I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
,
p
p
.
710
~
719
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
41
.
i
2
.
pp
7
1
0
-
7
1
9
710
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Integ
ra
ting co
ntr
a
stiv
e and g
enera
tive AI w
ith
RA
G
f
o
r
respo
nsibl
e and
f
a
ir CV
cla
ss
ifi
ca
ti
o
n
So
um
ia
Cha
f
i
,
M
us
t
a
ph
a
K
a
bil
,
Abde
s
s
a
m
a
d
K
a
m
o
us
s
La
b
o
r
a
t
o
r
y
o
f
M
a
t
h
e
ma
t
i
c
s,
C
o
mp
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 (L
M
C
S
A
)
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
s a
n
d
T
e
c
h
n
o
l
o
g
y
o
f
M
o
h
a
m
me
d
i
a
(
F
S
TM
)
,
H
a
ss
a
n
I
I
U
n
i
v
e
r
si
t
y
o
f
C
a
s
a
b
l
a
n
c
a
,
M
o
h
a
mm
e
d
i
a
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
8
,
2
0
2
5
R
ev
is
ed
J
an
1
,
2
0
2
6
Acc
ep
ted
J
an
16
,
2
0
2
6
Th
e
a
u
t
o
m
a
ti
o
n
o
f
c
u
rricu
l
u
m
v
it
a
e
(CV)
c
las
sifica
ti
o
n
ra
ise
s
m
a
jo
r
c
h
a
ll
e
n
g
e
s
re
late
d
to
a
c
c
u
ra
c
y
,
f
a
irn
e
ss
,
a
n
d
th
e
h
e
tero
g
e
n
e
it
y
o
f
c
a
n
d
id
a
te
d
o
c
u
m
e
n
ts.
Ex
isti
n
g
a
p
p
ro
a
c
h
e
s
o
ften
a
d
d
re
ss
t
h
e
se
d
ime
n
sio
n
s
s
e
p
a
ra
tely
a
n
d
str
u
g
g
le
t
o
re
d
u
c
e
d
e
m
o
g
ra
p
h
ic
b
ias
wh
il
e
m
a
in
tain
i
n
g
h
ig
h
p
re
d
ictiv
e
p
e
rf
o
rm
a
n
c
e
.
T
h
is
st
u
d
y
a
d
d
re
ss
e
s
th
is
g
a
p
b
y
p
r
o
p
o
sin
g
a
h
y
b
ri
d
p
ip
e
li
n
e
th
a
t
c
o
m
b
in
e
s
c
o
n
tras
ti
v
e
lea
rn
i
n
g
fo
r
re
p
re
se
n
tatio
n
wit
h
a
li
g
h
twe
ig
h
t
g
e
n
e
ra
ti
v
e
m
o
d
e
l
wit
h
in
a
re
tri
e
v
a
l
-
a
u
g
m
e
n
ted
g
e
n
e
ra
ti
o
n
(RAG
)
fra
m
e
wo
rk
.
Th
e
m
e
th
o
d
is
e
v
a
lu
a
ted
o
n
a
lar
g
e
d
a
ta
se
t
o
f
5
0
,
0
0
0
C
Vs
,
u
sin
g
sta
n
d
a
rd
c
las
sifica
ti
o
n
m
e
tri
c
s
a
s
we
ll
a
s
fa
irn
e
ss
in
d
ica
to
rs
b
a
se
d
o
n
re
d
u
c
ti
o
n
s
i
n
d
e
m
o
g
ra
p
h
ic
d
i
sp
a
rit
ies
a
n
d
e
q
u
a
li
ty
o
f
o
p
p
o
rt
u
n
it
y
.
Ex
p
e
rime
n
ts
sh
o
w
th
a
t
o
u
r
a
p
p
r
o
a
c
h
a
c
h
iev
e
s
a
n
a
c
c
u
ra
c
y
o
f
9
5
.
6
%
a
n
d
a
fa
irn
e
ss
in
d
e
x
o
f
0
.
9
4
,
re
d
u
c
i
n
g
g
e
n
d
e
r
-
re
late
d
d
isp
a
rit
ies
fro
m
4
.
8
%
t
o
0
.
3
%
.
Th
e
se
re
su
lt
s
d
e
m
o
n
stra
te
th
a
t
i
t
is
p
o
ss
i
b
le
to
sim
u
l
tan
e
o
u
sl
y
imp
ro
v
e
p
re
d
ictiv
e
p
e
rf
o
rm
a
n
c
e
a
n
d
fa
irn
e
ss
th
ro
u
g
h
a
m
u
lt
i
-
lev
e
l
fa
irn
e
ss
stra
teg
y
.
Th
e
p
r
o
p
o
se
d
sy
ste
m
th
u
s
re
p
re
se
n
ts
a
p
ra
c
ti
c
a
l
a
n
d
re
sp
o
n
sib
le
so
lu
ti
o
n
fo
r
i
n
teg
ra
ti
n
g
AI i
n
to
re
c
ru
it
m
e
n
t
p
r
o
c
e
ss
e
s.
K
ey
w
o
r
d
s
:
C
o
n
tr
asti
v
e
lear
n
in
g
C
V
clas
s
if
icatio
n
Fair
n
ess
Gen
er
ativ
e
lear
n
in
g
HR
I
S/S
I
R
H
R
AG
R
esp
o
n
s
ib
le
AI
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
:
So
u
m
ia
C
h
af
i
L
ab
o
r
ato
r
y
o
f
Ma
th
em
atics,
C
o
m
p
u
ter
Scien
ce
a
n
d
Ap
p
licatio
n
s
(
L
MCS
A)
Facu
lty
o
f
Scien
ce
s
an
d
T
ec
h
n
o
lo
g
y
o
f
Mo
h
am
m
ed
ia
(
FS
T
M)
Hass
an
I
I
Un
iv
er
s
ity
o
f
C
asab
lan
ca
Mo
h
am
m
ed
ia,
M
o
r
o
cc
o
E
m
ail:
s
o
u
m
ia.
ch
af
i@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
d
ig
italizatio
n
o
f
r
ec
r
u
it
m
en
t
p
r
o
ce
s
s
es
h
as
b
ee
n
ac
co
m
p
an
ied
b
y
th
e
g
r
o
win
g
ad
o
p
tio
n
o
f
ar
tific
ial
in
tellig
en
ce
(
AI
)
to
au
to
m
ate
ca
n
d
id
ate
p
r
e
-
s
cr
ee
n
in
g
an
d
class
if
icatio
n
.
Ho
wev
er
,
th
e
cu
r
r
icu
lu
m
v
itae
(
C
V)
,
wh
ich
r
e
m
ain
s
th
e
ce
n
tr
al
d
o
cu
m
en
t
in
th
e
ev
alu
atio
n
p
r
o
ce
s
s
,
p
r
esen
ts
s
ig
n
if
ican
t
co
m
p
lex
ity
d
u
e
to
th
e
d
iv
e
r
s
ity
o
f
f
o
r
m
ats,
th
e
h
ete
r
o
g
e
n
eity
o
f
wr
itin
g
s
ty
les,
an
d
th
e
in
cr
ea
s
in
g
p
r
es
en
ce
o
f
m
u
ltimo
d
al
elem
en
ts
(
lo
g
o
s
,
g
r
ap
h
ics,
an
d
p
h
o
to
g
r
a
p
h
s
)
.
T
h
is
v
ar
ia
b
ilit
y
m
ak
es
it
d
if
f
icu
lt
to
ap
p
l
y
tr
ad
itio
n
al
tex
t
-
class
if
icatio
n
m
eth
o
d
s
,
wh
ich
s
tr
u
g
g
le
to
ca
p
t
u
r
e
th
e
c
o
n
tex
t
u
al
an
d
s
em
an
tic
r
ic
h
n
ess
o
f
C
Vs [
1
]
.
R
ec
en
t
ad
v
an
ce
s
in
co
n
tr
asti
v
e
lear
n
in
g
an
d
g
e
n
er
ativ
e
m
o
d
els
h
av
e
m
ar
k
ed
a
m
ajo
r
b
r
e
ak
th
r
o
u
g
h
in
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
.
Ap
p
r
o
ac
h
es
s
u
ch
as
Si
m
C
SE
an
d
C
o
n
tr
iev
er
p
r
o
d
u
ce
r
o
b
u
s
t
s
em
an
tic
r
ep
r
esen
tatio
n
s
th
at
im
p
r
o
v
e
s
im
ilar
ity
m
atch
in
g
b
etwe
en
C
Vs
an
d
jo
b
d
escr
ip
tio
n
s
,
wh
ile
g
en
er
ativ
e
m
o
d
els
s
u
ch
as
L
L
aM
A
en
r
ich
t
h
e
co
n
tex
t
b
y
g
e
n
er
atin
g
s
u
m
m
ar
ie
s
o
r
r
ele
v
an
t
co
m
p
lem
en
tar
y
i
n
f
o
r
m
atio
n
[
2
]
,
[
3
]
.
T
h
ese
m
eth
o
d
s
s
u
r
p
ass
tr
ad
itio
n
al
ap
p
r
o
ac
h
es
(
T
F
-
I
DF,
SVM)
in
b
o
th
ac
c
u
r
ac
y
a
n
d
th
ei
r
ab
ilit
y
to
h
an
d
le
u
n
s
tr
u
ctu
r
ed
tex
t [
4
]
.
Ho
wev
er
,
b
ey
o
n
d
p
er
f
o
r
m
a
n
c
e,
a
cr
u
cial
ch
allen
g
e
r
em
ain
s
:
en
s
u
r
in
g
th
e
r
esp
o
n
s
ib
ilit
y
an
d
f
air
n
ess
o
f
AI
ap
p
lied
t
o
r
ec
r
u
itm
e
n
t.
R
ec
en
t
s
tu
d
ies
h
ig
h
lig
h
t
th
at
m
o
d
els
tr
ain
ed
o
n
b
iased
d
ata
m
ay
r
ep
r
o
d
u
ce
o
r
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
I
n
teg
r
a
tin
g
c
o
n
tr
a
s
tive
a
n
d
g
e
n
era
tive
A
I
w
ith
R
A
G
fo
r
r
esp
o
n
s
ib
le
…
(
S
o
u
mia
C
h
a
fi
)
711
ev
en
am
p
lify
d
is
cr
im
in
atio
n
r
elate
d
to
g
e
n
d
er
,
ag
e,
o
r
o
r
ig
i
n
[
5
]
,
[
6
]
.
I
n
s
u
c
h
a
s
en
s
itiv
e
d
o
m
ain
as
h
u
m
an
r
es
o
u
r
ce
s
,
it
is
th
er
e
f
o
r
e
ess
en
tial
to
d
esig
n
ap
p
r
o
ac
h
es
th
at
en
s
u
r
e
ef
f
icien
cy
,
t
r
an
s
p
ar
en
c
y
,
an
d
alg
o
r
ith
m
ic
f
air
n
ess
.
Alth
o
u
g
h
m
an
y
s
tu
d
ies
h
a
v
e
i
n
v
esti
g
ated
au
to
m
atic
C
V
class
if
icatio
n
,
m
o
s
t
ap
p
r
o
ac
h
es
f
o
cu
s
eith
e
r
o
n
im
p
r
o
v
i
n
g
r
ep
r
esen
tatio
n
m
o
d
els
o
r
o
n
r
e
d
u
cin
g
b
ias
—
b
u
t
r
ar
ely
o
n
b
o
th
s
im
u
ltan
eo
u
s
ly
.
M
o
r
eo
v
er
,
ex
is
tin
g
s
o
lu
tio
n
s
g
en
er
ally
f
ail
to
ac
co
u
n
t
f
o
r
th
e
h
eter
o
g
en
eity
o
f
m
o
d
er
n
C
Vs,
wh
ich
m
ay
co
m
b
i
n
e
tex
t,
im
ag
es,
tab
les,
an
d
o
th
er
v
is
u
al
elem
en
ts
.
R
esear
ch
o
n
f
air
n
ess
in
r
ec
r
u
itm
en
t
r
em
ain
s
li
m
ited
an
d
o
f
ten
f
o
cu
s
es
o
n
a
s
in
g
le
s
en
s
itiv
e
attr
ib
u
te,
with
o
u
t
o
f
f
er
in
g
a
co
m
p
r
e
h
en
s
iv
e
m
itig
atio
n
f
r
am
ewo
r
k
.
T
o
o
u
r
k
n
o
wled
g
e,
n
o
p
r
io
r
wo
r
k
h
as
co
m
b
in
e
d
c
o
n
tr
asti
v
e
le
ar
n
in
g
,
r
etr
iev
al
-
a
u
g
m
e
n
ted
g
en
er
atio
n
(
R
AG)
,
m
u
ltimo
d
ality
,
an
d
f
air
n
ess
with
in
a
u
n
if
ied
p
i
p
elin
e.
T
h
is
wo
r
k
aim
s
to
f
ill
th
is
g
ap
b
y
p
r
o
p
o
s
in
g
an
in
teg
r
ated
ap
p
r
o
ac
h
ca
p
a
b
le
o
f
s
im
u
ltan
eo
u
s
ly
im
p
r
o
v
in
g
p
er
f
o
r
m
a
n
ce
,
tr
an
s
p
ar
en
c
y
,
an
d
f
air
n
ess
in
C
V
class
if
icatio
n
.
