I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
2
0
2
5
,
p
p
.
5
0
2
7
~
5
0
3
7
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
6
.
p
p
5
0
2
7
-
5
0
3
7
5027
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
M
etaheuris
tic
o
pt
imiza
tion for sa
rc
a
sm
det
ect
io
n in s
o
cia
l
media
wit
h
embe
dding
and pa
ddin
g
t
echniqu
es
G
ee
t
a
Sa
hu
1
,
M
a
no
j
H
ud
nu
r
k
a
r
2
1
D
e
p
a
r
t
m
e
n
t
o
f
I
n
f
o
r
ma
t
i
o
n
T
e
c
h
n
o
l
o
g
y
a
n
d
D
a
t
a
S
c
i
e
n
c
e
,
V
i
d
y
a
l
a
n
k
a
r
S
c
h
o
o
l
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
M
u
mb
a
i
,
I
n
d
i
a
2
O
p
e
r
a
t
i
o
n
s a
n
d
A
n
a
l
y
t
i
c
s,
S
y
m
b
i
o
si
s
C
e
n
t
r
e
f
o
r
M
a
n
a
g
e
me
n
t
a
n
d
H
u
ma
n
R
e
s
o
u
r
c
e
D
e
v
e
l
o
p
me
n
t
(
S
C
M
H
R
D
)
,
S
y
mb
i
o
si
s I
n
t
e
r
n
a
t
i
o
n
a
l
(
D
e
e
me
d
U
n
i
v
e
r
si
t
y
)
,
P
u
n
e
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
18
,
2
0
2
4
R
ev
is
ed
Oct
3
,
2025
Acc
ep
ted
Oct
16
,
2
0
2
5
S
a
rc
a
sm
is
a
so
p
h
isti
c
a
ted
m
o
d
e
o
f
e
x
p
re
ss
io
n
t
h
a
t
a
ll
o
ws
s
p
e
a
k
e
rs
to
e
x
p
re
ss
th
e
ir
o
p
i
n
io
n
s
su
b
t
ly
.
S
t
a
k
e
h
o
l
d
e
rs
p
ro
v
id
e
u
n
str
u
c
tu
re
d
m
e
ss
a
g
e
s
with
e
x
ten
d
e
d
p
h
ra
se
s,
m
a
k
i
n
g
it
d
iffi
c
u
lt
fo
r
c
o
m
p
u
ters
a
n
d
p
e
o
p
le
t
o
u
n
d
e
rsta
n
d
.
Th
is
re
se
a
rc
h
a
ims
t
o
d
e
v
e
lo
p
a
sa
rc
a
sm
d
e
tec
ti
o
n
m
e
th
o
d
to
id
e
n
ti
f
y
wo
rd
s
i
n
p
h
ra
se
s
a
s
sa
r
c
a
stic
o
r
n
o
n
-
sa
rc
a
stic
fro
m
tex
t
,
u
ti
li
z
i
n
g
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
a
p
p
li
a
n
c
e
s.
Th
e
first
ste
p
is
p
re
-
p
ro
c
e
ss
in
g
,
wh
e
n
t
h
e
p
a
d
d
in
g
a
n
d
e
m
b
e
d
d
i
n
g
a
re
p
e
rfo
rm
e
d
.
Zero
p
a
d
d
i
n
g
a
n
d
e
n
d
p
a
d
d
i
n
g
a
re
u
se
d
f
o
r
th
e
p
a
d
d
i
n
g
.
At
th
e
sa
m
e
ti
m
e
,
d
iffere
n
t
e
m
b
e
d
d
i
n
g
tec
h
n
iq
u
e
s,
su
c
h
a
s
wo
r
d
2
v
e
c
,
G
lo
v
e
,
a
n
d
BERT
,
a
re
u
se
d
.
F
o
ll
o
win
g
p
re
-
p
ro
c
e
ss
in
g
,
t
h
e
fe
a
tu
re
s
a
re
e
x
tra
c
ted
fro
m
th
e
p
re
-
p
ro
c
e
ss
e
d
d
a
ta,
in
c
l
u
d
i
n
g
"
in
f
o
rm
a
ti
o
n
g
a
i
n
,
c
h
i
-
sq
u
a
re
,
m
u
tu
a
l
i
n
fo
rm
a
ti
o
n
,
a
n
d
sy
m
m
e
tri
c
a
l
u
n
c
e
rtain
t
y
-
b
a
se
d
fe
a
tu
re
s."
T
h
e
n
,
a
h
y
b
ri
d
o
p
ti
m
iza
ti
o
n
tec
h
n
i
q
u
e
k
n
o
w
n
a
s
c
lan
-
u
p
d
a
ted
g
re
y
wo
lf
o
p
ti
m
iz
a
ti
o
n
(CU
-
G
WO)
is
u
se
d
f
o
r
o
p
ti
m
ize
d
fe
a
tu
re
s
a
n
d
we
i
g
h
t
se
lec
ti
o
n
.
An
e
n
se
m
b
le
tec
h
n
i
q
u
e
wa
s
a
p
p
l
ied
to
e
x
trac
t
o
p
ti
m
a
l
fe
a
t
u
re
s.
T
h
e
c
las
sifiers
in
th
e
p
r
o
p
o
se
d
su
g
g
e
ste
d
e
n
se
m
b
le
tec
h
n
iq
u
e
with
d
e
e
p
c
o
n
v
o
lu
ti
o
n
n
e
u
ra
l
n
e
two
r
k
(DCN
N).
DCN
N
o
ffe
rs fi
n
e
we
ig
h
t
tu
n
in
g
a
n
d
d
e
tec
ti
o
n
re
su
l
t
s
.
Th
e
p
e
rfo
rm
a
n
c
e
a
n
a
ly
sis
a
n
d
it
s
imp
a
c
t
o
n
th
e
p
ro
p
o
se
d
m
o
d
e
l
fo
r
sa
rc
a
sm
d
e
tec
ti
o
n
a
re
c
las
sified
with
g
o
o
d
a
c
c
u
ra
c
y
in
to
sa
rc
a
stic
a
n
d
n
o
n
-
sa
rc
a
stic
c
a
teg
o
ries
.
Th
e
re
su
lt
s
a
re
a
lso
c
o
m
p
a
re
d
with
a
g
a
in
st t
h
o
se
o
f
t
h
e
G
lo
Ve
a
n
d
BERT
tec
h
n
i
q
u
e
s.
K
ey
w
o
r
d
s
:
D
e
e
p
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
E
m
b
ed
d
in
g
E
n
s
em
b
le
class
if
ier
Op
tim
al
f
ea
tu
r
es
Pad
d
in
g
Sar
ca
s
m
id
en
tific
atio
n
W
eig
h
t o
p
tim
izatio
n
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
:
Gee
ta
Sah
u
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atio
n
T
e
ch
n
o
lo
g
y
an
d
Data
Scien
ce
,
V
id
y
alan
k
ar
Sch
o
o
l o
f
I
n
f
o
r
m
at
io
n
T
ec
h
n
o
lo
g
y
Mu
m
b
ai
,
I
n
d
ia
E
m
ail: g
ee
ta.
s
ah
u
@
v
s
it
.
ed
u
.
in
1.
I
NT
RO
D
UCT
I
O
N
Sar
ca
s
m
ex
am
in
es
an
d
class
if
ies
an
in
d
i
v
id
u
al'
s
th
o
u
g
h
ts
,
s
en
tim
en
ts
,
an
d
em
o
ti
o
n
s
i
n
to
th
r
ee
ca
teg
o
r
ies:
p
o
s
itiv
e,
n
eg
ativ
e
,
an
d
n
e
u
tr
al.
Peo
p
le
co
m
m
u
n
icate
th
eir
s
en
tim
en
ts
an
d
f
ee
lin
g
s
in
v
a
r
io
u
s
way
s
.
