re
-
t
ra
i
n
e
d
l
a
n
g
u
a
g
e
m
o
d
e
l
fo
r
I
n
d
o
n
e
s
i
a
n
N
L
P
,
”
CO
L
ING
2
0
2
0
-
2
8
t
h
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
e
r
e
n
c
e
o
n
Co
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
Co
n
f
e
r
e
n
c
e
,
p
p
.
7
5
7
–
7
7
0
,
2
0
2
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
c
o
l
i
n
g
-
m
a
i
n
.
6
6
.
[2
1
]
T
.
P
i
re
s
,
E
.
S
c
h
l
i
n
g
e
r,
a
n
d
D
.
G
a
rre
t
t
e
,
“
H
o
w
m
u
l
t
i
l
i
n
g
u
a
l
i
s
m
u
l
t
i
l
i
n
g
u
a
l
BE
R
T
?
,
”
A
CL
2
0
1
9
-
5
7
t
h
A
n
n
u
a
l
M
e
e
t
i
n
g
o
f
t
h
e
A
s
s
o
c
i
a
t
i
o
n
f
o
r
Co
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
Co
n
f
e
r
e
n
c
e
,
p
p
.
4
9
9
6
–
5
0
0
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
p
1
9
-
1
4
9
3
.
[2
2
]
H
.
M
.
Z
a
h
e
ra
,
I
.
E
l
g
e
n
d
y
,
R.
J
a
l
o
t
a
,
a
n
d
M
.
A
.
S
h
e
ri
f
,
“
F
i
n
e
-
t
u
n
e
d
BE
RT
m
o
d
e
l
fo
r
m
u
l
t
i
-
l
a
b
e
l
t
w
e
e
t
s
c
l
a
s
s
i
f
i
c
a
t
i
o
n
,
”
i
n
2
8
t
h
T
e
x
t
R
E
t
r
i
e
v
a
l
Co
n
f
e
r
e
n
c
e
,
T
R
E
C
2
0
1
9
-
P
r
o
c
e
e
d
i
n
g
s
,
p
p
.
1
–
7
,
2
0
1
9
.
[2
3
]
M
.
Bi
l
a
l
a
n
d
A
.
A
.
A
l
m
a
z
ro
i
,
“
E
ffe
c
t
i
v
e
n
e
s
s
o
f
fi
n
e
-
t
u
n
e
d
BE
RT
m
o
d
e
l
i
n
c
l
a
s
s
i
fi
c
a
t
i
o
n
o
f
h
e
l
p
f
u
l
a
n
d
u
n
h
e
l
p
fu
l
o
n
l
i
n
e
c
u
s
t
o
m
e
r
re
v
i
e
w
s
,
”
E
l
e
c
t
r
o
n
i
c
Co
m
m
e
r
c
e
R
e
s
e
a
r
c
h
,
v
o
l
.
2
3
,
n
o
.
4
,
p
p
.
2
7
3
7
–
2
7
5
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
6
6
0
-
0
2
2
-
0
9
5
6
0
-
w.
[2
4
]
Y
.
M
a
o
e
t
a
l
.
,
“
A
s
u
rv
e
y
o
n
L
o
RA
o
f
l
a
rg
e
l
a
n
g
u
a
g
e
m
o
d
e
l
s
,
”
F
r
o
n
t
i
e
r
s
o
f
Co
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
9
,
n
o
.
7
,
2
0
2
5
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
7
0
4
-
0
2
4
-
4
0
6
6
3
-
9.
[2
5
]
K
.
D
h
o
l
e
e
t
a
l
.
,
“
N
L
-
a
u
g
m
e
n
t
e
r
a
fr
a
m
e
w
o
rk
fo
r
t
a
s
k
-
s
e
n
s
i
t
i
v
e
n
a
t
u
ra
l
l
a
n
g
u
a
g
e
a
u
g
m
e
n
t
a
t
i
o
n
,
”
No
r
t
h
e
r
n
E
u
r
o
p
e
a
n
J
o
u
r
n
a
l
o
f
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
9
,
n
o
.
1
,
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
4
/
n
e
j
l
t
.
2
0
0
0
-
1
5
3
3
.
2
0
2
3
.
4
7
2
5
.
[2
6
]
S
.
E
d
u
n
o
v
,
M
.
O
t
t
,
M
.
A
u
l
i
,
a
n
d
D
.
G
ra
n
g
i
e
r,
“
U
n
d
e
rs
t
a
n
d
i
n
g
b
a
c
k
-
t
ra
n
s
l
a
t
i
o
n
a
t
s
c
a
l
e
,
”
i
n
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
8
Co
n
f
e
r
e
n
c
e
o
n
E
m
p
i
r
i
c
a
l
M
e
t
h
o
d
s
i
n
Na
t
u
r
a
l
L
a
n
g
u
a
g
e
P
r
o
c
e
s
s
i
n
g
,
E
M
NL
P
2
0
1
8
,
2
0
1
8
,
p
p
.