T
h
is
s
tu
d
y
b
u
ild
s
u
p
o
n
o
u
r
p
r
ev
io
u
s
co
n
tr
ib
u
tio
n
s
[
7
]
,
wh
ile
p
a
v
in
g
th
e
wa
y
to
war
d
r
esp
o
n
s
ib
le
an
d
f
air
AI
f
o
r
r
ec
r
u
itm
en
t.
2.
RE
L
AT
E
D
WO
RK
S
T
h
e
au
to
m
atic
class
if
icatio
n
o
f
C
Vs
f
alls
w
ith
in
th
e
b
r
o
ad
e
r
r
esear
ch
ar
ea
o
f
tex
t
class
if
icatio
n
an
d
AI
ap
p
lied
to
h
u
m
a
n
r
eso
u
r
ce
s
.
T
r
ad
itio
n
al
ap
p
r
o
ac
h
es,
b
ased
o
n
s
tatis
tical
r
ep
r
esen
tati
o
n
s
s
u
ch
as
T
F
-
I
DF
co
m
b
in
ed
with
class
if
ier
s
lik
e
SVMs,
h
av
e
lo
n
g
d
o
m
in
ated
th
e
f
ield
.
Ho
wev
er
,
th
ese
m
et
h
o
d
s
r
e
m
ain
lim
ited
in
th
eir
ab
ilit
y
to
ca
p
tu
r
e
c
o
n
t
ex
tu
al
s
u
b
tleties
an
d
co
m
p
lex
s
em
an
tic
r
elatio
n
s
h
ip
s
[
1
]
.
T
h
e
em
er
g
en
ce
o
f
p
r
et
r
ain
ed
l
an
g
u
ag
e
m
o
d
e
ls
(
PLM
s
)
an
d
,
m
o
r
e
r
ec
en
tly
,
L
ar
g
e
L
an
g
u
ag
e
Mo
d
els
(
L
L
Ms)
,
h
as
p
r
o
f
o
u
n
d
ly
tr
an
s
f
o
r
m
ed
th
is
d
o
m
ain
.
Mo
d
els
s
u
ch
as
B
E
R
T
,
R
o
B
E
R
T
a,
an
d
L
L
aM
A
en
ab
le
r
ich
er
co
n
tex
tu
al
u
n
d
er
s
tan
d
in
g
an
d
o
p
e
n
th
e
d
o
o
r
to
ze
r
o
-
s
h
o
t
an
d
f
ew
-
s
h
o
t
s
tr
ateg
ies
s
u
itab
le
f
o
r
u
n
s
tr
u
ctu
r
ed
d
ata
s
u
ch
as
C
Vs
[
3
]
,
[
8
]
.
Mo
r
e
o
v
er
,
r
ec
en
t
s
u
r
v
ey
s
tu
d
ies
co
n
f
ir
m
th
e
ir
s
u
p
er
io
r
ity
o
v
er
class
ical
m
eth
o
d
s
wh
ile
h
ig
h
lig
h
tin
g
c
h
allen
g
es r
elate
d
to
c
o
m
p
u
tatio
n
al
co
s
t a
n
d
d
o
m
ain
a
d
ap
tatio
n
[
4
]
.
C
o
n
tr
asti
v
e
lear
n
in
g
h
as
b
ec
o
m
e
a
k
ey
ad
v
an
ce
m
e
n
t.
Mo
d
els
s
u
ch
as
SimC
SE
an
d
C
o
n
tr
iev
er
p
r
o
d
u
ce
r
o
b
u
s
t
s
em
an
tic
r
ep
r
esen
tatio
n
s
th
at
ar
e
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
f
o
r
te
x
t
s
im
ilar
ity
task
s
an
d
C
V
-
jo
b
m
atch
in
g
[
8
]
.
E
x
te
n
s
io
n
s
s
u
ch
as
SimCS
E
++
f
u
r
th
er
s
tr
en
g
t
h
en
r
ep
r
esen
tatio
n
s
tab
ilit
y
an
d
g
e
n
er
aliza
tio
n
i
n
d
iv
er
s
e
co
n
tex
ts
.
I
n
p
a
r
allel,
g
en
er
ativ
e
lear
n
in
g
with
m
o
d
els
s
u
ch
as
L
L
aM
A
en
r
ich
es
r
ep
r
esen
tatio
n
s
b
y
p
r
o
d
u
cin
g
co
n
te
x
tu
alize
d
s
u
m
m
ar
ies
o
f
C
Vs
an
d
f
illi
n
g
m
is
s
in
g
in
f
o
r
m
atio
n
.
T
h
ese
m
o
d
els
ar
e
esp
ec
ially
u
s
ef
u
l
f
o
r
s
k
ills
n
o
r
m
aliza
tio
n
o
r
f
o
r
g
en
er
atin
g
ca
n
d
i
d
ate
p
r
o
f
iles
th
at
ca
n
b
e
in
teg
r
ate
d
in
to
an
HR
I
S
[
3
]
.
T
h
e
in
teg
r
atio
n
o
f
r
etr
iev
al
-
au
g
m
en
te
d
g
en
er
atio
n
(
R
AG)
,
wh
ich
co
m
b
in
es
co
n
tex
tu
al
r
etr
iev
al
with
g
en
er
atio
n
,
h
as
also
p
r
o
v
en
to
b
e
a
p
o
wer
f
u
l
lev
er
f
o
r
im
p
r
o
v
in
g
th
e
ac
cu
r
ac
y
an
d
r
eliab
ilit
y
o
f
AI
s
y
s
t
em
s
ap
p
lied
to
h
u
m
an
r
eso
u
r
ce
s
[
9
]
.
An
o
th
er
m
ajo
r
r
esear
ch
d
ir
ec
t
io
n
co
n
ce
r
n
s
b
ias
m
itig
atio
n
an
d
r
esp
o
n
s
ib
le
AI
.
R
ec
en
t
w
o
r
k
s
h
o
ws
th
at
class
if
icat
io
n
m
o
d
els
tr
ain
ed
o
n
b
iased
d
ata
m
ay
r
e
p
r
o
d
u
ce
o
r
ev
en
am
p
lify
d
is
cr
im
i
n
atio
n
,
p
ar
ticu
lar
l
y
r
elate
d
to
g
e
n
d
er
an
d
eth
n
icit
y
[
6
]
.
Var
io
u
s
m
itig
atio
n
s
tr
at
eg
ies
h
av
e
b
ee
n
p
r
o
p
o
s
ed
:
d
at
a
r
eb
alan
cin
g
(
p
r
e
-
p
r
o
ce
s
s
in
g
)
,
co
n
s
tr
ain
ed
lear
n
in
g
(
in
-
p
r
o
ce
s
s
in
g
)
,
a
n
d
d
ec
is
io
n
th
r
esh
o
ld
ca
lib
r
atio
n
(
p
o
s
t
-
p
r
o
ce
s
s
in
g
)
.
T
h
ese
m
u
lti
-
lev
el
ap
p
r
o
ac
h
es
p
av
e
th
e
way
f
o
r
f
air
er
an
d
m
o
r
e
tr
a
n
s
p
ar
en
t
r
ec
r
u
itm
en
t
s
y
s
tem
s
.
T
h
u
s
,
th
e
liter
atu
r
e
co
n
v
er
g
es
to
war
d
th
e
id
ea
o
f
h
y
b
r
id
p
ip
elin
es
co
m
b
in
i
n
g
ad
v
a
n
ce
d
class
if
icatio
n
,
co
n
tr
asti
v
e
lear
n
in
g
,
g
en
er
ativ
e
m
o
d
els,
a
n
d
f
air
n
es
s
m
ec
h
an
is
m
s
.
Ou
r
wo
r
k
f
o
llo
ws
th
is
d
ir
ec
tio
n
b
y
p
r
o
p
o
s
in
g
a
m
u
l
tim
o
d
al
a
n
d
r
esp
o
n
s
ib
le
ap
p
r
o
ac
h
i
n
teg
r
ate
d
in
to
HR
I
S.
Sev
er
al
p
r
io
r
s
tu
d
ies
h
av
e
im
p
r
o
v
e
d
d
o
cu
m
e
n
t
class
if
icatio
n
b
u
t
r
em
ain
f
r
ag
m
en
ted
an
d
i
n
s
u
f
f
icien
t
f
o
r
s
en
s
itiv
e
HR
ap
p
licatio
n
s
.
So
m
e
wo
r
k
s
p
r
o
p
o
s
e
d
ee
p
l
ea
r
n
in
g
a
r
ch
itectu
r
es
ap
p
lie
d
to
C
Vs
[
1
0
]
,
wh
i
le
o
th
er
s
f
o
cu
s
o
n
m
u
ltimo
d
al
d
o
cu
m
en
t u
n
d
er
s
tan
d
in
g
[
1
1
]
.
M
o
r
e
r
ec
en
t a
p
p
r
o
ac
h
es h
ig
h
lig
h
t th
e
u
s
ef
u
ln
ess
o
f
g
en
er
ativ
e
m
o
d
els
f
o
r
co
m
p
letin
g
m
is
s
in
g
in
f
o
r
m
atio
n
[
1
2
]
,
an
d
ad
d
itio
n
al
s
tu
d
i
es
em
p
h
asize
th
e
co
n
tr
ib
u
tio
n
o
f
co
n
tr
asti
v
e
lea
r
n
in
g
in
s
tr
en
g
th
e
n
in
g
th
e
r
o
b
u
s
tn
ess
o
f
r
ep
r
esen
tatio
n
s
[
1
3
]
.
Mo
r
eo
v
er
,
s
ev
er
al
an
aly
s
es
s
h
o
w
th
at
au
to
m
ated
r
ec
r
u
itm
en
t
s
y
s
tem
s
ca
n
am
p
lify
d
is
cr
im
in
atio
n
wh
en
n
o
f
air
n
ess
m
ec
h
an
is
m
is
in
teg
r
ated
[
1
4
]
.
I
n
co
n
tr
ast to
th
ese
is
o
lated
ap
p
r
o
ac
h
es,
o
u
r
u
n
if
ied
p
ip
elin
e
co
m
b
in
es
co
n
tr
asti
v
e
lear
n
in
g
,
g
en
er
ativ
e
m
o
d
elin
g
,
R
AG,
an
d
f
air
n
ess
with
in
a
co
h
er
e
n
t
ar
ch
itectu
r
e,
p
r
o
v
i
d
in
g
a
m
o
r
e
co
m
p
lete
an
d
r
esp
o
n
s
ib
le
s
o
lu
tio
n
f
o
r
C
V
class
if
icatio
n
.
T
r
ad
itio
n
al
C
V
class
if
icatio
n
ap
p
r
o
ac
h
es
b
ased
o
n
T
F
-
I
DF
an
d
SVM
s
u
f
f
er
f
r
o
m
a
lim
ite
d
ab
ilit
y
to
ca
p
tu
r
e
th
e
c
o
m
p
lex
s
em
an
tic
r
elatio
n
s
h
ip
s
p
r
esen
t
in
ca
n
d
id
ate
d
o
cu
m
e
n
ts
[
4
]
.
T
r
an
s
f
o
r
m
er
-
b
ased
m
o
d
els
s
u
ch
as
B
E
R
T
an
d
R
o
B
E
R
T
a
h
av
e
b
r
o
u
g
h
t
s
u
b
s
tan
tial
im
p
r
o
v
em
en
ts
b
u
t
r
em
ain
s
en
s
itiv
e
to
th
e
v
a
r
iab
ilit
y
o
f
C
V
s
tr
u
ctu
r
es
an
d
th
e
q
u
ali
ty
o
f
ex
tr
ac
ted
tex
t
[
1
]
.
C
o
n
tr
asti
v
e
m
eth
o
d
s
(
SimCS
E
,
C
o
n
tr
iev
er
)
h
av
e
p
r
o
v
e
n
ef
f
ec
tiv
e
f
o
r
C
V
–
jo
b
s
im
ilar
ity
[
1
5
]
,
wh
ile
m
u
ltim
o
d
al
m
o
d
els
lik
e
C
L
I
P
ar
e
o
n
ly
b
e
g
in
n
in
g
t
o
b
e
ex
p
lo
r
ed
f
o
r
tex
t
–
im
ag
e
f
u
s
io
n
[
1
6
]
.
Alth
o
u
g
h
s
ev
er
al
s
tu
d
ies
h
av
e
h
ig
h
lig
h
ted
t
h
e
n
ee
d
to
r
ed
u
ce
b
ias
i
n
au
to
m
ated
r
ec
r
u
itm
en
t
s
y
s
te
m
s
[
6
]
,
a
n
d
tec
h
n
iq
u
es
s
u
ch
as
d
ata
r
ewe
ig
h
tin
g
[
1
7
]
,
f
air
n
ess
-
awa
r
e
r
eg
u
lar
izatio
n
[
1
8
]
,
o
r
ad
v
er
s
a
r
ial
m
eth
o
d
s
[
19
]
h
a
v
e
b
ee
n
p
r
o
p
o
s
ed
to
a
d
d
r
ess
th
ese
d
is
p
ar
ities
,
f
ew
s
tu
d
ies
o
f
f
er
a
co
m
p
r
eh
en
s
iv
e
i
n
teg
r
a
tio
n
o
f
f
air
n
ess
with
in
a
co
m
p
lete
p
ip
elin
e.