So
cial
n
etwo
r
k
s
an
d
m
icr
o
b
lo
g
g
in
g
web
s
ites
f
r
eq
u
en
t
ly
em
p
lo
y
s
ar
ca
s
m
,
a
s
o
p
h
is
ti
ca
ted
k
in
d
o
f
ir
o
n
y
,
b
ec
au
s
e
th
ey
o
f
ten
p
r
o
m
o
te
t
r
o
llin
g
an
d
cr
iticis
m
o
f
o
th
e
r
u
s
er
s
.
Sar
ca
s
tic
ev
alu
atio
n
is
cr
itical
f
o
r
o
b
tain
in
g
m
ea
n
in
g
f
u
l
in
f
o
r
m
atio
n
f
r
o
m
am
o
r
p
h
o
u
s
d
ata
s
o
u
r
ce
s
s
u
ch
as
r
ev
iews
an
d
twee
ts
[
1
]
.
Sar
ca
s
m
is
a
s
en
tim
en
tality
ca
teg
o
r
y
u
s
ed
to
co
m
m
u
n
icate
g
o
o
d
,
n
eg
ati
v
e,
an
d
n
eu
tr
al
m
o
o
d
s
th
r
o
u
g
h
wo
r
d
s
,
te
x
t,
o
r
s
en
ten
ce
s
[
2
]
,
[
3
]
.
Sar
ca
s
m
is
o
f
ten
p
er
ce
iv
ed
as
a
lan
g
u
ag
e
elem
en
t
th
at
ch
ar
ac
ter
izes
o
n
lin
e
co
n
ten
t
an
d
ex
p
r
ess
es st
r
o
n
g
ly
h
eld
o
p
in
io
n
s
an
d
s
u
b
jectiv
ity
[
4
]
.
Sen
tim
en
t
an
aly
s
is
,
a
tech
n
iq
u
e
u
s
ed
in
ad
v
er
tis
in
g
an
d
o
p
i
n
io
n
m
in
in
g
,
is
v
alu
ab
le
f
o
r
d
eter
m
in
in
g
attitu
d
es.
T
h
er
ef
o
r
e,
s
ar
ca
s
m
d
etec
tio
n
p
r
ec
ed
es
th
at
o
f
th
e
p
r
im
ar
y
NL
P
d
ev
ices
[
5
]
.
User
s
ca
n
ex
ch
an
g
e
in
f
o
r
m
atio
n
an
d
th
o
u
g
h
ts
o
n
s
o
cial
m
ed
ia
p
latf
o
r
m
s
[
6
]
.
S
o
cial
m
ed
ia
u
s
er
s
ca
n
s
u
b
m
it
co
n
ten
t,
in
clu
d
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
2
7
-
5
0
3
7
5028
p
h
o
to
g
r
ap
h
s
,
v
i
d
eo
s
,
an
d
wo
r
d
s
,
r
eg
ar
d
in
g
an
y
s
itu
atio
n
to
ex
p
r
ess
th
eir
f
ee
lin
g
s
[
7
]
.
H
o
wev
er
,
n
eg
ativ
e
tex
t
co
m
m
en
ts
u
n
d
er
m
in
e
th
e
e
x
c
ellen
t
s
en
tim
en
t
o
f
s
o
cial
m
e
d
ia
u
s
er
s
.
Sar
ca
s
m
is
a
v
er
b
al
ex
p
r
ess
io
n
s
ty
le
in
wh
ich
wo
r
d
s
ar
e
u
s
ed
to
o
f
f
e
n
d
o
r
o
f
f
en
d
an
o
th
er
p
er
s
o
n
b
y
d
ep
ar
tin
g
f
r
o
m
th
eir
liter
a
l
m
ea
n
in
g
s
.
Peo
p
le
o
f
ten
b
ec
o
m
e
f
u
r
i
o
u
s
an
d
co
n
d
em
n
th
em
[
8
]
.
Sar
ca
s
m
ca
n
b
e
id
en
tifie
d
in
wr
itten
tex
t,
g
estu
r
es,
f
ac
ial
ex
p
r
ess
io
n
s
,
an
d
o
t
h
er
f
o
r
m
s
o
f
co
m
m
u
n
icatio
n
.
Sar
ca
s
m
is
a
co
m
b
in
atio
n
o
f
b
o
th
p
o
s
itiv
e
an
d
n
eg
ativ
e
o
b
s
er
v
atio
n
s
[
9
]
.
Sar
ca
s
m
'
s
in
ter
p
r
etatio
n
an
d
u
s
ag
e
ar
e
g
r
e
atly
in
f
lu
en
ce
d
b
y
th
e
cu
ltu
r
al
co
n
tex
t
in
wh
ich
it
is
em
p
lo
y
ed
.
T
h
is
m
ay
in
d
u
ce
v
ar
iatio
n
s
in
th
e
ex
am
in
atio
n
o
f
th
e
en
tire
p
o
lar
ity
an
d
alter
th
e
p
o
lar
ity
o
f
th
e
ex
p
r
ess
io
n
wh
ile
it
is
b
ein
g
ev
alu
ated
.
Sen
tim
en
t
an
aly
s
is
is
es
s
en
tial
f
o
r
s
elf
-
ex
am
in
in
g
co
n
d
u
ct
an
d
ass
es
s
in
g
s
en
tim
en
ts
ex
p
r
ess
ed
o
n
s
o
cial
m
e
d
ia
u
s
in
g
s
o
cial
lis
ten
in
g
tech
n
iq
u
es
[
1
0
]
.
Dete
ctin
g
s
ar
ca
s
m
in
u
n
s
tr
u
ctu
r
ed
tex
t
o
r
s
p
ee
ch
,
s
u
ch
as
in
s
tr
u
ctio
n
s
,
b
lo
g
s
,
co
m
m
en
ts
o
n
in
d
iv
id
u
als
o
r
e
v
en
ts
,
a
n
d
p
r
o
d
u
ct
o
r
s
er
v
ice
e
v
alu
atio
n
s
,
is
k
n
o
wn
as
s
ar
ca
s
m
d
etec
tio
n
.
I
t
also
e
n
h
an
ce
s
h
u
m
an
-
m
ac
h
in
e
co
m
m
u
n
icatio
n
ef
f
icac
y
b
y
p
r
o
v
i
d
in
g
i
n
s
ig
h
ts
in
to
an
in
d
i
v
id
u
al'
s
em
o
tio
n
s
,
p
s
y
ch
o
l
o
g
y
,
an
d
,
o
cc
asio
n
ally
ev
e
n
h
ea
lth
.
Sen
tim
en
t
ty
p
es
ar
e
well
d
ef
in
ed
:
n
o
m
atter
wh
o
m
y
o
u
ask
o
r
wh
at
lan
g
u
ag
e
y
o
u
u
s
e,
l
o
v
e
is
alwa
y
s
a
p
o
s
itiv
e
s
en
tim
en
t,
an
d
h
atr
ed
is
al
way
s
a
n
eg
ativ
e
s
en
tim
en
t
[
1
1
]
.
Hash
tag
s
o
f
ten
ac
co
m
p
an
y
a
s
ar
ca
s
tically
wo
r
d
ed
s
en
ten
ce
.
Fo
r
in
s
tan
ce
,
"I
lo
v
e
b
o
r
in
g
f
o
o
d
.
#
n
o
t."
Say
in
g
"I
ad
o
r
e
b
lan
d
f
o
o
d
"
i
n
th
is
c
o
n
tex
t
m
ig
h
t
n
o
t
b
e
ir
o
n
ic.
Ho
we
v
er
,
th
er
e
is
s
ar
ca
s
m
in
"
#
n
o
t,"
NL
P
o
u
g
h
t to
m
ak
e
th
e
d
ata
v
is
ib
le
u
s
in
g
h
ash
tag
s
.
T
h
er
ef
o
r
e,
it is
es
s
en
tial to
r
ec
o
g
n
ize
ir
o
n
y
in
ev
er
y
d
a
y
co
n
tact
an
d
d
is
co
u
r
s
e
to
a
v
o
id
m
is
u
n
d
er
s
tan
d
in
g
s
an
d
e
n
h
an
ce
s
en
ti
m
en
tal
m
ea
n
in
g
[
1
2
]
.