4
8
9
–
5
0
0
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
d
1
8
-
1
0
4
5
.
[2
7
]
R.
S
e
n
n
ri
c
h
,
B.
H
a
d
d
o
w
,
a
n
d
A
.
Bi
rc
h
,
“
Im
p
ro
v
i
n
g
n
e
u
ra
l
m
a
c
h
i
n
e
t
ra
n
s
l
a
t
i
o
n
m
o
d
e
l
s
w
i
t
h
m
o
n
o
l
i
n
g
u
a
l
d
a
t
a
,
”
5
4
t
h
A
n
n
u
a
l
M
e
e
t
i
n
g
o
f
t
h
e
A
s
s
o
c
i
a
t
i
o
n
f
o
r
Co
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
A
C
L
2
0
1
6
-
L
o
n
g
P
a
p
e
r
s
,
v
o
l
.
1
,
p
p
.
8
6
–
9
6
,
2
0
1
6
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
p
1
6
-
1
0
0
9
.
[2
8
]
M
.
G
ra
ç
a
,
Y
.
K
i
m
,
J
.
S
c
h
a
m
p
e
r,
S
.
K
h
a
d
i
v
i
,
a
n
d
H
.
N
e
y
,
“
G
e
n
e
ra
l
i
z
i
n
g
b
a
c
k
-
t
ra
n
s
l
a
t
i
o
n
i
n
n
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
i
n
W
M
T
2
0
1
9
-
4
t
h
C
o
n
f
e
r
e
n
c
e
o
n
M
a
c
h
i
n
e
T
r
a
n
s
l
a
t
i
o
n
,
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
C
o
n
f
e
r
e
n
c
e
,
2
0
1
9
,
v
o
l
.
1
,
p
p
.
4
5
–
5
2
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
w
1
9
-
5
2
0
5
.
[2
9
]
V
.
G
a
n
g
a
n
w
a
r
a
n
d
R.
Ra
j
a
l
a
k
s
h
m
i
,
“
M
T
D
O
T
:
A
m
u
l
t
i
l
i
n
g
u
a
l
t
ra
n
s
l
a
t
i
o
n
-
b
a
s
e
d
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
t
e
c
h
n
i
q
u
e
fo
r
o
ffe
n
s
i
v
e
c
o
n
t
e
n
t
i
d
e
n
t
i
fi
c
a
t
i
o
n
i
n
T
a
m
i
l
t
e
x
t
d
a
t
a
,
”
E
l
e
c
t
r
o
n
i
c
s
,
v
o
l
.
1
1
,
n
o
.
2
1
,
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
ro
n
i
c
s
1
1
2
1
3
5
7
4
.
[3
0
]
J
.
V
a
n
N
o
o
t
e
n
a
n
d
W
.
D
a
e
l
e
m
a
n
s
,
“
Im
p
ro
v
i
n
g
d
u
t
c
h
v
a
c
c
i
n
e
h
e
s
i
t
a
n
c
y
m
o
n
i
t
o
ri
n
g
v
i
a
m
u
l
t
i
-
l
a
b
e
l
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
w
i
t
h
G
P
T
-
3
.
5
,
”
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
A
n
n
u
a
l
M
e
e
t
i
n
g
o
f
t
h
e
A
s
s
o
c
i
a
t
i
o
n
f
o
r
Co
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
p
p
.
2
5
1
–
2
7
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
3
.
w
a
s
s
a
-
1
.
2
3
.
[3
1
]
D
.
D
e
v
i
,
S
.
K
.
Bi
s
w
a
s
,
a
n
d
B.
P
u
rk
a
y
a
s
t
h
a
,
“
A
re
v
i
e
w
o
n
s
o
l
u
t
i
o
n
t
o
c
l
a
s
s
i
m
b
a
l
a
n
c
e
p
ro
b
l
e
m
:
u
n
d
e
rs
a
m
p
l
i
n
g
a
p
p
ro
a
c
h
e
s
,
”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
e
r
e
n
c
e
o
n
Co
m
p
u
t
a
t
i
o
n
a
l
P
e
r
f
o
r
m
a
n
c
e
E
v
a
l
u
a
t
i
o
n
,
Co
m
P
E
2
0
2
0
,
2
0
2
0
,
p
p
.
6
2
6
–
6
3
1
,
d
o
i
:
1
0
.
1
1
0
9
/
Co
m
P
E
4
9
3
2
5
.
2
0
2
0
.
9
2
0
0
0
8
7
.