T
h
ese
a
p
p
r
o
ac
h
e
s
ar
e
o
f
ten
a
p
p
lied
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
7
1
0
-
7
1
9
712
in
is
o
latio
n
an
d
r
ar
ely
ad
ap
t
ed
to
th
e
HR
co
n
tex
t.
I
n
c
o
n
tr
ast,
o
u
r
wo
r
k
co
m
b
in
es,
f
o
r
t
h
e
f
ir
s
t
tim
e,
co
n
tr
asti
v
e
lear
n
in
g
,
g
e
n
er
ati
o
n
,
m
u
ltimo
d
ality
,
an
d
R
AG
with
a
m
u
lt
i
-
lev
el
f
air
n
ess
m
o
d
u
le,
p
r
o
v
id
i
n
g
a
h
o
lis
tic
s
o
lu
tio
n
f
o
r
s
im
u
ltan
e
o
u
s
ly
im
p
r
o
v
in
g
p
er
f
o
r
m
an
ce
an
d
tr
an
s
p
ar
e
n
cy
in
C
V
class
if
icatio
n
.
3.
M
E
T
H
O
DO
L
O
G
Y
AND
P
R
O
P
O
SE
D
AP
P
RO
ACH
T
h
e
m
eth
o
d
o
lo
g
y
is
b
ase
d
o
n
a
h
y
b
r
id
p
ip
elin
e
co
m
b
in
in
g
p
r
e
p
r
o
ce
s
s
in
g
,
m
u
ltimo
d
al
r
ep
r
esen
tatio
n
,
g
e
n
er
ativ
e
e
n
r
ich
m
en
t,
an
d
alg
o
r
ith
m
ic
f
air
n
ess
.
T
h
e
p
r
ep
r
o
ce
s
s
in
g
s
tag
e
p
r
ep
ar
es
h
eter
o
g
en
e
o
u
s
C
Vs
(
tex
t,
s
c
an
s
,
v
is
u
al
elem
en
ts
)
u
s
in
g
OC
R
,
n
o
r
m
aliza
tio
n
,
s
eg
m
en
tatio
n
,
an
d
p
ar
tial
an
o
n
y
m
izatio
n
in
o
r
d
er
to
r
ed
u
ce
th
e
im
p
ac
t
o
f
s
en
s
itiv
e
attr
ib
u
tes at
th
e
s
o
u
r
ce
(
Fig
u
r
e
1
)
.
T
h
e
f
o
u
r
m
o
d
u
les
o
f
th
e
p
ip
el
in
e
ar
e
in
ter
d
ep
e
n
d
en
t,
an
d
ea
ch
p
lay
s
an
ess
en
tial
r
o
le:
r
em
o
v
in
g
an
y
o
n
e
o
f
th
em
lea
d
s
to
a
s
ig
n
if
ican
t
d
ec
r
ea
s
e
in
o
v
e
r
all
q
u
ality
.
W
ith
o
u
t
th
e
co
n
tr
asti
v
e
m
o
d
u
l
e
(
SimCS
E
+
c
o
n
tr
iev
er
)
,
th
e
tex
tu
al
r
ep
r
esen
tatio
n
s
b
ec
o
m
e
l
ess
r
o
b
u
s
t
an
d
less
d
is
cr
im
in
a
tiv
e,
wh
ich
d
ir
ec
tly
d
eg
r
ad
es
th
e
ac
cu
r
ac
y
o
f
C
V
–
jo
b
m
atch
in
g
.
W
ith
o
u
t
t
h
e
g
e
n
er
ativ
e
m
o
d
u
le
(
L
L
aM
A)
,
co
n
tex
tu
al
en
r
ich
m
e
n
t
an
d
th
e
n
o
r
m
aliza
tio
n
o
f
h
et
er
o
g
en
e
o
u
s
C
Vs
d
is
ap
p
ea
r
,
r
esu
ltin
g
in
m
o
r
e
er
r
o
r
s
f
o
r
in
co
m
p
lete
o
r
p
o
o
r
l
y
s
tr
u
ctu
r
ed
p
r
o
f
iles
.
T
h
e
ab
s
en
ce
o
f
th
e
R
AG
m
o
d
u
le
r
em
o
v
es
th
e
g
r
o
u
n
d
in
g
in
r
ea
l
s
k
ills
d
ata,
in
cr
ea
s
in
g
m
o
d
el
h
allu
cin
atio
n
s
an
d
r
e
d
u
cin
g
th
e
co
h
er
en
ce
o
f
p
r
e
d
ictio
n
s
.
Fin
ally
,
with
o
u
t
th
e
f
air
n
ess
m
o
d
u
le,
d
em
o
g
r
a
p
h
ic
b
iases
re
-
em
er
g
e,
p
ar
ticu
lar
l
y
a
g
ain
s
t
wo
m
e
n
,
an
d
t
h
e
s
y
s
tem
ca
n
n
o
l
o
n
g
er
b
e
c
o
n
s
id
er
ed
r
esp
o
n
s
ib
le
o
r
s
u
itab
le
f
o
r
r
e
al
-
wo
r
ld
HR
ap
p
licatio
n
s
.
T
h
u
s
,
ea
ch
m
o
d
u
le
c
o
n
tr
ib
u
tes
i
n
an
in
d
is
p
en
s
ab
le
way
to
th
e
p
ip
elin
e’
s
p
er
f
o
r
m
an
ce
,
r
o
b
u
s
tn
ess
,
an
d
a
lg
o
r
ith
m
ic
f
air
n
ess
,
an
d
r
em
o
v
in
g
a
n
y
o
f
th
em
s
u
b
s
tan
tially
wea
k
en
s
th
e
en
tire
s
y
s
tem
.
R
etr
iev
a
l
-
au
g
m
en
ted
g
en
er
ati
o
n
(
R
AG
)
is
a
tech
n
iq
u
e
t
h
at
c
o
m
b
in
es:
-
R
etr
iev
al
→
r
etr
iev
in
g
in
f
o
r
m
atio
n
f
r
o
m
an
e
x
ter
n
al
k
n
o
wle
d
g
e
b
ase
(
d
o
cu
m
e
n
ts
,
d
atab
as
es,
etc.
)
.
-
Au
g
m
en
ted
g
en
er
atio
n
→
a
L
L
M
s
u
ch
as
L
L
aM
A
u
s
es
th
is
in
f
o
r
m
atio
n
to
g
en
e
r
ate
a
m
o
r
e
p
r
ec
is
e
an
d
co
n
tex
tu
alize
d
o
u
tp
u
t.
I
n
o
u
r
p
ip
elin
e,
R
AG
is
u
s
ed
ex
clu
s
iv
ely
to
en
r
ich
th
e
v
e
cto
r
d
atab
ase
with
th
e
co
m
p
a
n
y
’
s
s
k
ills
r
ep
o
s
ito
r
y
a
n
d
t
o
im
p
r
o
v
e
t
h
e
r
elev
an
ce
o
f
C
V
-
jo
b
m
atch
in
g
.
On
ce
th
e
m
o
s
t
s
u
itab
le
p
r
o
f
iles
ar
e
id
en
tifie
d
th
r
o
u
g
h
th
is
o
p
tim
ized
r
etr
iev
al,
th
e
f
air
n
ess
m
o
d
u
le
o
p
er
at
es
d
o
wn
s
tr
ea
m
to
an
aly
ze
an
d
co
r
r
ec
t
p
o
te
n
tial
d
em
o
g
r
a
p
h
ic
d
is
p
ar
ities
.
T
h
u
s
,
R
AG
im
p
r
o
v
es
s
elec
tio
n
q
u
ality
,
wh
ile
th
e
f
air
n
ess
s
tag
e
e
n
s
u
r
es
n
eu
tr
ality
in
th
e
f
in
al
d
ec
is
io
n
s
.
Fig
u
r
e
1
.
Diag
r
a
m
o
f
th
e
class
if
icatio
n
p
ip
elin
e
Fin
ally
,
a
m
u
lti
-
lev
el
f
air
n
ess
m
o
d
u
le
is
in
teg
r
ate
d
,
r
el
y
in
g
o
n
th
r
ee
c
o
m
p
lem
e
n
tar
y
s
tag
es.
Du
r
in
g
p
r
ep
r
o
ce
s
s
in
g
,
class
r
eb
alan
c
in
g
an
d
co
u
n
ter
f
ac
t
u
al
C
V
g
en
er
atio
n
d
iv
e
r
s
if
y
s
en
s
itiv
e
ex
am
p
les
wh
ile
p
r
eser
v
in
g
p
r
o
f
ess
io
n
al
s
k
ills
.
Du
r
in
g
tr
ain
in
g
,
we
u
s
e
a
r
eg
u
lar
ized
lo
s
s
f
u
n
ctio
n
t
h
at
in
co
r
p
o
r
ates
a
d
em
o
g
r
a
p
h
ic
p
a
r
ity
co
n
s
tr
ain
t:
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
I
n
teg
r
a
tin
g
c
o
n
tr
a
s
tive
a
n
d
g
e
n
era
tive
A
I
w
ith
R
A
G
fo
r
r
esp
o
n
s
ib
le
…
(
S
o
u
mia
C
h
a
fi
)
713
ℒ
=
ℒ
+
⋋
|
1
−
2
|
with
λ
=
0
.
1
,
to
p
e
n
alize
p
er
f
o
r
m
an
ce
d
is
p
ar
ities
b
etwe
en
g
r
o
u
p
s
.
Fin
ally
,
in
th
e
p
o
s
t
-
p
r
o
ce
s
s
in
g
s
tag
e,
a
th
r
esh
o
ld
ca
lib
r
atio
n
in
s
p
ir
e
d
b
y
E
q
u
alize
d
Od
d
s
a
d
ju
s
ts
th
e
f
in
al
p
r
o
b
a
b
ilit
ies
s
e
p
ar
ately
to
r
ed
u
ce
d
if
f
er
en
ce
s
in
f
alse
p
o
s
itiv
es
an
d
f
alse
n
eg
ati
v
es
ac
r
o
s
s
d
em
o
g
r
ap
h
ic
g
r
o
u
p
s
.
T
h
is
co
m
b
in
ed
m
ec
h
a
n
is
m
r
ed
u
ce
s
b
ias wh
ile
p
r
eser
v
i
n
g
h
ig
h
o
v
er
all
ac
cu
r
ac
y
.
-
Pre
-
p
r
o
ce
s
s
in
g
:
d
ata
r
eb
ala
n
cin
g
an
d
c
o
u
n
ter
f
ac
tu
al
g
en
e
r
at
io
n
(
e.
g
.
,
cr
ea
tin
g
s
y
n
t
h
etic
C
Vs
th
at
m
o
d
if
y
g
en
d
er
o
r
o
r
ig
in
with
o
u
t a
lter
i
n
g
p
r
o
f
ess
io
n
al
s
k
ills
)
.
-
In
-
p
r
o
ce
s
s
in
g
:
co
n
s
tr
ain
ed
le
ar
n
in
g
th
r
o
u
g
h
f
air
n
ess
-
awa
r
e
em
b
ed
d
in
g
s
an
d
ad
v
e
r
s
ar
ial
p
en
alties
to
r
ed
u
ce
p
r
ed
ictio
n
g
ap
s
b
etwe
e
n
d
em
o
g
r
ap
h
ic
g
r
o
u
p
s
.
-
Po
s
t
-
p
r
o
ce
s
s
in
g
:
d
ec
is
io
n
t
h
r
e
s
h
o
ld
ca
lib
r
atio
n
an
d
p
r
o
b
ab
il
ity
ad
ju
s
tm
en
t
t
o
en
s
u
r
e
d
e
m
o
g
r
ap
h
ic
p
a
r
ity
an
d
eq
u
al
o
p
p
o
r
tu
n
ity
.
T
h
is
m
o
d
u
le
m
itig
ates
g
en
d
e
r
-
r
elate
d
b
iases
.
T
h
e
p
ip
elin
e
is
in
teg
r
ate
d
in
t
o
a
n
HR
in
f
o
r
m
atio
n
s
y
s
tem
v
ia
API
,
p
r
o
v
id
in
g
cl
ass
if
icatio
n
,
m
atch
in
g
s
co
r
es,
g
en
er
ated
s
u
m
m
ar
ies,
an
d
f
a
ir
n
ess
in
d
icato
r
s
,
in
ac
co
r
d
an
ce
with
th
e
p
r
in
cip
le
s
o
f
r
esp
o
n
s
ib
le,
tr
an
s
p
a
r
en
t,
an
d
ex
p
lain
ab
le
AI
ad
a
p
ted
to
d
iv
er
s
e
r
ec
r
u
itm
en
t
co
n
tex
ts
.
T
h
e
ev
alu
atio
n
r
elies
o
n
s
tan
d
a
r
d
class
if
icatio
n
m
etr
ics
(
ac
cu
r
ac
y
,
F1
-
s
co
r
e
,
p
r
ec
is
io
n
,
r
ec
all)
as
well
as f
air
n
ess
in
d
icato
r
s
.
4.