Sar
ca
s
m
d
etec
tio
n
h
as b
ee
n
ac
h
iev
ed
u
s
in
g
n
u
m
er
o
u
s
m
a
ch
in
e
lear
n
in
g
an
d
ar
tific
ial
in
tellig
en
ce
alg
o
r
ith
m
s
,
s
u
ch
as
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
,
r
a
n
d
o
m
f
o
r
es
t
(
RF
)
,
co
n
v
o
lu
tio
n
n
e
u
r
al
n
et
wo
r
k
(
C
NN)
,
an
d
n
eu
r
al
n
etw
o
r
k
(
NN
)
[
1
3
]
.
T
h
ey
p
r
o
p
o
s
ed
a
b
id
ir
ec
tio
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
B
i
-
L
STM
)
m
o
d
el
to
id
e
n
tify
s
ar
c
asm
.
T
h
e
d
ataset
was
ac
q
u
ir
ed
f
r
o
m
Ka
g
g
le
[
1
4
]
.
New
s
h
ea
d
lin
es
wer
e
in
clu
d
e
d
i
n
th
e
co
llectio
n
.
T
h
e
d
ataset
was
p
r
e
-
p
r
o
ce
s
s
ed
b
y
r
em
o
v
in
g
s
p
ec
ial
s
y
m
b
o
ls
,
lo
wer
in
g
,
s
to
p
p
in
g
wo
r
d
r
em
o
v
al,
to
k
e
n
izatio
n
,
an
d
p
u
n
ctu
atio
n
,
f
o
llo
wed
b
y
w
o
r
d
em
b
ed
d
i
n
g
to
o
b
tain
wo
r
d
v
ec
to
r
s
with
Glo
Ve.
T
h
e
v
ec
to
r
s
g
e
n
er
ated
u
s
in
g
th
is
m
eth
o
d
wer
e
f
ed
in
to
th
e
B
i
-
L
STM
m
o
d
el.
P
r
o
p
o
s
ed
a
n
o
v
el
en
s
em
b
le
m
eth
o
d
b
ased
o
n
tex
t
em
b
ed
d
i
n
g
u
s
in
g
f
u
zz
y
ev
o
lu
tio
n
ar
y
lo
g
ic
in
th
e
to
p
lay
er
.
T
h
is
m
eth
o
d
u
tili
ze
s
f
u
zz
y
lo
g
ic
to
en
s
em
b
le
e
m
b
ed
d
in
g
f
r
o
m
th
e
wo
r
d
to
v
ec
to
r
(
W
o
r
d
2
v
ec
)
,
g
lo
b
al
v
ec
t
o
r
s
f
o
r
wo
r
d
r
ep
r
es
en
tatio
n
(
Glo
Ve
)
,
an
d
b
i
d
ir
ec
t
io
n
al
en
co
d
er
r
ep
r
esen
tatio
n
s
f
r
o
m
t
r
an
s
f
o
r
m
e
r
s
(
B
E
R
T
)
m
o
d
els
b
ef
o
r
e
d
ec
id
i
n
g
o
n
th
e
u
ltima
te
ca
teg
o
r
izat
io
n
[
1
5
]
.
T
h
e
th
r
ee
s
o
cial
m
e
d
ia
d
atasets
u
s
ed
to
v
alid
ate
th
e
s
u
g
g
ested
m
o
d
el
wer
e
th
e
h
ea
d
lin
es
d
ataset,
th
e
"
s
elf
-
an
n
o
tated
r
ed
d
it
co
r
p
u
s
"
(
SA
R
C
)
,
an
d
th
e
X
(
f
o
r
m
er
l
y
T
witter
)
ap
p
d
ataset.
Acc
u
r
ac
y
r
ates
o
f
9
0
.
8
1
%,
8
5
.
3
8
%,
an
d
8
6
.
8
0
%
wer
e
o
b
tain
ed
,
r
esp
ec
tiv
ely
[
1
6
]
,
[
1
7
]
.
A
n
o
v
el
m
eth
o
d
f
o
r
im
p
lem
en
tin
g
a
h
y
b
r
id
o
p
tim
izatio
n
s
tr
ate
g
y
h
elp
s
to
id
en
tify
s
ar
ca
s
m
[
1
8
]
.
C
o
n
tex
tu
alizin
g
a
ter
m
d
em
an
d
s
f
lu
en
c
y
th
r
o
u
g
h
ar
ticu
latio
n
[
1
9
]
.
T
y
p
ically
,
s
en
tim
en
t
ch
ar
ac
ter
is
tics
ar
e
m
an
u
ally
b
u
ilt u
s
in
g
a
s
ar
ca
s
m
-
d
etec
tio
n
alg
o
r
ith
m
[
2
0
]
.
Dee
p
f
ak
e
v
id
eo
d
etec
to
r
s
ex
am
in
e
v
o
ice
to
n
es,
lip
-
s
y
n
c,
f
a
cial
ex
p
r
ess
io
n
s
,
an
d
m
icr
o
ex
p
r
ess
io
n
s
.
Sar
ca
s
m
d
etec
to
r
s
,
o
n
th
e
o
th
er
h
an
d
,
e
x
am
in
e
tex
tu
al
c
o
n
t
en
t,
to
n
e
,
g
estu
r
es,
an
d
f
ac
ial
ex
p
r
ess
io
n
s
,
am
o
n
g
o
th
er
th
in
g
s
.
C
U
-
GW
O
o
p
tim
izatio
n
tech
n
iq
u
es
aid
i
n
s
elec
tin
g
im
p
o
r
tan
t
ch
a
r
ac
ter
is
tics
an
d
f
in
e
-
t
u
n
in
g
m
o
d
el
p
ar
am
eter
s
.
Dis
am
b
ig
u
atio
n
:
d
e
p
en
d
i
n
g
o
n
th
e
c
o
n
tex
t,
wo
r
d
s
lik
e
"g
r
ea
t,"
"lo
v
e,
"
o
r
"f
in
e"
m
ig
h
t
h
av
e
q
u
ite
d
if
f
er
en
t
m
ea
n
in
g
s
.
ca
p
tu
r
i
n
g
s
y
n
tactic
an
d
s
eq
u
en
tial
d
ep
en
d
en
cies:
co
n
tex
tu
al
an
d
tr
an
s
f
o
r
m
er
m
o
d
els ar
e
ad
ju
s
ted
ac
c
o
r
d
in
g
o
n
p
latf
o
r
m
-
s
p
ec
if
ic
in
f
o
r
m
at
io
n
.
T
h
e
ap
p
r
o
ac
h
em
p
lo
y
ed
is
d
y
n
am
ic
co
n
tex
tu
al
m
o
d
u
lati
o
n
an
d
e
m
o
tio
n
-
e
m
b
ed
d
ed
v
ec
to
r
s
to
ca
p
tu
r
e
th
e
tex
t'
s
em
o
tio
n
al
co
n
ten
t
,
b
o
th
th
e
m
o
d
el'
s
co
n
tex
tu
al
ad
a
p
tatio
n
a
n
d
t
h
e
h
ier
ar
ch
ical
atten
tio
n
m
ec
h
an
is
m
f
o
r
tex
t
s
eg
m
e
n
tatio
n
.
T
h
e
m
o
d
el
o
b
tain
e
d
a
n
F1
-
s
co
r
e
o
f
0
.
9
0
an
d
a
n
ac
c
u
r
ac
y
o
f
8
9
%
o
n
th
e
Mu
s
tar
d
d
ataset.
P
r
o
p
o
s
ed
m
o
d
el
u
tili
ze
s
th
e
n
at
u
r
e
i
n
s
p
ir
ed
s
war
m
-
b
ased
b
io
h
y
b
r
id
o
p
tim
izatio
n
tech
n
iq
u
es
,
m
in
g
led
-
elep
h
an
t
h
er
d
in
g
&
g
r
ey
wo
lf
o
p
ti
m
izatio
n
(
GW
O)
,
i.e
.
C
lan
u
p
d
ated
GW
O
.