BI
O
G
R
A
P
H
I
ES
O
F
A
U
T
H
O
R
S
D
e
l
a
N
u
r
l
ai
l
a
i
s
a
g
r
a
d
ua
t
e
s
t
ude
n
t
a
t
B
i
na
N
us
a
nt
a
r
a
U
n
i
ve
r
s
i
t
y
,
pur
s
u
i
ng
a
m
a
s
t
e
r
’
s
de
g
r
e
e
i
n
c
o
m
p
ut
e
r
s
c
i
e
nc
e
.
S
he
c
ur
r
e
n
t
l
y
w
or
k
s
a
t
t
he
M
i
ni
s
t
r
y
o
f
F
or
e
i
gn
A
f
f
a
i
r
s
o
f
t
he
R
e
pub
l
i
c
of
I
nd
one
s
i
a
.
H
e
r
r
e
s
e
a
r
c
h
i
nt
e
r
e
s
t
s
i
nc
l
ud
e
t
h
e
a
ppl
i
c
a
t
i
on
of
a
r
t
i
f
i
c
i
a
l
i
nt
e
l
l
i
ge
n
c
e
a
nd
d
a
t
a
a
n
a
l
yt
i
c
s
i
n
di
pl
om
a
t
i
c
c
o
m
m
un
i
c
a
t
i
on
s
a
n
d
i
nt
e
r
na
t
i
ona
l
r
e
l
a
t
i
o
ns
.
S
he
c
a
n
be
c
o
nt
a
c
t
e
d
a
t
e
m
a
i
l
:
de
l
a
.
nu
r
l
a
i
l
a
@
bi
nus
.
a
c
.
i
d
.
A
b
b
a
S
u
gan
d
a
G
i
r
s
an
g
i
s
c
ur
r
e
nt
l
y
a
l
e
c
t
u
r
e
r
a
t
M
a
s
t
e
r
o
f
C
om
p
ut
e
r
S
c
i
e
n
c
e
,
B
i
na
N
us
a
nt
a
r
a
U
n
i
v
e
r
s
i
t
y
s
i
nc
e
20
15.
H
e
go
t
P
h.
D
.
i
n
t
he
I
ns
t
i
t
ut
e
o
f
C
om
pu
t
e
r
a
n
d
C
om
m
un
i
c
a
t
i
on
E
n
gi
ne
e
r
i
ng,
D
e
pa
r
t
m
e
nt
o
f
E
l
e
c
t
r
i
c
a
l
E
ng
i
ne
e
r
i
ng,
N
a
t
i
ona
l
C
he
n
g
K
u
ng
U
ni
ve
r
s
i
t
y,
T
a
i
n
a
n
,
T
a
i
w
a
n,
h
e
g
r
a
dua
t
e
d
b
a
c
h
e
l
o
r
f
r
o
m
t
h
e
D
e
pa
r
t
m
e
n
t
of
E
l
e
c
t
r
i
c
a
l
E
ngi
n
e
e
r
i
ng,
G
a
dj
a
h
M
a
da
U
n
i
v
e
r
s
i
t
y
(
U
G
M
)
,
Y
ogy
a
ka
r
t
a
,
I
nd
one
s
i
a
,
i
n
2000
.
H
e
t
he
n
c
ont
i
n
ue
d
h
i
s
m
a
s
t
e
r
’
s
d
e
gr
e
e
i
n
t
he
D
e
p
a
r
t
m
e
n
t
o
f
C
o
m
pu
t
e
r
S
c
i
e
nc
e
i
n
t
he
s
a
m
e
u
ni
v
e
r
s
i
t
y
i
n
200
6
–
20
08.
H
e
w
a
s
a
s
t
a
f
f
c
ons
ul
t
a
n
t
pr
ogr
a
m
m
e
r
i
n
B
e
t
he
s
d
a
H
o
s
p
i
t
a
l
,
Y
ogya
ka
r
t
a
,
i
n
2001
a
n
d
w
o
r
k
e
d
a
s
a
w
e
b
d
e
ve
l
o
pe
r
i
n
200
2
–
20
03
.
H
e
t
he
n
j
oi
ne
d
t
he
D
e
p
a
r
t
m
e
n
t
of
I
nf
or
m
a
t
i
c
s
E
ng
i
n
e
e
r
i
n
g
i
n
J
a
na
ba
d
r
a
U
ni
ve
r
s
i
t
y
a
s
a
l
e
c
t
ur
e
r
i
n
20
03
-
201
5.
H
e
c
a
n
be
c
ont
a
c
t
e
d
a
t
e
m
a
i
l
:
a
g
i
r
s
a
ng
@b
i
nu
s
.
e
du
.
Evaluation Warning : The document was created with Spire.PDF for Python.