E
XP
E
R
I
M
E
N
T
S
T
h
e
d
ataset
u
s
ed
in
o
u
r
ex
p
e
r
im
en
t
s
co
n
s
is
ts
o
f
5
0
,
0
0
0
C
Vs
an
d
1
0
0
jo
b
o
f
f
e
r
s
s
o
u
r
ce
d
f
r
o
m
an
in
ter
n
al
d
atab
ase,
s
u
p
p
lem
en
t
ed
b
y
a
class
if
icatio
n
d
etail
f
ile
o
b
tain
ed
f
r
o
m
o
p
en
-
d
ata
j
o
b
p
latf
o
r
m
s
.
T
h
e
an
n
o
tatio
n
o
f
th
ese
C
Vs
was
ca
r
r
ied
o
u
t
b
y
HR
ex
p
er
ts
.
T
h
e
d
em
o
g
r
ap
h
ic
d
is
tr
ib
u
tio
n
o
f
th
e
C
Vs
in
clu
d
es
5
4
%
m
ale
an
d
4
6
%
f
em
ale
c
an
d
id
ates.
I
n
ter
m
s
o
f
d
o
cu
m
en
t
ty
p
es,
th
e
d
ataset
co
n
tain
s
6
2
%
tex
tu
al
C
V
s
,
2
4
%
s
ca
n
n
ed
C
Vs
r
eq
u
ir
in
g
OC
R
ex
tr
ac
tio
n
,
an
d
1
4
%
co
n
tain
in
g
v
is
u
al
elem
en
ts
(
lo
g
o
s
,
g
r
ap
h
ics,
ico
n
s
)
.
Sen
s
itiv
e
attr
ib
u
tes
(
n
am
e,
ag
e,
p
h
o
to
,
a
d
d
r
ess
)
wer
e
r
em
o
v
ed
o
r
r
ep
lace
d
with
n
eu
tr
al
to
k
en
s
in
ac
co
r
d
a
n
ce
with
p
r
iv
ac
y
b
est
p
r
ac
tices.
T
h
e
d
ata
we
r
e
s
p
lit
in
to
8
0
%
tr
ain
in
g
,
1
0
%
v
alid
atio
n
,
an
d
1
0
%
test
in
g
.
T
h
is
lev
el
o
f
tr
an
s
p
a
r
en
cy
is
ess
en
tial f
o
r
en
s
u
r
i
n
g
r
e
p
r
o
d
u
c
ib
ilit
y
an
d
en
a
b
lin
g
cr
itical
ev
alu
atio
n
o
f
th
e
ap
p
r
o
ac
h
.
T
o
f
u
r
th
er
e
n
h
an
ce
p
r
e
p
r
o
ce
s
s
in
g
tr
an
s
p
ar
e
n
cy
,
we
ev
alu
ate
d
th
e
q
u
ality
o
f
th
e
d
if
f
er
e
n
t
o
p
er
atio
n
s
.
T
h
e
OC
R
co
m
p
o
n
en
t
ac
h
iev
e
d
an
av
er
ag
e
ac
cu
r
ac
y
o
f
9
6
.
8
%
o
n
a
s
am
p
le
o
f
5
0
0
s
ca
n
n
ed
C
Vs,
en
s
u
r
in
g
r
eliab
le
tex
t
ex
tr
ac
tio
n
f
o
r
s
u
b
s
eq
u
en
t
s
tep
s
.
R
eg
ar
d
in
g
a
n
o
n
y
m
izatio
n
,
we
m
ea
s
u
r
ed
a
d
etec
tio
n
r
ate
o
f
9
8
.
1
%
f
o
r
s
en
s
itiv
e
en
titi
es
(
n
am
e,
ag
e,
ad
d
r
ess
,
p
h
o
n
e
n
u
m
b
er
)
,
with
an
er
r
o
r
r
ate
b
elo
w
1
%.
T
h
ese
r
esu
lts
co
n
f
ir
m
th
at
th
e
p
ip
elin
e
o
p
er
ates
o
n
n
o
r
m
al
ized
a
n
d
p
r
o
p
er
ly
an
o
n
y
m
ized
d
ata
p
r
io
r
to
r
ep
r
esen
tatio
n
,
th
er
eb
y
lim
itin
g
t
h
e
in
tr
o
d
u
cti
o
n
o
f
b
ias r
elate
d
to
p
er
s
o
n
al
i
n
f
o
r
m
atio
n
.
E
ac
h
C
V
in
th
e
d
ataset
is
ass
o
ciate
d
with
a
class
if
icatio
n
lab
el
co
r
r
esp
o
n
d
in
g
to
th
e
c
an
d
id
ate’
s
p
r
o
f
ess
io
n
al
d
o
m
ain
(
I
T
,
f
in
an
ce
,
en
g
in
ee
r
i
n
g
,
an
d
h
ea
lth
ca
r
e)
,
en
ab
lin
g
th
e
ev
alu
atio
n
o
f
m
o
d
el
p
er
f
o
r
m
a
n
c
e
o
n
a
m
u
lti
-
lab
el
class
if
icatio
n
task
.
Fo
llo
win
g
th
e
an
n
o
tatio
n
p
r
o
ce
s
s
,
we
id
en
tifie
d
eig
h
t
m
ajo
r
p
r
o
f
ess
io
n
al
ca
teg
o
r
ies
(
Fig
u
r
e
2
)
:
I
T
(
2
5
%),
Fin
an
ce
(
1
8
%),
E
n
g
in
ee
r
in
g
(
1
5
%),
Hea
lth
ca
r
e
(
1
2
%),
Ma
r
k
etin
g
(
1
0
%),
E
d
u
ca
tio
n
(
8
%),
L
aw
(
7
%),
an
d
Oth
er
(
5
%).
T
h
e
m
o
d
e
ls
wer
e
tr
ain
ed
a
n
d
ev
alu
ate
d
o
n
a
d
is
tr
ib
u
ted
in
f
r
astru
ctu
r
e,
u
s
in
g
a
v
ec
to
r
d
atab
ase
(
FAI
SS
)
f
o
r
th
e
R
AG
co
m
p
o
n
en
t a
n
d
L
L
Ms (
L
L
aM
A)
f
o
r
g
en
er
atio
n
.
Fig
u
r
e
2
.
Dis
tr
ib
u
tio
n
o
f
C
V
b
y
p
r
o
f
ess
io
n
al
d
o
m
ain
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
7
1
0
-
7
1
9
714
4
.
1
.
E
x
perim
ent
a
l
s
et
up
T
h
e
ex
p
er
im
en
ts
wer
e
co
n
d
u
cted
o
n
a
Dell
Vo
s
tr
o
3
5
1
0
lap
to
p
e
q
u
ip
p
ed
with
an
I
n
tel
C
o
r
e
i7
-
1
1
6
5
G7
(
1
1
th
g
en
er
atio
n
)
p
r
o
ce
s
s
o
r
,
1
6
GB
o
f
R
AM
,
an
d
an
in
teg
r
ated
I
n
tel
I
r
is
Xe
GPU.
G
iv
en
th
ese
lim
ited
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
,
h
ea
v
y
m
o
d
els
wer
e
n
o
t
ex
ec
u
ted
lo
ca
lly
:
lig
h
tweig
h
t
co
n
tr
asti
v
e
m
o
d
els
(
SimCS
E
-
b
ase,
C
o
n
tr
iev
er
-
b
a
s
e)
wer
e
u
s
ed
in
in
f
e
r
en
ce
m
o
d
e,
wh
ile
th
e
g
en
er
ativ
e
m
o
d
el
(
L
L
aM
A)
was
ac
ce
s
s
ed
v
ia
API
o
r
in
a
q
u
an
t
ized
C
PU
-
opt
im
ized
v
er
s
io
n
.
T
h
e
d
ataset
o
f
5
0
,
0
0
0
C
Vs
was
s
p
lit
in
to
8
0
%
f
o
r
tr
ain
in
g
,
1
0
%
f
o
r
v
alid
atio
n
,
an
d
1
0
%
f
o
r
test
in
g
to
en
s
u
r
e
r
ig
o
r
o
u
s
ev
alu
atio
n
.
T
ex
tu
al
a
n
d
v
is
u
al
em
b
ed
d
in
g
s
wer
e
p
r
o
jecte
d
in
t
o
5
1
2
d
im
en
s
io
n
s
b
ef
o
r
e
n
o
r
m
aliza
tio
n
an
d
in
d
e
x
in
g
i
n
FAI
SS
f
o
r
r
etr
iev
al
o
p
e
r
atio
n
s
.
T
h
e
m
ain
h
y
p
er
p
a
r
am
ete
r
s
u
s
ed
wer
e:
b
atch
s
ize
=
1
6
,
m
ax
le
n
g
th
=
2
5
6
to
k
en
s
,
an
d
lea
r
n
in
g
r
ate
=
2
e
-
5
f
o
r
m
o
d
u
les r
eq
u
ir
in
g
a
d
ju
s
tm
en
t.
Fair
n
ess
ev
alu
atio
n
r
elied
o
n
m
ea
s
u
r
in
g
d
em
o
g
r
ap
h
ic
d
is
p
ar
ities
b
etwe
en
s
en
s
iti
v
e
g
r
o
u
p
s
,
p
ar
ti
cu
lar
ly
th
e
g
en
d
er
g
a
p
,
ca
lcu
lated
b
ef
o
r
e
an
d
af
ter
a
p
p
ly
in
g
th
e
f
air
n
ess
m
o
d
u
le.
A
f
air
n
ess
in
d
e
x
r
an
g
in
g
f
r
o
m
0
to
1
co
m
p
lem
en
ts
th
is
an
aly
s
is
,
with
v
alu
es
clo
s
er
to
1
in
d
icatin
g
a
s
ig
n
i
f
ican
t
r
ed
u
ctio
n
in
d
is
p
ar
ities
.
4
.
2
.
B
a
s
elines
T
o
ass
ess
th
e
r
elev
an
ce
o
f
o
u
r
ap
p
r
o
ac
h
,
we
co
m
p
ar
ed
its
p
er
f
o
r
m
an
ce
with
s
ev
er
al
m
o
d
els:
-
SVM+
T
F
-
I
DF:
a
clas
s
ical
b
as
elin
e
f
o
r
tex
t c
lass
if
icatio
n
.
-
B
E
R
T
/
R
o
B
E
R
T
a
:
b
en
ch
m
ar
k
m
o
d
els f
o
r
NL
P c
lass
if
icatio
n
.
-
C
o
n
tr
asti
v
e+
g
en
er
ativ
e
h
y
b
r
id
:
a
p
ip
elin
e
in
teg
r
atin
g
SimCS
E
an
d
L
L
aMA.
-
Hy
b
r
id
+RAG:
an
en
r
ich
e
d
co
m
b
in
atio
n
with
e
x
ter
n
al
k
n
o
w
led
g
e
an
ch
o
r
in
g
.
-
Pro
p
o
s
ed
m
u
ltim
o
d
al+
f
air
n
es
s
Pip
elin
e:
f
u
ll
in
teg
r
atio
n
o
f
co
n
tr
asti
v
e
lear
n
i
n
g
,
g
en
er
ativ
e
m
o
d
elin
g
,
R
AG,
an
d
f
air
n
ess
m
ec
h
an
is
m
s
.
5.
RE
SU
L
T
S
T
h
e
ex
p
er
im
e
n
tal
r
esu
lts
h
ig
h
lig
h
t
th
e
clea
r
s
u
p
er
io
r
ity
o
f
h
y
b
r
i
d
an
d
f
air
n
ess
-
awa
r
e
ap
p
r
o
ac
h
es
co
m
p
ar
ed
to
class
ical
m
eth
o
d
s
(
T
ab
le
1
)
.
T
h
e
SVM+
T
F
-
I
DF
m
o
d
el
ac
h
iev
e
d
o
n
ly
7
6
.
2
%
ac
cu
r
ac
y
,
with
lim
ited
g
en
er
aliza
tio
n
ca
p
a
b
ilit
y
,
co
n
f
ir
m
i
n
g
th
e
wea
k
n
ess
o
f
tr
ad
itio
n
al
m
eth
o
d
s
wh
e
n
d
ea
lin
g
with
th
e
h
eter
o
g
en
e
o
u
s
s
tr
u
ctu
r
e
o
f
C
Vs.
T
r
an
s
f
o
r
m
er
-
b
ased
m
o
d
el
s
s
u
ch
as
B
E
R
T
(
8
4
.
9
%
)
a
n
d
R
o
B
E
R
T
a
(
8
6
.
4
%)
s
ig
n
if
ican
tly
im
p
r
o
v
e
p
er
f
o
r
m
an
ce
th
a
n
k
s
to
th
eir
co
n
tex
tu
al
r
e
p
r
esen
tatio
n
s
,
wh
ic
h
ca
p
tu
r
e
s
em
an
ti
c
r
elatio
n
s
h
ip
s
ac
r
o
s
s
d
if
f
er
en
t s
ec
tio
n
s
o
f
a
C
V.
T
h
e
in
itial
h
y
b
r
id
ap
p
r
o
ac
h
,
co
m
b
in
in
g
co
n
tr
asti
v
e
lear
n
in
g
(
SimCS
E
,
c
o
n
tr
iev
er
)
wit
h
g
en
er
ativ
e
lear
n
in
g
(
L
L
aM
A)
,
y
ield
e
d
a
n
o
tab
le
im
p
r
o
v
em
en
t,
r
ea
ch
i
n
g
9
1
.