T
h
e
p
o
s
itiv
e
Me
tr
ic
-
F_
m
ea
s
u
r
e
ev
alu
ated
o
v
er
W
o
r
d
2
v
ec
,
Glo
v
e,
an
d
B
E
R
T
ar
e
9
3
%,
9
1
%,
an
d
9
0
%,
r
esp
ec
tiv
ely
.
T
h
is
ar
ticle
u
s
es
au
d
io
d
ata
f
r
o
m
s
p
o
n
ta
n
eo
u
s
,
r
ea
l
-
wo
r
ld
,
m
o
n
o
lin
g
u
al
d
at
asets
to
attem
p
t
to
d
etec
t
s
ar
ca
s
m
in
s
p
ee
ch
.
I
r
o
n
y
,
ex
ag
g
e
r
atio
n
,
s
u
b
tlety
,
s
em
an
tics
,
an
d
p
r
ag
m
atics
ar
e
n
o
t
tak
en
in
to
ac
co
u
n
t.
T
h
e
d
ataset
is
s
m
all,
an
d
th
er
e
is
n
o
b
aselin
e
co
m
p
ar
is
o
n
.
T
h
is
wo
r
k
p
r
esen
ts
s
ar
ca
s
m
i
d
en
tific
atio
n
f
r
o
m
s
o
cial
m
ed
ia
u
s
in
g
th
e
in
n
o
v
ativ
e
C
U
-
GW
O
-
b
ased
d
ee
p
en
s
em
b
le
tech
n
iq
u
e
,
a
h
y
b
r
id
n
atu
r
e
in
s
p
ir
e
d
s
war
m
b
ased
o
p
tim
izatio
n
tech
n
i
q
u
e
,
i
.
e.
elep
h
an
t
h
er
d
in
g
o
p
tim
izatio
n
(
E
HO)
+
GW
O
,
with
p
r
e
-
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
s
u
ch
as
em
b
ed
d
in
g
(
W
o
r
d
2
v
ec
,
Glo
v
e
,
an
d
B
E
R
T
)
an
d
p
ad
d
in
g
(
ze
r
o
an
d
en
d
)
,
f
ea
tu
r
e
ex
tr
ac
tio
n
wit
h
o
p
tim
al
f
ea
tu
r
e
s
elec
tio
n
,
f
o
llo
wed
b
y
en
s
em
b
le
class
if
ier
with
R
F,
SVM
,
an
d
NN,
th
e
o
u
tp
u
t
o
f
th
ese
clas
s
if
ier
s
g
iv
en
as
i
n
p
u
t
to
d
ee
p
co
n
v
o
l
u
tio
n
n
eu
r
al
n
etwo
r
k
(
DC
NN)
,
weig
h
t
o
p
tim
izatio
n
is
d
o
n
e
an
d
DC
NN
p
r
o
d
u
ce
s
th
e
d
esire
d
r
esu
lt.
C
o
m
p
ar
in
g
th
e
m
o
d
el’
s
r
esu
lts
at
v
ar
io
u
s
s
tag
es
with
an
d
with
o
u
t
f
ea
t
u
r
e,
weig
h
t
o
p
tim
izatio
n
an
d
with
f
ea
tu
r
e,
weig
h
t
o
p
tim
izatio
n
.
T
h
e
ev
al
u
atio
n
o
f
p
er
f
o
r
m
an
ce
f
o
r
s
ar
ca
s
m
ty
p
e
d
etec
tio
n
f
r
o
m
s
o
cial
m
ed
ia
u
s
in
g
em
b
ed
d
in
g
an
d
p
a
d
d
in
g
tech
n
iq
u
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Meta
h
eu
r
is
tic
o
p
timiz
a
tio
n
fo
r
s
a
r
ca
s
m
d
etec
tio
n
in
s
o
cia
l m
ed
ia
w
ith
…
(
Gee
ta
S
a
h
u
)
5029
2.
M
E
T
H
O
D
S
tep
s
in
id
en
tify
in
g
s
ar
ca
s
m
f
r
o
m
tex
t
a
r
e
p
r
e
-
p
r
o
ce
s
s
in
g
,
em
b
e
d
d
in
g
an
d
p
ad
d
in
g
t
ec
h
n
iq
u
es,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
id
ea
l
f
ea
tu
r
e
s
elec
tio
n
,
o
p
tim
ized
wei
g
h
t
s
elec
tio
n
,
an
d
d
ee
p
lear
n
in
g
-
b
ased
d
etec
tio
n
m
eth
o
d
s
.
I
n
th
e
in
itial
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e,
to
k
en
izatio
n
wa
s
o
n
th
e
d
ataset.
T
h
e
k
ey
wo
r
d
s
ar
e
th
en
ex
tr
ac
te
d
f
r
o
m
ea
ch
d
o
m
ai
n
.
T
h
is
is
f
o
ll
o
wed
b
y
e
m
b
ed
d
in
g
,
s
u
ch
as Wo
r
d
2
v
ec
,
Glo
v
e
,
an
d
B
E
R
T
,
k
n
o
wn
as
t
h
e
wo
r
d
to
v
ec
to
r
,
wh
ich
is
u
s
ed
,
f
o
llo
wed
b
y
ze
r
o
an
d
en
d
p
ad
d
in
g
.
M
an
y
f
ea
tu
r
es a
r
e
ex
tr
ac
ted
,
s
u
ch
as sy
m
m
etr
ical
u
n
ce
r
tain
ty
-
b
ased
f
ea
tu
r
es,
m
u
tu
al
in
f
o
r
m
atio
n
,
c
h
i
-
s
q
u
ar
e
,
an
d
in
f
o
r
m
atio
n
g
ain
.
T
h
e
b
es
t
ch
ar
ac
ter
is
tics
ar
e
f
ed
in
to
a
co
m
b
in
atio
n
ap
p
r
o
ac
h
th
at
co
m
b
in
es
SVM,
NN,
DC
NN,
an
d
R
F
to
d
etec
t
s
ar
ca
s
tic
s
ta
tem
en
ts
in
th
e
in
p
u
t
tex
t.
T
h
is
wo
r
k
s
u
g
g
ests
a
n
ew
C
U
-
GW
O
h
y
b
r
id
m
o
d
el
(
E
HO+
GW
O)
th
at
in
clu
d
es
weig
h
t
ad
ju
s
tm
en
t
u
s
in
g
a
DC
NN,
s
ea
r
ch
in
g
f
o
r
o
p
tim
al
f
ea
tu
r
es.
T
h
e
o
b
jectiv
e
f
u
n
ctio
n
o
f
th
e
DC
NN
m
o
d
el
is
d
eter
m
in
ed
with
f
ea
tu
r
es
a
n
d
weig
h
ts
,
as
g
iv
en
in
(
1
)
.
Fig
u
r
e
1
d
e
p
icts
th
e
ar
ch
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
s
ar
ca
s
m
d
etec
tio
n
s
y
s
tem
.
=
(
)
(
1
)
Fig
u
r
e
1
.
Ar
c
h
itectu
r
e
o
f
th
e
s
u
g
g
ested
s
ar
ca
s
m
d
etec
tio
n
s
y
s
tem
2
.
1
.
P
re
-
pro
ce
s
s
ing
Pre
-
p
r
o
ce
s
s
in
g
s
tep
s
ap
p
lied
to
th
e
d
ata:
clea
n
in
g
te
x
t
.
On
e
m
ajo
r
d
is
ad
v
a
n
tag
e
o
f
u
s
in
g
X
d
ata
s
ets
is
th
e
n
o
is
e
in
th
e
d
ata.
User
m
en
tio
n
s
(
@
u
s
er
)
,
u
n
if
o
r
m
r
e
s
o
u
r
ce
lo
ca
to
r
r
e
f
er
en
ce
s
(
UR
L
)
,
in
tr
o
d
u
cto
r
y
tex
t,
an
d
co
n
ten
t
tag
s
(
#
)
,
co
m
m
o
n
ly
r
ef
er
r
ed
to
as
h
ash
tag
s
,
m
ak
e
u
p
X
d
ata,
o
r
twee
ts
.
I
n
th
is
s
tep
,
X
d
ata
is
p
r
e
-
p
r
o
ce
s
s
ed
b
ef
o
r
e
f
ea
t
u
r
e
ex
tr
ac
tio
n
an
d
class
if
icatio
n
.
Ad
d
itio
n
al
b
asic
p
r
e
-
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
th
at
ch
an
g
e
th
e
i
n
p
u
t
tex
t
t
o
lo
w
er
ca
s
e
in
clu
d
e
to
k
e
n
izatio
n
,
s
to
p
wo
r
d
r
em
o
v
al,
an
d
p
a
r
ts
o
f
s
p
ee
c
h
(
POS)
tag
g
in
g
.
Key
wo
r
d
ex
t
r
ac
tio
n
an
d
s
to
p
w
o
r
d
r
em
o
v
al:
i
n
th
is
in
itial
s
tep
,
ea
c
h
d
o
m
a
in
'
s
s
to
p
wo
r
d
s
ar
e
elim
in
ated
b
ef
o
r
e
t
h
e
k
e
y
wo
r
d
s
ar
e
r
ec
o
v
e
r
ed
.
T
o
k
e
n
izatio
n
:
s
p
ec
if
ic
wo
r
d
s
o
r
p
h
r
as
es
ar
e
d
e
n
o
ted
b
y
to
k
en
s
.
T
o
k
e
n
i
za
ti
o
n
is
u
s
e
d
t
o
d
iv
id
e
a
t
ex
t
s
tr
ea
m
i
n
t
o
d
is
c
r
e
te
t
o
k
e
n
s
.
V
ec
t
o
r
i
za
t
io
n
is
t
h
e
p
r
o
c
ess
o
f
r
e
p
r
ese
n
t
in
g
ea
ch
s
e
n
te
n
c
e
o
r
t
ex
t
as a
v
ec
t
o
r
,
w
h
e
r
e
ea
c
h
ele
m
e
n
t r
e
p
r
ese
n
ts
a
w
o
r
d
i
n
t
h
e
v
o
ca
b
u
l
ar
y
a
n
d
t
h
e
v
al
u
e
o
f
e
ac
h
co
m
p
o
n
en
t
i
n
d
i
ca
t
es
w
h
e
r
e
t
h
at
w
o
r
d
a
p
p
ea
r
s
in
t
h
e
d
o
c
u
m
e
n
t
.
T
h
is
is
s
o
m
eti
m
es
r
e
f
er
r
e
d
to
as
ter
m
f
r
e
q
u
e
n
c
y
(
T
F
)
.
E
.
g
.,
"I
c
an
n
o
t
b
eli
ev
e
y
o
u
d
i
d
t
h
at
!
"
c
a
n
b
e
ex
p
r
ess
e
d
i
n
te
r
m
s
o
f
[
0
,
1
,
1
,
0
,
0
,
1
,
1
]
.
2
.
2
.
F
e
a
t
ure
ex
t
r
a
ct
io
n
Featu
r
e
ex
tr
ac
tio
n
p
r
o
ce
s
s
:
s
i
n
ce
th
e
f
ea
tu
r
es
ar
e
s
elec
ted
a
n
d
ex
tr
ac
ted
ac
c
o
r
d
i
n
g
to
h
o
w
well
th
ey
f
it
with
in
th
e
d
ata
ty
p
e,
ch
i
-
s
q
u
ar
e,
in
f
o
r
m
atio
n
g
ain
,
a
n
d
r
elate
d
f
ea
tu
r
e
s
.
T
o
id
e
n
tify
s
ar
ca
s
m
,
th
e
s
u
b
tle
lin
g
u
is
tic,
s
y
n
tactic
,
s
em
an
tic,
an
d
p
r
ag
m
atic
i
n
f
o
r
m
atio
n
p
r
esen
ted
in
a
h
ig
h
-
d
im
en
s
io
n
s
p
ac
e
is
ex
am
in
ed
.
C
h
i
-
s
q
u
ar
e
ev
alu
ates
th
e
d
e
g
r
ee
o
f
in
d
ep
en
d
en
ce
b
etwe
e
n
a
f
ea
tu
r
e
an
d
th
e
ta
r
g
et
v
a
r
iab
l
e.
B
y
m
ea
s
u
r
in
g
t
h
e
r
ed
u
ctio
n
i
n
en
tr
o
p
y
,
f
ea
tu
r
es
ar
e
ex
p
l
o
ited
,
an
d
in
f
o
r
m
ati
o
n
g
ain
is
q
u
an
tifie
d
.
T
h
e
f
o
llo
win
g
ex
tr
ac
tio
n
m
eth
o
d
s
ar
e
u
tili
ze
d
.
2
.
2
.
1
.
Chi
-
s
qu
a
re
(
χ²
):
C
hi
-
s
q
u
ar
e,
wh
o
s
e
v
alu
e
is
d
eter
m
in
ed
b
y
d
iv
i
d
in
g
th
e
tar
g
et
f
ea
tu
r
e
b
y
th
e
r
em
ain
in
g
f
e
atu
r
e
[
2
1
]
.
T
h
er
ef
o
r
e,
in
(
2
)
ex
p
r
ess
es
d
eter
m
in
in
g
th
e
b
est
ch
i
-
s
q
u
ar
e
v
alu
e/f
ea
tu
r
e
(
χ²
f
)
an
d
c
h
o
o
s
in
g
th
e
attr
ib
u
tes
wh
er
e
ℎ
o
b
s
er
v
ed
f
r
e
q
u
en
c
y
an
d
is
th
e
ex
p
ec
ted
f
r
eq
u
e
n
cy
[
2
2
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
2
7
-
5
0
3
7
5030
²
=
(
−
)
2
(
2
)
2
.
2
.
2
.
M
utua
l
info
rm
a
t
io
n
T
h
e
co
m
p
u
tatio
n
o
f
th
e
s
h
ar
e
d
in
f
o
r
m
atio
n
b
etwe
en
two
co
m
b
in
atio
n
s
o
f
r
a
n
d
o
m
v
ar
ia
b
les,
a
an
d
b
,
is
ca
lled
th
e
m
u
tu
al
in
f
o
r
m
at
io
n
f
ea
tu
r
e
(
MI
f
)
.
T
h
e
MI
f
a
r
e
ex
tr
ac
ted
an
d
s
h
o
wn
in
(
3
)
,
wh
er
e
is
th
e
p
r
o
b
a
b
ilit
y
.
=
∑
[
(
,
)
×
2
(
(
,
)
(
)
×
(
)
)
]
(
3
)
2
.
2
.
3
.
I
nfo
rm
a
t
io
n
g
a
in
f
e
a
t
ures
I
n
a
an
d
b
a
r
e
r
an
d
o
m
v
ar
i
ab
les,
E
is
en
tr
o
p
y
,
an
d
w
is
weig
h
t
[
2
3
]
.
T
h
e
ex
t
r
ac
ted
f
ea
tu
r
e
i
n
f
o
r
m
atio
n
g
ain
f
ea
tu
r
es
(IG
f
)
ar
e
ca
lcu
lated
u
s
in
g
(
4
)
an
d
is
g
iv
en
as:
=
(
)
−
(
×
(
)
)
(
4
)
2
.
2
.
4
.
F
ea
t
ures
wit
h sy
mm
et
ric
un
ce
rt
a
inty
ba
s
ed
I
t
is
ex
p
ec
ted
,
an
d
th
e
e
x
tr
ac
tio
n
o
f
h
o
w
f
ea
tu
r
es
ar
e
ex
t
r
ac
ted
is
s
h
o
wn
in
(
5
)
,
wh
e
r
e
E
d
en
o
tes
th
e
en
tr
o
p
y
.
SU
f
d
en
o
tes th
e
f
ea
tu
r
es e
x
tr
ac
ted
b
ased
o
n
s
y
m
m
et
r
ic
u
n
ce
r
tain
ty
(
SU)
.