7
%
ac
cu
r
ac
y
an
d
8
9
.
4
%
Ma
cr
o
-
F1
.
T
h
is
en
h
an
ce
m
e
n
t
illu
s
tr
ates
th
e
co
m
p
lem
en
tar
ity
o
f
co
n
tr
asti
v
e
m
o
d
els,
wh
ich
p
r
o
d
u
ce
r
o
b
u
s
t
r
ep
r
esen
tatio
n
s
,
an
d
g
e
n
er
ativ
e
m
o
d
els,
wh
ich
en
r
ich
an
d
n
o
r
m
alize
ca
n
d
i
d
ate
p
r
o
f
iles
.
Ad
d
in
g
R
AG
p
r
o
v
id
ed
a
n
ad
d
itio
n
al
g
ain
:
b
y
g
r
o
u
n
d
in
g
m
o
d
el
o
u
tp
u
ts
in
s
tr
u
ctu
r
e
d
k
n
o
wled
g
e
b
ases
,
th
e
s
y
s
tem
ac
h
iev
ed
9
4
.
2
%
ac
cu
r
ac
y
an
d
9
2
.
3
%
Ma
cr
o
-
F1
,
s
ig
n
if
ica
n
tly
r
ed
u
cin
g
h
allu
cin
atio
n
s
an
d
im
p
r
o
v
in
g
th
e
c
o
n
s
is
ten
cy
o
f
s
k
ill
-
b
ased
class
if
icatio
n
s
.
Fin
ally
,
th
e
m
u
ltimo
d
al
p
ip
elin
e
with
f
air
n
ess
ac
h
iev
ed
th
e
b
est
o
v
er
all
p
er
f
o
r
m
an
ce
:
9
5
.
6
%
ac
cu
r
ac
y
,
9
3
.
2
%
m
ac
r
o
-
F1
,
9
2
.
0
%
r
ec
all,
an
d
9
4
.
4
%
p
r
ec
i
s
io
n
,
o
u
tp
er
f
o
r
m
i
n
g
all
o
th
er
ap
p
r
o
ac
h
es.
T
h
ese
r
esu
lts
co
n
f
ir
m
th
e
r
o
b
u
s
tn
ess
o
f
m
u
ltimo
d
al
r
ep
r
esen
tatio
n
s
an
d
d
em
o
n
s
tr
ate
th
at
f
air
n
ess
ad
ju
s
tm
en
ts
d
o
n
o
t e
n
tail a
lo
s
s
in
p
r
e
d
ictiv
e
p
er
f
o
r
m
a
n
ce
(
T
ab
le
1
)
.
T
ab
le
1
.
C
o
m
p
a
r
is
o
n
o
f
m
o
d
el
p
er
f
o
r
m
an
ce
s
M
o
d
e
l
A
c
c
u
r
a
c
y
M
a
c
r
o
F
1
-
c
o
r
e
R
e
c
a
l
l
P
r
e
c
i
s
i
o
n
F
a
i
r
n
e
ss
i
n
d
e
x
S
V
M
(
TF
-
I
D
F
)
7
6
.
2
%
7
3
.
5
%
7
1
.
8
%
7
5
.
3
%
0
.
7
0
B
ER
T
8
4
.
9
%
8
2
.
1
%
8
0
.
5
%
8
3
.
7
%
0
.
7
8
R
o
B
E
R
Ta
8
6
.
4
%
8
3
.
8
%
8
2
.
2
%
8
5
.
1
%
0
7
9
H
y
b
r
i
d
a
p
p
r
o
a
c
h
(
c
o
n
t
r
a
st
i
v
e
+
g
e
n
e
r
a
t
i
v
e
)
9
1
.
7
%
8
9
.
4
%
8
8
.
1
%
9
0
.
6
%
0
.
8
5
H
y
b
r
i
d
a
p
p
r
o
a
c
h
(
c
o
n
t
r
a
st
i
v
e
+
g
e
n
e
r
a
t
i
v
e
+
R
A
G
)
9
4
.
2
%
9
2
.
3
%
9
1
.
1
%
9
3
.
5
%
0
.
8
9
P
r
o
p
o
se
d
m
u
l
t
i
mo
d
a
l
p
i
p
e
l
i
n
e
+
f
a
i
r
n
e
s
s mo
d
e
l
9
5
.
6
%
9
3
.
2
%
9
2
.
0
%
9
4
.
4
%
0
.
9
4
5
.
1
.
F
a
irness
e
v
a
lua
t
io
n
B
ey
o
n
d
r
aw
p
er
f
o
r
m
an
ce
,
f
air
n
ess
ev
alu
atio
n
is
a
ce
n
tr
al
co
m
p
o
n
e
n
t
o
f
th
is
s
tu
d
y
.
C
lass
ic
al
m
o
d
els
s
u
ch
as
SVM
an
d
B
E
R
T
ex
h
ib
it
s
ig
n
if
ican
t
d
is
p
ar
ities
b
etwe
en
d
em
o
g
r
a
p
h
ic
g
r
o
u
p
s
,
with
d
em
o
g
r
a
p
h
ic
p
ar
ity
g
a
p
s
ex
ce
ed
in
g
1
0
%.
T
h
ese
g
ap
s
r
ef
lect
b
iases
in
h
er
ited
f
r
o
m
th
e
tr
ain
in
g
d
at
a,
p
ar
ticu
lar
ly
th
e
o
v
er
r
e
p
r
esen
tatio
n
o
f
ce
r
tain
p
r
o
f
ess
io
n
al
d
o
m
ai
n
s
(
s
u
ch
as I
T
an
d
f
i
n
an
ce
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
I
n
teg
r
a
tin
g
c
o
n
tr
a
s
tive
a
n
d
g
e
n
era
tive
A
I
w
ith
R
A
G
fo
r
r
esp
o
n
s
ib
le
…
(
S
o
u
mia
C
h
a
fi
)
715
T
h
e
in
tr
o
d
u
ctio
n
o
f
th
e
f
air
n
ess
m
o
d
u
le
d
r
asti
ca
lly
r
ed
u
ce
d
th
ese
d
is
p
ar
ities
.
B
y
co
m
b
in
in
g
p
r
e
-
p
r
o
ce
s
s
in
g
(
r
e
b
alan
cin
g
,
c
o
u
n
ter
f
ac
t
u
al
au
g
m
e
n
tatio
n
)
,
in
-
p
r
o
ce
s
s
in
g
(
f
air
n
ess
-
a
war
e
em
b
ed
d
i
n
g
s
,
ad
v
er
s
ar
ial
r
e
g
u
lar
izatio
n
)
,
a
n
d
p
o
s
t
-
p
r
o
ce
s
s
in
g
(
d
ec
is
io
n
th
r
esh
o
l
d
ca
lib
r
atio
n
)
,
t
h
e
p
r
o
p
o
s
ed
p
i
p
elin
e
r
ed
u
ce
s
in
ter
-
g
r
o
u
p
g
a
p
s
to
b
elo
w
3
%
wh
ile
m
ain
tain
in
g
h
ig
h
p
er
f
o
r
m
an
ce
.
T
h
e
f
air
n
ess
in
d
ex
r
is
es
f
r
o
m
0
.
7
0
(
SVM)
an
d
0
.
7
8
(
B
E
R
T
)
to
0
.
9
4
in
th
e
f
in
al
p
ip
elin
e,
d
em
o
n
s
tr
atin
g
th
e
f
ea
s
ib
ilit
y
o
f
r
esp
o
n
s
ib
le
AI
in
r
ec
r
u
i
tm
en
t (
T
a
b
le
2
)
.
Gen
d
er
-
b
ased
ev
al
u
atio
n
s
h
o
ws
th
at,
with
o
u
t
co
r
r
ec
tio
n
,
th
e
m
o
d
el
f
av
o
r
e
d
m
ale
C
Vs
(
9
4
.
8
%)
o
v
e
r
f
em
ale
C
Vs
(
8
9
.
2
%),
r
esu
ltin
g
in
a
5
.
6
-
p
o
in
t
g
a
p
.
Af
te
r
in
teg
r
atin
g
th
e
f
air
n
ess
m
o
d
u
le
,
p
er
f
o
r
m
an
ce
b
ec
o
m
es
b
alan
ce
d
(
9
5
.
4
%
v
s
.
9
5
.
1
%),
r
ed
u
ci
n
g
th
e
g
a
p
to
o
n
ly
0
.
3
p
o
in
ts
.
Ov
er
all
ac
cu
r
ac
y
in
cr
ea
s
es
f
r
o
m
9
2
.
0
%
to
9
5
.
6
%,
an
d
t
h
e
f
ai
r
n
ess
in
d
ex
r
ea
ch
es
0
.
9
4
,
co
n
f
ir
m
in
g
t
h
at
f
air
n
ess
ca
n
b
e
im
p
r
o
v
e
d
with
o
u
t
co
m
p
r
o
m
is
in
g
p
er
f
o
r
m
a
n
ce
T
ab
le
3
an
d
Fig
u
r
e
3
.
T
ab
le
2
.
I
m
p
ac
t o
f
th
e
f
air
n
ess
m
o
d
u
le
o
n
p
er
f
o
r
m
an
ce
a
n
d
e
q
u
ity
M
e
t
r
i
c
B
e
f
o
r
e
f
a
i
r
n
e
ss
A
f
t
e
r
f
a
i
r
n
e
ss
I
mp
r
o
v
e
m
e
n
t
A
c
c
u
r
a
c
y
(
%)
9
4
.
2
9
5
.
6
+
1
.
4
M
a
c
r
o
-
F
1
(
%)
9
2
.
3
9
3
.
2
+
0
.
9
R
e
c
a
l
l
(
%)
9
1
.
1
9
2
.
0
+
0
.
9
P
r
e
c
i
s
i
o
n
(
%)
9
3
.
5
9
4
.
4
+
0
.
9
D
e
mo
g
r
a
p
h
i
c
p
a
r
i
t
y
g
ap
5
.
6
%
0
.
3
%
–
5
.
3
%
F
a
i
r
n
e
ss
i
n
d
e
x
0
.
8
9
0
.
9
4
+
0
.
0
5
T
ab
le
3
.
R
esu
lts
o
f
C
V
class
if
i
ca
tio
n
b
y
g
en
d
e
r
(
b
e
f
o
r
e
a
n
d
a
f
ter
th
e
f
air
n
ess
m
o
d
u
le)
G
e
n
r
e
A
c
c
u
r
a
c
y
(
%)
B
e
f
o
r
e
A
c
c
u
r
a
c
y
(
%) A
f
t
e
r
D
i
f
f
e
r
e
n
c
e
F
a
i
r
n
e
ss
i
n
d
e
x
M
a
l
e
9
4
.
8
9
5
.
4
+
0
.
6
F
e
mal
e
8
9
.
2
9
5
.
1
+
5
.
9
M
a
l
e
/
f
e
ma
l
e
g
r
a
p
5
.
6
0
.
3
–
5
.
3
O
v
e
r
a
l
l
9
2
.
0
9
5
.
6
+
3
.
6
0
.
9
4
T
h
e
f
air
n
ess
in
d
e
x
r
ef
lects
t
h
e
lev
el
o
f
eq
u
ality
b
etwe
en
s
en
s
itiv
e
g
r
o
u
p
s
;
a
v
alu
e
c
lo
s
e
to
1
in
d
icate
s
th
e
ab
s
en
ce
o
f
d
is
p
ar
ities
.
T
h
is
in
d
ex
is
ca
lcu
lated
b
y
m
ea
s
u
r
i
n
g
th
e
p
er
f
o
r
m
an
ce
g
ap
b
etwe
en
s
en
s
itiv
e
g
r
o
u
p
s
(
e.
g
.
,
m
ale/f
e
m
ale)
in
th
e
m
o
d
el’
s
p
r
e
d
ictio
n
s
.
I
t c
o
r
r
esp
o
n
d
s
to
:
=
1
−
i
n
o
u
r
ca
s
e:
I
n
itial g
en
d
er
g
ap
=
5
.
6
;
Fin
al
g
e
n
d
er
g
ap
=
0
.
3
.
F
1
=
1
−
0
.
3
5
.
6
=
1
-
0
.
0
5
3
5
7
=
0
.
9
4
6
4
≈
0
.
9
4
A
v
alu
e
clo
s
e
t
o
1
in
d
icate
s
t
h
at
th
e
m
o
d
el’
s
p
er
f
o
r
m
an
ce
is
s
im
ilar
ac
r
o
s
s
g
r
o
u
p
s
,
m
ea
n
in
g
d
is
p
ar
ities
ar
e
lo
w,
wh
er
ea
s
a
v
alu
e
cl
o
s
e
to
0
r
ef
lects st
r
o
n
g
in
e
q
u
ality
in
t
r
ea
tm
en
t.
F
ig
u
re
3
.
C
o
m
p
a
rin
g
CV cla
ss
ifi
c
a
ti
o
n
a
c
c
u
ra
c
y
b
y
g
e
n
d
e
r
b
e
fo
re
a
n
d
a
fter ap
p
ly
in
g
t
h
e
fa
irn
e
ss
m
o
d
u
le
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
7
1
0
-
7
1
9
716
5
.