=
2
×
(
(
,
)
(
)
∗
(
)
)
(
5
)
Fin
ally
,
th
e
to
tal
o
f
all
th
e
ex
tr
ac
ted
f
ea
tu
r
es,
wh
ich
in
clu
d
e
th
e
ch
i
-
s
q
u
ar
e,
m
u
tu
al
in
f
o
r
m
atio
n
,
in
f
o
r
m
atio
n
g
ain
,
an
d
s
y
m
m
etr
ic
u
n
ce
r
tain
ty
-
b
ased
f
ea
tu
r
es,
is
r
ep
r
esen
t
ed
b
y
t
h
e
letter
F,
wh
ich
ca
n
b
e
f
o
u
n
d
in
(
6
)
.
=
[
²
+
+
+
]
(
6
)
I
n
ter
ac
tio
n
with
th
e
o
p
tim
iza
tio
n
tech
n
i
q
u
e:
f
o
r
th
e
f
ea
tu
r
e
o
p
tim
izatio
n
d
ataset,
th
e
C
U
-
GW
O
m
o
d
el
is
u
tili
s
ed
.
I
t
g
o
es
th
r
o
u
g
h
a
p
r
ep
r
o
ce
s
s
in
g
p
h
ase
f
ir
s
t.
Fo
llo
win
g
th
e
ex
tr
ac
tio
n
o
f
all
th
e
co
m
p
o
n
e
n
ts
,
ce
r
tain
tr
aits
o
b
s
tr
u
ct
th
e
tr
ai
n
in
g
.
W
h
at
m
ig
h
t
b
e
r
ef
e
r
r
ed
to
as
th
e
“c
u
r
s
e
o
f
d
im
en
s
io
n
ality
”
af
f
ec
ts
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
.
B
alan
ce
b
etwe
en
ex
p
lo
r
atio
n
an
d
e
x
p
lo
itatio
n
:
C
U
-
GW
O
im
p
r
o
v
es
th
e
e
q
u
ilib
r
iu
m
b
etwe
en
ex
p
lo
itatio
n
(
id
e
n
tify
in
g
th
e
b
est
s
o
lu
tio
n
with
th
e
least
lo
s
s
an
d
m
ax
im
izin
g
m
o
d
el
p
er
f
o
r
m
an
ce
)
an
d
ex
p
lo
r
atio
n
(
f
in
d
i
n
g
s
ev
er
al
s
o
lu
tio
n
s
)
.
I
t a
cc
eler
ates th
e
co
n
v
er
g
en
ce
r
ate.
2
.
3
.
O
pti
m
a
l
f
e
a
t
ure
s
elec
t
io
n a
nd
deep
en
s
em
ble c
la
s
s
if
i
er
t
ec
hn
iqu
e
f
o
r
s
a
rc
a
s
m
re
co
g
nitio
n
T
h
e
tr
ain
in
g
p
r
o
ce
d
u
r
e
b
ec
a
m
e
o
v
er
b
u
r
d
en
ed
with
p
er
tin
en
t
an
d
u
n
n
ec
ess
ar
y
d
ata
f
o
r
all
d
er
iv
ed
f
ea
tu
r
es.
T
h
e
ter
m
‘
c
u
r
s
e
o
f
d
im
en
s
io
n
ality
’
h
as
b
ec
o
m
es
an
is
s
u
e.
T
h
is
s
tu
d
y
p
ap
er
p
r
o
v
id
es
a
n
ew
m
eta
-
h
eu
r
is
tic
-
b
ased
b
io
-
in
s
p
ir
e
d
h
y
b
r
id
o
p
tim
izatio
n
ap
p
r
o
ac
h
(
C
U
-
GW
O)
to
d
eter
m
in
e
th
e
b
est
ch
ar
ac
ter
is
tics
.
Fig
u
r
e
2
illu
s
tr
ates th
e
b
est an
d
o
p
tim
al
f
ea
tu
r
e
s
elec
tio
n
.
Fig
u
r
e
2
.
Op
tim
al
f
ea
tu
r
e
s
ele
ctio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Meta
h
eu
r
is
tic
o
p
timiz
a
tio
n
fo
r
s
a
r
ca
s
m
d
etec
tio
n
in
s
o
cia
l m
ed
ia
w
ith
…
(
Gee
ta
S
a
h
u
)
5031
2
.
4
.
E
ns
em
ble
cla
s
s
if
iers
T
h
e
d
ev
elo
p
e
d
en
s
em
b
le
tech
n
iq
u
e
u
s
es
v
ar
io
u
s
class
if
ier
s
,
in
clu
d
in
g
D
C
NN,
NN,
SV
M
,
an
d
R
F.
T
h
ese
class
if
ier
s
wer
e
ass
ig
n
ed
g
iv
e
n
th
e
b
est
f
e
atu
r
e
s
elec
t
io
n
.
T
h
e
class
if
ier
r
esu
lt
is
th
e
n
s
en
t
to
th
e
C
NN.
T
h
e
f
in
al
d
etec
tio
n
r
esu
lts
wer
e
ass
ess
ed
[
2
4
]
.
Fig
u
r
e
3
i
llu
s
tr
ates
th
e
en
s
em
b
le
class
i
f
ier
an
d
o
p
tim
is
ed
DC
NN
c
la
s
s
if
ier
.
RF
:
a
class
if
icatio
n
f
r
am
ewo
r
k
u
s
es
s
ev
er
al
wea
k
class
if
ier
s
to
cr
ea
te
a
r
o
b
u
s
t
class
if
ier
.
T
h
e
R
F
m
o
d
el
en
s
u
r
es
r
a
n
d
o
m
v
,
an
d
C
AR
T
DT
u
s
es
ev
er
y
f
ea
tu
r
e.
C
o
m
p
a
r
ed
with
DT
,
th
e
R
F
m
o
d
el
o
f
f
er
s
b
etter
class
if
icatio
n
ac
cu
r
ac
y
.
SVM
is
a
n
o
n
lin
ea
r
m
a
p
p
in
g
t
h
at
in
cr
ea
s
es
th
e
m
ag
n
itu
d
e
o
f
tr
ain
in
g
d
ata
f
r
o
m
ac
tu
al
tr
ain
in
g
d
ata.
I
t
lo
o
k
s
f
o
r
th
is
lin
ea
r
o
p
tim
al
h
y
p
er
p
l
an
e
o
f
th
is
ex
tr
a
d
im
en
s
io
n
.
I
n
(
7
)
ex
p
r
ess
es
th
e
ass
es
s
m
en
t
o
f
th
e
h
y
p
er
p
lan
e
in
two
s
ep
ar
ab
le
class
lin
ea
r
p
r
o
b
lem
s
,
wh
er
e
N
V
is
th
e
av
er
ag
e
v
ec
to
r
an
d
h
p
is
th
e
d
is
tan
ce
b
etwe
en
th
e
o
r
i
g
in
s
an
d
h
y
p
er
p
lan
es.
SVM
d
en
o
tes th
e
o
u
tp
u
t o
f
th
e
SVM
.
=
+
ℎ
=
0
(
7
)
Neu
r
al
n
etwo
r
k
:
i
n
p
u
t
f
o
r
th
i
s
is
th
e
id
ea
l
f
ea
tu
r
e
F,
d
ef
i
n
ed
b
y
(
8
)
,
wh
er
e
F
fc
d
en
o
tes
th
e
o
v
er
all
f
ea
tu
r
e
co
u
n
t.
=
[
+
+
+
⋯
.
+
]
(
8
)
Fig
u
r
e
3
.
E
n
s
em
b
le
class
if
ier
an
d
o
p
tim
is
ed
DC
NN
class
if
ie
r
2
.
5
.
E
nh
a
nced
deep
co
nv
o
lu
t
io
n neura
l net
wo
rk
cla
s
s
if
ica
t
io
n
T
h
e
f
ea
tu
r
e
ass
ess
m
en
t
at
th
at
p
o
s
itio
n
(
x
,
y
)
in
t
h
e
co
r
r
esp
o
n
d
in
g
f
ea
tu
r
e
m
ap
an
d
r
th
lay
er
is
ex
p
r
ess
ed
b
y
(
9
)
,
wh
e
r
e
P.