2
.
Dis
cus
s
io
n
C
o
m
p
ar
ed
with
ex
is
tin
g
wo
r
k
,
o
u
r
r
esu
lts
in
T
ab
le
4
s
h
o
w
a
s
ig
n
if
ican
t
im
p
r
o
v
em
e
n
t
in
b
o
th
p
er
f
o
r
m
an
ce
an
d
f
air
n
ess
.
C
la
s
s
ical
ap
p
r
o
ac
h
es
b
ased
o
n
T
F
-
I
DF
an
d
SVM
g
en
er
ally
ac
h
iev
e
b
etwe
en
7
0
%
an
d
8
0
%
ac
cu
r
ac
y
o
n
h
eter
o
g
en
eo
u
s
C
Vs
[
4
]
,
wh
il
e
T
r
a
n
s
f
o
r
m
er
m
o
d
els
s
u
ch
as
B
E
R
T
an
d
R
o
B
E
R
T
a
ty
p
ically
r
ea
ch
b
etwe
en
8
2
%
an
d
8
8
%
b
u
t
s
till
ex
h
ib
it
f
air
n
ess
g
ap
s
ex
ce
ed
in
g
1
0
%
b
et
wee
n
d
em
o
g
r
a
p
h
ic
g
r
o
u
p
s
[
1
]
.
C
o
n
tr
asti
v
e
m
eth
o
d
s
lik
e
SimC
SE
im
p
r
o
v
e
tex
tu
al
s
im
ilar
ity
,
y
et
th
eir
clas
s
if
ic
atio
n
p
er
f
o
r
m
an
ce
r
ar
ely
ex
ce
e
d
s
8
9
-
9
1
%
a
n
d
th
ey
d
o
n
o
t
in
c
o
r
p
o
r
ate
an
y
co
m
p
r
eh
e
n
s
iv
e
b
ias
m
itig
atio
n
m
ec
h
a
n
is
m
s
[
7
]
.
I
s
o
lated
f
air
n
ess
tech
n
iq
u
es
-
s
u
ch
as
r
ewe
ig
h
tin
g
,
r
e
g
u
lar
iza
tio
n
,
o
r
ad
v
er
s
ar
ial
lear
n
in
g
-
o
f
ten
r
ed
u
ce
g
a
p
s
to
ar
o
u
n
d
5
-
8
%,
b
u
t
at
th
e
co
s
t
o
f
a
p
er
f
o
r
m
an
c
e
d
r
o
p
o
f
2
to
5
p
o
in
ts
[
1
8
]
-
[
2
0
]
.
I
n
co
m
p
ar
is
o
n
,
o
u
r
in
teg
r
ated
p
ip
elin
e
ac
h
iev
es
9
5
.
6
%
ac
c
u
r
ac
y
wh
ile
r
ed
u
ci
n
g
th
e
d
em
o
g
r
ap
h
ic
g
ap
to
o
n
ly
0
.
3
%,
d
em
o
n
s
tr
atin
g
t
h
at
it
is
p
o
s
s
ib
le
to
s
im
u
ltan
eo
u
s
ly
im
p
r
o
v
e
b
o
t
h
p
e
r
f
o
r
m
an
ce
an
d
f
ai
r
n
ess
with
in
a
u
n
if
i
ed
m
e
th
o
d
o
lo
g
ical
f
r
am
ewo
r
k
.
T
ab
le
4
.
C
o
m
p
a
r
ativ
e
r
esu
lts
with
o
th
er
s
tu
d
ies in
th
e
liter
at
u
r
e
A
p
p
r
o
c
h
B
e
st
a
c
c
u
r
a
c
y
r
e
p
o
r
t
e
d
i
n
t
h
e
l
i
t
e
r
a
t
u
r
e
O
u
r
a
c
c
u
r
a
c
y
TF
-
I
D
F
+
S
V
M
[
2
1
]
8
7
.
8
%
7
6
.
2
%
B
ER
T
/
R
o
B
ER
Ta
[
2
2
]
8
5
.
6
5
%
8
6
.
4
%
S
i
mCSE
(
C
o
n
t
r
a
st
i
v
e
)
[
7
]
7
6
.
5
0
%
-
8
8
.
4
5
%
9
1
.
7
%
C
o
n
t
r
a
st
i
v
e
+
G
é
n
é
r
a
t
i
v
e
N
o
c
o
m
b
i
n
e
d
st
u
d
y
a
v
a
i
l
a
b
l
e
9
1
.
7
%
R
A
G
[
2
3
]
90
–
9
3
%
9
4
.
2
%
F
a
i
r
n
e
ss
M
L
(
F
a
i
r
n
e
ss
i
n
d
e
x
)
[
1
8
]
,
[
2
4
]
0
.
6
0
–
0
.
8
0
0
.
9
4
T
h
e
r
esu
lts
co
n
f
ir
m
th
at
ea
c
h
co
m
p
o
n
e
n
t
co
n
t
r
ib
u
tes
a
m
ea
s
u
r
ab
le
im
p
r
o
v
em
e
n
t,
b
u
t
it
is
th
e
co
m
b
in
atio
n
o
f
co
n
t
r
asti
v
e
an
d
g
en
e
r
ativ
e
m
o
d
u
les
th
at
p
r
o
v
id
es
th
e
f
ir
s
t
s
ig
n
if
ica
n
t
p
er
f
o
r
m
an
ce
g
ain
.
T
h
e
ad
d
itio
n
o
f
R
AG
ac
ts
as
an
ad
d
itio
n
al
lev
er
,
s
tr
en
g
th
e
n
i
n
g
r
ep
r
esen
tatio
n
r
o
b
u
s
tn
e
s
s
an
d
im
p
r
o
v
in
g
g
en
er
aliza
tio
n
,
p
ar
ticu
lar
ly
o
n
u
n
s
tr
u
ctu
r
e
d
C
Vs.
T
h
e
co
m
p
lete
p
ip
elin
e
th
u
s
ac
h
ie
v
es
th
e
b
est
p
er
f
o
r
m
an
ce
,
o
f
f
er
in
g
a
clea
r
a
d
v
an
tag
e
o
v
e
r
all
p
ar
tial c
o
n
f
i
g
u
r
atio
n
s
(
Fig
u
r
es 4
-
9
)
.
T
wo
k
ey
o
b
s
er
v
atio
n
s
em
e
r
g
e
f
r
o
m
th
is
s
tu
d
y
:
-
T
h
e
s
y
n
er
g
y
b
etwe
en
co
n
tr
asti
v
e+
g
en
er
ativ
e
+RAG
en
h
an
c
es
co
n
tex
tu
aliza
tio
n
,
r
ed
u
ce
s
h
allu
cin
atio
n
s
,
an
d
en
s
u
r
es r
o
b
u
s
t c
lass
if
icatio
n
ac
r
o
s
s
h
eter
o
g
en
e
o
u
s
C
Vs.
-
T
h
e
f
air
n
ess
m
o
d
u
le
d
em
o
n
s
tr
ates
th
at
it
is
p
o
s
s
ib
le
to
ac
h
iev
e
b
o
th
h
i
g
h
p
e
r
f
o
r
m
an
ce
a
n
d
a
s
ig
n
if
ican
t
r
ed
u
ctio
n
i
n
b
ias,
illu
s
tr
atin
g
t
h
e
f
ea
s
ib
ilit
y
o
f
r
esp
o
n
s
ib
le
A
I
ap
p
lied
to
r
ec
r
u
itm
e
n
t.
T
h
ese
r
esu
lts
co
n
f
ir
m
th
at
in
teg
r
atin
g
th
e
p
r
in
cip
les
o
f
f
a
ir
n
ess
,
tr
an
s
p
ar
en
cy
,
a
n
d
e
x
p
lain
ab
ilit
y
with
in
a
class
if
icatio
n
p
ip
elin
e
is
n
o
t
o
n
ly
f
ea
s
ib
le
b
u
t
c
o
n
s
titu
tes
a
k
ey
lev
e
r
f
o
r
b
u
ild
in
g
s
o
cially
ac
ce
p
tab
le
an
d
tr
u
s
two
r
th
y
r
ec
r
u
itm
en
t sy
s
tem
s
.
I
n
f
er
en
ce
tim
e
(
T
a
b
le
5
)
,
th
e
tim
e
r
eq
u
ir
ed
to
p
r
o
ce
s
s
a
C
V
an
d
p
r
o
d
u
ce
a
p
r
ed
ictio
n
,
is
a
k
ey
cr
iter
io
n
f
o
r
p
r
o
d
u
ctio
n
-
lev
el
class
if
icatio
n
s
y
s
tem
s
.
I
n
o
u
r
s
tu
d
y
,
tr
a
d
itio
n
al
m
o
d
els
s
u
ch
as
SVM
s
tan
d
o
u
t
f
o
r
th
eir
h
ig
h
s
p
ee
d
(
2
3
m
s
p
er
C
V)
,
wh
er
ea
s
T
r
an
s
f
o
r
m
e
r
m
o
d
els
lik
e
B
E
R
T
an
d
R
o
B
E
R
T
a
ex
h
ib
it
h
ig
h
er
av
er
ag
e
p
r
o
ce
s
s
in
g
tim
es,
at
1
5
8
m
s
an
d
1
7
4
m
s
,
r
esp
ec
tiv
el
y
.
T
h
e
h
y
b
r
id
a
p
p
r
o
ac
h
,
wh
ile
o
f
f
er
in
g
im
p
r
o
v
e
d
ac
cu
r
ac
y
,
in
c
u
r
s
a
s
lig
h
tly
h
ig
h
er
co
m
p
u
tatio
n
al
co
s
t
(
2
1
5
m
s
)
,
m
ain
ly
d
u
e
to
th
e
g
e
n
er
ativ
e
en
r
ich
m
en
t
m
o
d
u
le.
Fig
u
r
e
4
.
Mo
d
el
p
er
f
o
r
m
a
n
ce
Fig
u
r
e
5
.
SVM
-
a
cc
u
r
ac
y
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
I
n
teg
r
a
tin
g
c
o
n
tr
a
s
tive
a
n
d
g
e
n
era
tive
A
I
w
ith
R
A
G
fo
r
r
esp
o
n
s
ib
le
…
(
S
o
u
mia
C
h
a
fi
)
717
Fig
u
r
e
6
.
B
E
R
T
-
a
cc
u
r
ac
y
Fig
u
r
e
7
.
Hy
b
r
id
a
p
p
r
o
ac
h
-
a
cc
u
r
ac
y
with
o
u
t RAG
Fig
u
r
e
8
.
Hy
b
r
id
a
p
p
r
o
ac
h
-
a
cc
u
r
ac
y
with
R
AG
Fig
u
r
e
9
.
Hy
b
r
id
a
p
p
r
o
ac
h
-
a
cc
u
r
ac
y
with
R
AG
+
f
air
n
ess
m
o
d
u
le
T
a
b
le
5
.
A
v
e
r
a
g
e
i
n
f
e
r
e
n
c
e
ti
m
e
p
er
C
V
M
o
d
e
l
Te
mp
s m
o
y
e
n
(
ms)
S
V
M
2
3
ms
B
ER
T
1
5
8
m
s
R
o
B
E
R
Ta
1
7
4
m
s
H
y
b
r
i
d
e
2
1
5
m
s
6.
C
O
N
C
L
U
SIO
N
A
N
D
FU
T
U
R
E
WO
R
K
T
h
is
r
esear
ch
ac
h
iev
ed
its
o
b
jectiv
es
b
y
p
r
o
p
o
s
in
g
a
h
y
b
r
id
a
n
d
m
u
ltimo
d
al
p
ip
elin
e
f
o
r
C
V
class
if
icatio
n
,
ca
p
ab
le
o
f
s
ig
n
if
ican
tly
im
p
r
o
v
in
g
p
er
f
o
r
m
an
ce
wh
ile
r
ed
u
ci
n
g
g
en
d
er
b
ias.
T
h
e
an
aly
s
es
s
h
o
w
th
at
th
e
m
ain
d
r
iv
er
s
o
f
f
air
n
ess
im
p
r
o
v
em
en
t
s
tem
f
r
o
m
co
u
n
ter
f
ac
tu
al
g
en
er
atio
n
,
r
eg
u
lar
izatio
n
,
an
d
p
o
s
t
-
d
ec
is
io
n
ca
lib
r
atio
n
.
H
o
wev
er
,
th
e
wo
r
k
p
r
esen
ts
ce
r
ta
in
lim
itatio
n
s
,
n
o
tab
ly
th
e
e
v
alu
atio
n
f
o
c
u
s
ed
o
n
a
si
n
g
le
s
en
s
itiv
e
attr
ib
u
te
an
d
a
m
o
n
o
lin
g
u
al
co
r
p
u
s
,
wh
ic
h
m
ay
r
ed
u
ce
g
en
e
r
aliza
b
ilit
y
in
r
ea
l
-
wo
r
ld
s
ettin
g
s
.
B
u
ild
in
g
o
n
th
ese
r
esu
lts
,
f
u
t
u
r
e
wo
r
k
will
in
v
o
l
v
e
d
ee
p
e
n
in
g
th
e
an
aly
s
is
an
d
m
itig
atio
n
o
f
b
iases
r
elate
d
to
o
th
e
r
s
en
s
itiv
e
attr
i
b
u
tes,
s
u
ch
as
ag
e
o
r
eth
n
ic
o
r
ig
in
,
i
n
o
r
d
er
to
ex
ten
d
o
u
r
m
u
lti
-
lev
el
f
air
n
ess
ap
p
r
o
ac
h
b
ey
o
n
d
g
en
d
er
an
d
s
tr
en
g
th
en
th
e
eth
ical
r
o
b
u
s
tn
ess
o
f
th
e
p
r
o
p
o
s
ed
p
ip
elin
e.
Fu
r
th
er
m
o
r
e,
in
teg
r
atin
g
t
h
is
ap
p
r
o
ac
h
in
t
o
an
HR
I
S
o
f
f
er
s
c
o
n
cr
ete
p
r
o
s
p
ec
ts
f
o
r
m
o
r
e
tr
an
s
p
a
r
en
t,
ex
p
lain
a
b
le,
an
d
s
o
cially
r
esp
o
n
s
ib
le
r
ec
r
u
itm
e
n
t
p
r
o
ce
s
s
es.
W
e
also
p
lan
to
o
p
tim
ize
th
e
m
o
d
el’
s
in
f
er
en
c
e
tim
e
to
m
ak
e
th
e
p
ip
elin
e
lig
h
te
r
an
d
b
etter
s
u
it
ed
f
o
r
p
r
o
d
u
ctio
n
d
ep
lo
y
m
en
t
.
Fin
ally
,
a
n
ev
al
u
atio
n
i
n
v
o
lv
in
g
en
d
u
s
er
s
will
b
e
in
clu
d
ed
in
f
u
tu
r
e
wo
r
k
to
f
u
lly
v
alid
ate
th
e
in
ter
p
r
etab
il
ity
an
d
u
s
ef
u
ln
ess
o
f
th
e
ex
p
lan
atio
n
s
g
en
er
ated
b
y
th
e
m
o
d
el.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
is
in
v
o
lv
ed
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
a
r
y
20
2
6
:
7
1
0
-
7
1
9
718
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
So
u
m
ia
C
h
af
i
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Mu
s
tap
h
a
Kab
il
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ab
d
ess
am
ad
Kam
o
u
s
s
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
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
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
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
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
ter
est.
I
NF
O
RM
E
D
CO
NS
E
N
T
W
e
h
av
e
o
b
tain
ed
in
f
o
r
m
ed
c
o
n
s
en
t f
r
o
m
all
in
d
iv
id
u
als in
c
lu
d
ed
in
t
h
is
s
tu
d
y
.
DATA AV
AI
L
AB
I
L
I
T
Y
Der
iv
ed
d
ata
s
u
p
p
o
r
tin
g
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
ailab
le
f
r
o
m
th
e
c
o
r
r
esp
o
n
d
i
n
g
au
t
h
o
r
S.C
h
af
i o
n
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
P
.
Tr
u
s
t
a
n
d
R
.
M
i
n
g
h
i
m,
“
A
st
u
d
y
o
n
t
e
x
t
c
l
a
ssi
f
i
c
a
t
i
o
n
i
n
t
h
e
a
g
e
o
f
l
a
r
g
e
l
a
n
g
u
a
g
e
m
o
d
e
l
s
,
”
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
K
n
o
w
l
e
d
g
e
Ex
t
r
a
c
t
i
o
n
,
v
o
l
.
6
,
n
o
.
4
,
p
p
.
2
6
8
8
-
2
7
2
1
,
N
o
v
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
ma
k
e
6
0
4
0
1
2
9
.
[
2
]
G
.
I
z
a
c
a
r
d
e
t
a
l
.
,
“
U
n
s
u
p
e
r
v
i
se
d
d
e
n
se
i
n
f
o
r
mat
i
o
n
r
e
t
r
i
e
v
a
l
w
i
t
h
c
o
n
t
r
a
st
i
v
e
l
e
a
r
n
i
n
g
,
”
T
r
a
n
sa
c
t
i
o
n
s
o
n
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
Re
se
a
rc
h
,
v
o
l
.
2
0
2
2
-
A
u
g
u
s
t
.
A
u
g
.
2
9
,
2
0
2
2
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
1
1
2
.
0
9
1
1
8
.
[
3
]
T.
B
.
B
r
o
w
n
e
t
a
l
.
,
“
La
n
g
u
a
g
e
mo
d
e
l
s
a
r
e
f
e
w
-
s
h
o
t
l
e
a
r
n
e
r
s,
”
A
d
v
a
n
c
e
s
i
n
N
e
u
ra
l
I
n
f
o
rm
a
t
i
o
n
Pro
c
e
ssi
n
g
S
y
st
e
m
s
,
v
o
l
.
2
0
2
0
-
D
e
c
e
m
b
e
r
.
Ju
l
.
2
2
,
2
0
2
0
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
0
0
5
.
1
4
1
6
5
.
[
4
]
S
.
M
i
n
a
e
e
,
N
.
K
a
l
c
h
b
r
e
n
n
e
r
,
E.
C
a
mb
r
i
a
,
N
.
N
i
k
z
a
d
,
M
.
C
h
e
n
a
g
h
l
u
,
a
n
d
J
.
G
a
o
,
“
D
e
e
p
l
e
a
r
n
i
n
g
--
b
a
se
d
t
e
x
t
c
l
a
s
si
f
i
c
a
t
i
o
n
:
a
c
o
m
p
r
e
h
e
n
si
v
e
r
e
v
i
e
w
,
”
AC
M
C
o
m
p
u
t
i
n
g
S
u
rv
e
y
s
,
v
o
l
.
5
4
,
n
o
.
3
,
p
p
.
1
-
4
0
,
A
p
r
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
4
5
/
3
4
3
9
7
2
6
.
[
5
]
S
.
B
a
r
o
c
a
s
,
M
.
H
a
r
d
t
,
a
n
d
A
.
N
a
r
a
y
a
n
a
n
,
“
F
a
i
r
n
e
s
s
a
n
d
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
-
l
i
m
i
t
a
t
i
o
n
s
a
n
d
o
p
p
o
r
t
u
n
i
t
i
e
s
,
”
i
n
M
I
T
P
r
e
s
s
,
2
0
2
3
,
p
.
2
9
4
.
[
6
]
N
.
M
e
h
r
a
b
i
,
F
.
M
o
r
st
a
t
t
e
r
,
N
.
S
a
x
e
n
a
,
K
.
L
e
r
ma
n
,
a
n
d
A
.
G
a
l
s
t
y
a
n
,
“
A
su
r
v
e
y
o
n
b
i
a
s
a
n
d
f
a
i
r
n
e
ss
i
n
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
”
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
.
6
,
p
p
.
1
-
3
5
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
4
5
/
3
4
5
7
6
0
7
.
[
7
]
S
.
C
H
A
F
I
,
M
.
K
a
b
i
l
,
a
n
d
A
.
K
a
m
o
u
ss,
“
O
p
t
i
mi
z
i
n
g
a
u
t
o
m
a
t
i
c
C
V
c
l
a
s
si
f
i
c
a
t
i
o
n
w
i
t
h
c
o
n
t
r
a
st
i
v
e
a
n
d
g
e
n
e
r
a
t
i
v
e
l
e
a
r
n
i
n
g
,
”
Pro
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
6
5
,
p
p
.
3
4
2
-
3
4
9
,
2
0
2
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.2
0
2
5
.
0
7
.
1
9
0
.
[
8
]
T.
G
a
o
,
X
.
Y
a
o
,
a
n
d
D
.
C
h
e
n
,
“
S
i
mCSE:
si
m
p
l
e
c
o
n
t
r
a
st
i
v
e
l
e
a
r
n
i
n
g
o
f
se
n
t
e
n
c
e
e
m
b
e
d
d
i
n
g
s
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
1
C
o
n
f
e
re
n
c
e
o
n
Em
p
i
r
i
c
a
l
Me
t
h
o
d
s
i
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pr
o
c
e
ss
i
n
g
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
ss
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
Li
n
g
u
i
s
t
i
c
s,
2
0
2
1
,
p
p
.
6
8
9
4
–
6
9
1
0
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
1
.
e
mn
l
p
-
m
a
i
n
.
5
5
2
.
[
9
]
Y
.
G
a
o
e
t
a
l
.
,
“
R
e
t
r
i
e
v
a
l
-
a
u
g
m
e
n
t
e
d
g
e
n
e
r
a
t
i
o
n
f
o
r
l
a
r
g
e
l
a
n
g
u
a
g
e
m
o
d
e
l
s
:
a
s
u
r
v
e
y
.
”
M
a
r
.
2
7
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s
/
2
3
1
2
.
1
0
9
9
7
.
[
1
0
]
K
.
C
h
a
r
a
n
a
n
d
R
.
R
a
j
e
s
h
,
“
A
u
t
o
ma
t
e
d
r
e
su
m
e
scre
e
n
i
n
g
u
si
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
S
c
i
e
n
t
i
f
i
c
J
o
u
r
n
a
l
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
Ma
n
a
g
e
m
e
n
t
,
v
o
l
.
0
4
,
n
o
.
0
7
,
p
p
.
1
-
9
,
J
u
l
.
2
0
2
5
,
d
o
i
:
1
0
.
5
5
0
4
1
/
i
sj
e
m0
4
8
6
0
.
[
1
1
]
Y
.
X
u
,
M
.
L
i
,
L.
C
u
i
,
S
.
H
u
a
n
g
,
F
.
W
e
i
,
a
n
d
M
.
Zh
o
u
,
“
La
y
o
u
t
L
M
:
p
r
e
-
t
r
a
i
n
i
n
g
o
f
t
e
x
t
a
n
d
l
a
y
o
u
t
f
o
r
d
o
c
u
me
n
t
i
m
a
g
e
u
n
d
e
r
s
t
a
n
d
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
6
t
h
AC
M
S
I
G
K
D
D
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
K
n
o
w
l
e
d
g
e
D
i
sc
o
v
e
ry
&
D
a
t
a
Mi
n
i
n
g
,
N
e
w
Y
o
r
k
,
N
Y
,
U
S
A
:
A
C
M
,
A
u
g
.
2
0
2
0
,
p
p
.
1
1
9
2
-
1
2
0
0
.
d
o
i
:
1
0
.
1
1
4
5
/
3
3
9
4
4
8
6
.
3
4
0
3
1
7
2
.
[
1
2
]
M
.
L
e
w
i
s
e
t
a
l
.
,
“
B
A
R
T
:
D
e
n
o
i
si
n
g
se
q
u
e
n
c
e
-
to
-
s
e
q
u
e
n
c
e
p
r
e
-
t
r
a
i
n
i
n
g
f
o
r
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
g
e
n
e
r
a
t
i
o
n
,
t
r
a
n
s
l
a
t
i
o
n
,
a
n
d
c
o
m
p
r
e
h
e
n
s
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
5
8
t
h
a
n
n
u
a
l
m
e
e
t
i
n
g
o
f
t
h
e
a
sso
c
i
a
t
i
o
n
f
o
r
c
o
m
p
u
t
a
t
i
o
n
a
l
l
i
n
g
u
i
st
i
c
s
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
Li
n
g
u
i
st
i
c
s
,
2
0
2
0
,
p
p
.
7
8
7
1
-
7
8
8
0
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
a
c
l
-
ma
i
n
.
7
0
3
.
[1
3]
Y
.
S
.
C
h
u
a
n
g
e
t
a
l
.
,
“
D
i
f
f
C
S
E
:
D
i
f
f
e
r
e
n
c
e
-
b
a
se
d
c
o
n
t
r
a
st
i
v
e
l
e
a
r
n
i
n
g
f
o
r
sen
t
e
n
c
e
e
m
b
e
d
d
i
n
g
s
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
2
C
o
n
f
e
re
n
c
e
o
f
t
h
e
N
o
r
t
h
Am
e
ri
c
a
n
C
h
a
p
t
e
r
o
f
t
h
e
As
so
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s:
H
u
m
a
n
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
i
e
s
,
2
0
2
2
,
p
p
.
4
2
0
7
-
4
2
1
8
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
2
.
n
a
a
c
l
-
m
a
i
n
.
3
1
1
.
[
1
4
]
M
.
R
a
g
h
a
v
a
n
,
S
.
B
a
r
o
c
a
s
,
J
.
K
l
e
i
n
b
e
r
g
,
a
n
d
K
.
L
e
v
y
,
“
M
i
t
i
g
a
t
i
n
g
b
i
a
s
i
n
a
l
g
o
r
i
t
h
mi
c
h
i
r
i
n
g
:
E
v
a
l
u
a
t
i
n
g
c
l
a
i
ms
a
n
d
p
r
a
c
t
i
c
e
s,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
0
c
o
n
f
e
re
n
c
e
o
n
f
a
i
r
n
e
ss
,
a
c
c
o
u
n
t
a
b
i
l
i
t
y
,
N
e
w
Y
o
r
k
,
N
Y
,
U
S
A
:
A
C
M
,
J
a
n
.
2
0
2
0
,
p
p
.
4
6
9
-
4
8
1
.
d
o
i
:
1
0
.