,
E
r
ep
r
esen
ts
th
e
p
o
o
lin
g
lay
er
f
u
n
ctio
n
with
a
f
ix
ed
in
p
u
t
at
th
e
lay
er
p
o
s
itio
n
(
x
,
y
)
.
,
,
=
*
P
,
+
(
9
)
I
n
teg
r
atio
n
o
f
C
U
-
GW
O
with
DC
NN:
t
o
r
ef
lect
th
e
d
etec
tio
n
o
f
s
ar
ca
s
m
in
th
e
o
u
tp
u
t,
f
ea
tu
r
es
-
lex
ical
cu
es,
s
en
tim
en
t,
an
d
co
n
tex
tu
al
in
f
o
r
m
atio
n
n
ee
d
to
b
e
ap
p
r
o
p
r
i
ately
weig
h
ted
.
Dim
en
s
io
n
ality
r
ed
u
ctio
n
is
d
o
n
e
wh
er
e
weig
h
t
s
elec
tio
n
h
elp
s
less
en
th
e
im
p
ac
t
o
f
r
ed
u
n
d
a
n
t,
u
n
n
ec
ess
ar
y
in
f
o
r
m
atio
n
t
h
at
co
u
ld
in
tr
o
d
u
ce
n
o
is
e
in
to
th
e
d
etec
tio
n
p
r
o
ce
s
s
.
E
x
p
lo
r
atio
n
an
d
e
x
p
lo
itatio
n
:
C
U
-
GW
O
im
p
r
o
v
es
th
e
e
q
u
ilib
r
iu
m
b
etwe
en
ex
p
lo
itatio
n
(
id
en
tif
y
in
g
th
e
b
est
s
o
lu
tio
n
with
th
e
least
lo
s
s
an
d
m
a
x
im
izin
g
m
o
d
el
p
er
f
o
r
m
an
ce
)
an
d
ex
p
lo
r
atio
n
(
id
e
n
tify
in
g
s
ev
e
r
al
s
o
lu
tio
n
s
)
.
I
t
ac
ce
ler
ates
th
e
co
n
v
er
g
e
n
ce
r
ate
an
d
clea
r
o
f
o
v
er
f
itti
n
g
p
r
o
b
lem
s
.
2
.
6
.
P
r
o
po
s
ed
CU
-
G
WO
a
lg
o
rit
hm
f
o
r
t
he
o
ptim
iza
t
io
n
pro
ce
s
s
T
h
e
m
o
s
t
cr
iti
ca
l
wo
lv
es
i
n
t
h
e
h
u
n
ti
n
g
p
r
o
ce
s
s
ar
e
r
e
p
r
ese
n
t
e
d
b
y
al
p
h
a
(
α
)
,
b
et
a
(
β
)
,
g
am
m
a
(
¥
)
,
an
d
d
el
ta
(
§
)
.
L
ea
d
e
r
α
i
n
t
h
is
g
r
o
u
p
ca
n
d
e
ci
d
e
ab
o
u
t
t
h
e
h
u
n
ti
n
g
p
r
o
c
ess
,
w
h
i
le
w
o
l
v
es
β
in
th
e
s
ec
o
n
d
a
n
d
th
ir
d
p
o
s
iti
o
n
s
¥
s
u
p
p
o
r
t t
h
e
l
ea
d
e
r
i
n
m
a
k
i
n
g
d
ec
is
io
n
s
[
2
5
]
.
T
h
e
p
r
o
p
o
s
e
d
C
U
-
GW
O
m
o
d
e
l
is
d
es
cr
ib
ed
i
n
(
1
0
)
.
+
1
=
+
Ø
(
-
)
+
δ *
(
−
)
+
ψ
*
r
an
d
(
1
0
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
2
7
-
5
0
3
7
5032
CU
-
GW
O
is
a
b
etter
ap
p
r
o
ac
h
co
m
p
ar
e
d
to
ex
is
tin
g
m
eth
o
d
s
.
I
t
is
a
m
o
d
el
with
f
iv
e
s
tag
es.
T
wo
m
etah
eu
r
is
tic
o
p
tim
izatio
n
t
ec
h
n
iq
u
es
:
E
HO+
GW
O.
Featu
r
es
o
f
th
e
f
o
llo
win
g
ty
p
es
ar
e
em
p
lo
y
ed
:
p
r
ag
m
atic,
le
x
ical,
s
y
n
tactic,
s
em
an
tic,
an
d
co
n
tex
t
-
b
ased
.
Selecte
d
f
ea
tu
r
es
ar
e
u
s
ed
to
g
et
r
id
o
f
th
e
“
cu
r
s
e
o
f
d
im
en
s
io
n
ality
”.
I
t
m
an
a
g
e
s
d
ata
with
n
o
n
-
lin
ea
r
r
elatio
n
s
h
ip
s
an
d
h
ig
h
d
im
en
s
io
n
s
a
n
d
elim
in
ates
n
o
is
y
an
d
am
b
ig
u
o
u
s
d
ata
.
R
F,
SVM,
NN,
an
d
DC
NN
en
s
em
b
l
e
class
if
ier
s
ar
e
u
s
ed
.
DC
NN
is
u
s
ed
f
o
r
weig
h
t
f
in
e
-
tu
n
in
g
.
Op
tim
al
weig
h
ts
a
n
d
f
in
e
-
tu
n
in
g
ar
e
co
m
p
u
ted
.
E
x
p
lo
r
atio
n
an
d
e
x
p
lo
itatio
n
:
C
U
-
GW
O
im
p
r
o
v
es
th
e
eq
u
ilib
r
iu
m
b
etwe
en
ex
p
l
o
itatio
n
(
i
d
en
tify
in
g
th
e
b
est
s
o
lu
tio
n
with
th
e
least
lo
s
s
an
d
m
ax
im
izin
g
m
o
d
el
p
er
f
o
r
m
a
n
ce
)
an
d
ex
p
l
o
r
atio
n
(
id
en
tify
in
g
s
ev
e
r
al
s
o
lu
tio
n
s
)
.
W
o
r
d
2
v
ec
h
as
th
e
h
ig
h
est
s
en
s
itiv
ity
(
0
.
9
5
)
a
n
d
ze
r
o
p
ad
d
in
g
.
I
t
ac
ce
ler
ates
th
e
co
n
v
er
g
en
ce
r
ate
&
ca
n
i
n
v
esti
g
ate
v
ar
i
o
u
s
f
e
atu
r
es
an
d
o
p
tim
izatio
n
s
tr
ateg
ies
to
en
h
an
ce
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
.
T
h
e
ac
cu
r
ac
y
o
f
th
e
b
alan
ce
d
d
ataset
-
0
.
9
2
,
th
e
s
en
s
itiv
ity
was 0
.
9
5
,
an
d
t
h
e
F
-
m
ea
s
u
r
e
was 0
.
9
3
.
H
o
w
C
U
-
GW
O
wo
r
k
s
:
two
d
atasets
ar
e
in
clu
d
ed
in
th
is
m
o
d
el:
o
n
e
was
b
alan
ce
d
,
wh
il
e
th
e
o
th
er
was
u
n
b
alan
ce
d
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
id
e
n
tify
in
g
s
a
r
ca
s
m
in
th
e
tex
t
d
ataset
co
n
s
is
t
s
o
f
f
iv
e
s
tep
s
:
p
r
ep
r
o
ce
s
s
in
g
,
ap
p
r
o
p
r
iate
f
ea
tu
r
e
ex
tr
ac
tio
n
,
id
ea
l
f
ea
tu
r
e
s
elec
tio
n
,
an
d
d
ee
p
lear
n
in
g
-
b
a
s
ed
d
etec
tio
n
.
T
h
e
d
ataset
f
ir
s
t
u
n
d
e
r
g
o
es
a
p
r
e
p
r
o
ce
s
s
in
g
s
tep
t
h
at
in
clu
d
es
to
k
en
izatio
n
an
d
s
to
p
-
wo
r
d
r
em
o
v
al.