1
1
4
5
/
3
3
5
1
0
9
5
.
3
3
7
2
8
2
8
.
[
1
5
]
N
.
R
e
i
mers
a
n
d
I
.
G
u
r
e
v
y
c
h
,
“
S
e
n
t
e
n
c
e
-
B
E
R
T
:
S
e
n
t
e
n
c
e
e
mb
e
d
d
i
n
g
s
u
si
n
g
si
a
mes
e
B
ER
T
-
n
e
t
w
o
r
k
s,
”
EM
N
L
P
-
I
J
C
N
L
P
2
0
1
9
-
2
0
1
9
C
o
n
f
e
r
e
n
c
e
o
n
Em
p
i
ri
c
a
l
Me
t
h
o
d
s
i
n
N
a
t
u
ra
l
L
a
n
g
u
a
g
e
Pr
o
c
e
ssi
n
g
a
n
d
9
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
re
n
c
e
o
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pr
o
c
e
ssi
n
g
,
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
C
o
n
f
e
re
n
c
e
.
p
p
.
3
9
8
2
-
3
9
9
2
,
A
u
g
.
2
7
,
2
0
1
9
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
D
1
9
-
1
4
1
0
.
[
1
6
]
A
.
R
a
d
f
o
r
d
e
t
a
l
.
,
“
L
e
a
r
n
i
n
g
t
r
a
n
sf
e
r
a
b
l
e
v
i
su
a
l
m
o
d
e
l
s
f
r
o
m
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
s
u
p
e
r
v
i
si
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
Re
se
a
rc
h
,
F
e
b
.
2
0
2
1
,
p
p
.
8
7
4
8
–
8
7
6
3
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
1
0
3
.
0
0
0
2
0
.
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
I
n
teg
r
a
tin
g
c
o
n
tr
a
s
tive
a
n
d
g
e
n
era
tive
A
I
w
ith
R
A
G
fo
r
r
esp
o
n
s
ib
le
…
(
S
o
u
mia
C
h
a
fi
)
719
[
1
7
]
F
.
K
a
mi
r
a
n
a
n
d
T.
C
a
l
d
e
r
s,
“
D
a
t
a
p
r
e
p
r
o
c
e
ss
i
n
g
t
e
c
h
n
i
q
u
e
s
f
o
r
c
l
a
ssi
f
i
c
a
t
i
o
n
w
i
t
h
o
u
t
d
i
s
c
r
i
mi
n
a
t
i
o
n
,
”
K
n
o
w
l
e
d
g
e
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
y
st
e
m
s
,
v
o
l
.
3
3
,
n
o
.
1
,
p
p
.
1
-
3
3
,
O
c
t
.
2
0
1
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
1
1
5
-
0
1
1
-
0
4
6
3
-
8.
[
1
8
]
M
.
B
.
Za
f
a
r
,
I
.
V
a
l
e
r
a
,
M
.
G
.
R
o
d
r
i
g
u
e
z
,
a
n
d
K
.
P
.
G
u
mm
a
d
i
,
“
F
a
i
r
n
e
ss
c
o
n
st
r
a
i
n
t
s:
M
e
c
h
a
n
i
sms
f
o
r
f
a
i
r
c
l
a
s
si
f
i
c
a
t
i
o
n
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
S
t
a
t
i
st
i
c
s
,
2
0
1
7
.
[
1
9
]
H
.
Ed
w
a
r
d
s
a
n
d
A
.
S
t
o
r
k
e
y
,
“
C
e
n
so
r
i
n
g
r
e
p
r
e
s
e
n
t
a
t
i
o
n
s
w
i
t
h
a
n
a
d
v
e
r
sar
y
,
”
i
n
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
L
e
a
rn
i
n
g
Re
p
r
e
se
n
t
a
t
i
o
n
s
,
M
a
r
.
2
0
1
6
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
1
5
1
1
.
0
5
8
9
7
.
[2
0]
T.
K
a
mi
s
h
i
m
a
,
S
.
A
k
a
h
o
,
a
n
d
J.
S
a
k
u
ma
,
“
F
a
i
r
n
e
ss
-
a
w
a
r
e
l
e
a
r
n
i
n
g
t
h
r
o
u
g
h
r
e
g
u
l
a
r
i
z
a
t
i
o
n
a
p
p
r
o
a
c
h
,
”
i
n
2
0
1
1
I
EEE
1
1
t
h
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
D
a
t
a
Mi
n
i
n
g
W
o
rks
h
o
p
s
,
I
EEE,
D
e
c
.
2
0
1
1
,
p
p
.
6
4
3
-
6
5
0
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
D
M
W
.
2
0
1
1
.
8
3
.
[
2
1
]
A
.
A
.
J.
K
a
r
i
m,
K
.
H
.
M
.
A
sa
d
,
a
n
d
A
.
A
z
a
m,
“
S
t
r
e
n
g
t
h
e
n
i
n
g
f
a
k
e
n
e
w
s
d
e
t
e
c
t
i
o
n
:
l
e
v
e
r
a
g
i
n
g
S
V
M
a
n
d
so
p
h
i
st
i
c
a
t
e
d
t
e
x
t
v
e
c
t
o
r
i
z
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
.
d
e
f
y
i
n
g
B
E
R
T?
”
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
4
1
1
.
1
2
7
0
3
.
[
2
2
]
J.
D
e
v
l
i
n
,
M
.
-
W
.
C
h
a
n
g
,
K
.
Le
e
,
K
.
T.
G
o
o
g
l
e
,
a
n
d
A
.
I
.
La
n
g
u
a
g
e
,
“
B
E
R
T:
p
r
e
-
t
r
a
i
n
i
n
g
o
f
d
e
e
p
b
i
d
i
r
e
c
t
i
o
n
a
l
t
r
a
n
sf
o
r
mers
f
o
r
l
a
n
g
u
a
g
e
u
n
d
e
r
s
t
a
n
d
i
n
g
,
”
i
n
2
0
1
9
C
o
n
f
e
re
n
c
e
o
f
t
h
e
N
o
rt
h
Am
e
ri
c
a
n
C
h
a
p
t
e
r
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
:
H
u
m
a
n
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
i
e
s,
V
o
l
u
m
e
1
(
L
o
n
g
a
n
d
S
h
o
rt
Pa
p
e
rs)
,
2
0
1
8
,
p
p
.
4
1
7
1
-
4
1
8
6
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
a
c
l
a
n
t
h
o
l
o
g
y
.
o
r
g
/
N
1
9
-
1
4
2
3
.
p
d
f
.
[
2
3
]
P
.
Le
w
i
s
e
t
a
l
.
,
“
R
e
t
r
i
e
v
a
l
-
a
u
g
m
e
n
t
e
d
g
e
n
e
r
a
t
i
o
n
f
o
r
k
n
o
w
l
e
d
g
e
-
i
n
t
e
n
si
v
e
N
LP
t
a
s
k
s,
”
Ad
v
a
n
c
e
s
i
n
N
e
u
ra
l
I
n
f
o
rm
a
t
i
o
n
Pro
c
e
ssi
n
g
S
y
s
t
e
m
s
,
v
o
l
.
2
0
2
0
-
D
e
c
e
m
b
e
r
.
A
p
r
.
1
2
,
2
0
2
0
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
0
0
5
.
1
1
4
0
1
.
[
2
4
]
M
.
H
a
r
d
t
,
E.
P
r
i
c
e
,
a
n
d
N
.
S
r
e
b
r
o
,
“
E
q
u
a
l
i
t
y
o
f
o
p
p
o
r
t
u
n
i
t
y
i
n
su
p
e
r
v
i
se
d
l
e
a
r
n
i
n
g
,
”
A
d
v
a
n
c
e
s
i
n
N
e
u
ra
l
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
S
y
s
t
e
m
s
.
p
p
.
3
3
2
3
–
3
3
3
1
,
O
c
t
.
0
7
,
2
0
1
6
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
1
6
1
0
.
0
2
4
1
3
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
S
o
u
m
ia
Ch
a
fi
is
a
d
o
c
to
ra
l
r
e
se
a
rc
h
e
r
a
t
Ha
ss
a
n
II
Un
i
v
e
rsity
Ca
sa
b
lan
c
a
with
re
se
a
rc
h
in
m
a
c
h
in
e
lea
rn
i
n
g
a
n
d
tex
t
c
las
sifica
ti
o
n
.
S
h
e
c
o
-
a
u
t
h
o
re
d
a
p
e
e
r
-
re
v
iew
e
d
re
se
a
rc
h
a
rti
c
le
in
2
0
2
4
o
n
a
tt
e
n
ti
o
n
-
b
a
se
d
c
las
sifica
ti
o
n
.
Re
se
a
rc
h
fiel
d
:
Co
m
p
u
ter
sc
ien
c
e
a
n
d
m
a
c
h
in
e
lea
rn
in
g
,
wit
h
a
fo
c
u
s
o
n
tex
t
c
las
sifica
ti
o
n
a
n
d
m
a
c
h
in
e
lea
rn
in
g
tec
h
n
iq
u
e
s.
Ba
se
d
o
n
h
is
a
c
a
d
e
m
ic
p
ro
fil
e
,
h
e
is
in
v
o
lv
e
d
in
re
se
a
rc
h
re
late
d
to
d
o
c
u
m
e
n
t/
CV
c
las
sifica
ti
o
n
with
a
tt
e
n
ti
o
n
m
e
c
h
a
n
ism
s
u
sin
g
d
e
e
p
lea
rn
in
g
m
o
d
e
ls.
Sh
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
so
u
m
ia.ch
a
fi@g
m
a
il
.
c
o
m
.
Mu
sta
p
h
a
K
a
b
il
is
a
u
n
iv
e
rs
it
y
p
ro
fe
ss
o
r
a
n
d
re
se
a
rc
h
e
r
a
t,
a
ffil
iate
d
wit
h
t
h
e
Lab
o
ra
to
r
y
o
f
M
a
th
e
m
a
ti
c
s,
Co
m
p
u
ter
S
c
ien
c
e
,
a
n
d
Ap
p
li
c
a
ti
o
n
s.
He
is
k
n
o
wn
fo
r
h
is
wo
r
k
in
:
M
a
th
e
m
a
ti
c
s
(fix
e
d
p
o
in
t
t
h
e
o
ry
,
c
o
m
b
i
n
a
to
r
ics
,
g
ra
p
h
t
h
e
o
ry
)
,
Co
m
p
u
tati
o
n
a
l
stru
c
t
u
re
s
a
n
d
th
e
ir
a
p
p
li
c
a
ti
o
n
s
,
Co
ll
a
b
o
r
a
ti
v
e
re
se
a
rc
h
in
m
a
c
h
in
e
lea
rn
in
g
a
n
d
d
a
ta
c
las
sifica
ti
o
n
m
e
th
o
d
s
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
u
sta
p
h
a
.
k
a
b
il
@fstm
.
a
c
.
m
a
.
Abd
e
ss
a
m
a
d
K
a
m
o
u
ss
a
t
Ha
ss
a
n
II
Un
iv
e
rsity
o
f
Ca
sa
b
lan
c
a
,
M
o
ro
c
c
o
,
w
h
e
re
h
e
is
a
ffil
iate
d
with
th
e
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s
a
n
d
Tec
h
n
o
lo
g
ies
(M
o
h
a
m
m
e
d
ia),
De
p
a
rtme
n
t
o
f
M
a
th
e
m
a
ti
c
s,
a
n
d
th
e
Lab
o
ra
to
r
y
o
f
M
a
th
e
m
a
ti
c
s
a
n
d
Ap
p
li
c
a
t
io
n
s.
His
re
se
a
rc
h
f
o
c
u
se
s
p
rima
ril
y
o
n
m
a
th
e
m
a
ti
c
a
l
a
n
a
ly
sis,
e
sp
e
c
ially
in
a
re
a
s
su
c
h
a
s
fix
e
d
-
p
o
i
n
t
t
h
e
o
ry
,
g
e
n
e
ra
li
z
e
d
m
e
tri
c
sp
a
c
e
s,
a
n
d
a
b
stra
c
t
fu
n
c
t
io
n
a
l
a
n
d
t
o
p
o
lo
g
ica
l
m
e
th
o
d
s
in
p
u
re
m
a
th
e
m
a
ti
c
s.
In
t
h
is
c
o
n
tex
t,
Ka
m
o
u
ss
h
a
s
c
o
-
a
u
th
o
re
d
m
u
l
ti
p
le
p
e
e
r
-
re
v
iew
e
d
a
rti
c
les
th
a
t
e
x
ten
d
c
las
sic
a
l
fix
e
d
-
p
o
i
n
t
th
e
o
re
m
s
(e
.
g
.
,
Ka
n
n
a
n
a
n
d
Ha
rd
y
-
Ro
g
e
rs
ty
p
e
re
su
lt
s)
to
g
e
n
e
ra
li
z
e
d
fra
m
e
wo
rk
s
su
c
h
a
s
C*
-
a
lg
e
b
ra
-
v
a
lu
e
d
m
e
tri
c
sp
a
c
e
s
a
n
d
g
ra
p
h
-
e
n
d
o
we
d
m
o
d
u
lar
s
p
a
c
e
s
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
k
a
m
o
u
ss
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.