T
h
e
b
est
-
s
elec
ted
f
ea
tu
r
es
ar
e
f
ed
in
t
o
a
co
m
b
in
atio
n
ap
p
r
o
ac
h
th
at
c
o
n
s
is
ts
o
f
R
F,
SVM,
NN,
an
d
DC
NN
,
wh
ich
th
en
g
en
er
ates th
e
in
ten
d
ed
o
u
tco
m
e.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
P
y
t
h
o
n
w
a
s
u
s
e
d
t
o
s
i
m
u
la
t
e
th
e
m
e
t
h
o
d
[
2
6
]
.
M
U
S
t
AR
D
co
n
s
i
s
t
s
o
f
a
u
d
i
o
-
v
i
s
u
al
s
ta
t
e
m
en
t
s
l
a
b
e
le
d
w
i
t
h
s
a
r
c
as
m
i
d
e
n
t
i
f
i
c
a
ti
o
n
.
T
h
e
i
n
p
u
t
p
a
r
a
m
e
t
e
r
s
a
r
e
:
a
l
p
h
a
=
0
.
0
2
5
,
e
m
b
e
d
d
ed
_
s
e
n
t
e
n
c
e
s
i
z
e
-
2
6
,
m
o
d
e
l
=
wo
r
d
2
v
ec
.
W
o
r
d
2
V
e
c
(
v
e
c
t
o
r
_
s
i
z
e
=
1
0
0
,
v
o
c
a
b
u
l
a
r
y
=
2
9
6
5
)
.
B
a
l
a
n
c
e
d
d
a
t
as
e
t
d
im
e
n
s
i
o
n
is
(
6
9
0
,
3
)
.
I
m
b
a
l
a
n
c
e
d
d
a
t
a
s
e
t
-
f
i
le
1
s
a
r
ca
s
t
ic
d
i
m
e
n
s
i
o
n
(
2
0
0
,
3
)
,
f
i
l
e
2
non
-
s
a
r
c
a
s
ti
c
d
i
m
e
n
s
i
o
n
(
3
0
0
,
3
)
.
I
m
b
a
l
a
n
c
e
d
-
f
i
l
e
1
,
fi
l
e
2
=
3
0
0
+
2
0
0
=
5
0
0
=
(
5
0
0
,
1
2
6
)
.
T
r
a
i
n
i
n
g
s
e
t
(
3
5
0
,
1
2
6
)
,
t
e
s
t
i
n
g
s
e
t
(
2
0
7
,
1
2
6
)
.
[
6
0
%
,
4
0
%
]
.
T
a
b
l
e
s
1
a
n
d
2
d
e
s
c
r
i
b
e
t
h
e
p
a
r
a
m
e
t
e
r
s
&
g
i
v
e
t
h
e
s
t
at
is
t
i
ca
l
p
a
r
a
m
e
te
r
s
o
n
a
b
a
l
a
n
c
e
d
d
a
t
as
e
t
f
o
r
t
r
a
i
n
i
n
g
a
n
d
t
e
s
ti
n
g
.
T
ab
le
1
.
E
x
p
er
im
en
tal
p
ar
am
e
ter
s
M
e
t
r
i
c
s
V
a
l
u
e
s
F
e
a
t
u
r
e
s
H
e
a
d
l
i
n
e
s
(
t
e
x
t
)
R
e
c
o
r
d
s
2
6
,
6
9
1
S
a
r
c
a
st
i
c
r
e
c
o
r
d
s
1
3
,
5
6
8
W
i
t
h
o
u
t
s
a
r
c
a
sm
1
3
,
1
2
3
Ep
o
c
h
25
S
i
z
e
o
f
p
o
p
u
l
a
t
i
o
n
5
EH
O
n
_
c
l
a
n
s
=
5
,
a
l
p
h
a
=
0
.
5
,
b
e
t
a
=
0
.
5
W
O
A
b
=
1
a
n
d
p
=
0
.
5
T
ab
le
2
.
Statis
tical
p
ar
am
eter
s
o
n
a
b
alan
ce
d
d
ataset
f
o
r
tr
ain
in
g
an
d
test
in
g
B
a
l
a
n
c
e
d
d
a
t
a
s
e
t
W
o
r
d
2
v
e
c
(
7
0
%
,
3
0
%)
G
l
o
v
e
B
ER
T
6
9
0
t
r
a
i
n
_
d
a
t
a
_
m
(
4
8
3
,
1
2
6
)
t
r
a
i
n
_
d
a
t
a
_
m
(
4
8
3
,
9
9
)
t
r
a
i
n
_
d
a
t
a
_
m
(
4
8
3
,
7
1
)
t
e
st
_
d
a
t
a
_
m (
2
0
7
,
1
2
6
)
t
e
st
_
d
a
t
a
_
m (
2
0
7
,
9
9
)
t
e
st
_
d
a
t
a
_
m (
2
0
7
,
7
1
)
3
.
1
.
O
pera
t
io
na
l
pro
ce
du
re
T
h
e
p
u
r
p
o
s
e
is
to
u
s
e
em
b
ed
d
in
g
an
d
p
ad
d
i
n
g
tech
n
iq
u
es
in
th
e
p
r
o
p
o
s
ed
C
U
-
GW
O
alg
o
r
ith
m
with
a
DC
NN
to
in
v
esti
g
ate
th
e
r
ec
en
tly
in
tr
o
d
u
ce
d
s
ar
ca
s
m
d
etec
tio
n
m
o
d
el.
T
h
e
r
ec
en
tly
ad
d
ed
s
tep
s
ar
e
as
f
o
llo
ws:
i)
Pre
-
p
r
o
c
ess
i
n
g
,
i
n
v
o
l
v
es
f
i
n
is
h
in
g
p
a
d
d
in
g
a
n
d
e
m
b
e
d
d
in
g
.
W
o
r
d
2
v
ec
em
b
e
d
d
i
n
g
is
u
s
e
d
f
o
r
e
m
b
e
d
d
i
n
g
,
an
d
z
e
r
o
p
ad
d
i
n
g
is
u
t
iliz
e
d
f
o
r
p
a
d
d
i
n
g
.
Af
t
e
r
p
r
e
-
p
r
o
ce
s
s
in
g
,
"i
n
f
o
r
m
ati
o
n
g
ai
n
,
c
h
i
-
s
q
u
a
r
e
,
m
u
t
u
a
l
in
f
o
r
m
at
io
n
,
an
d
s
y
m
m
et
r
i
ca
l
u
n
ce
r
t
ai
n
t
y
-
b
ase
d
f
e
at
u
r
es"
e
x
t
r
a
ct
th
e
f
e
at
u
r
es
f
r
o
m
t
h
e
d
at
a.
ii)
Su
b
s
eq
u
en
tly
,
a
h
y
b
r
i
d
o
p
tim
i
za
tio
n
m
eth
o
d
k
n
o
w
n
as
clan
-
u
p
d
ated
g
r
ey
wo
lf
o
p
tim
izatio
n
(
C
U
-
GW
O)
is
u
s
ed
to
ch
o
o
s
e
th
e
b
est
f
ea
t
u
r
es.
T
h
e
p
r
o
p
o
s
ed
s
u
g
g
ested
en
s
em
b
le
tech
n
iq
u
e
in
clu
d
es
t
h
e
NN,
SVM,
R
F,
an
d
DC
NN
clas
s
if
ier
s
.
iii)
E
v
alu
ated
t
h
e
im
p
ac
ts
a
n
d
p
e
r
f
o
r
m
a
n
ce
o
f
v
ar
io
u
s
e
m
b
ed
d
in
g
tech
n
iq
u
es,
s
u
c
h
as
W
o
r
d
2
v
ec
,
Glo
Ve
,
an
d
B
E
R
T
in
th
e
b
alan
ce
d
an
d
u
n
b
alan
ce
d
d
ataset.
3
.
2
.
T
o
k
eniza
t
io
n
T
h
e
d
ataset
f
ir
s
t
u
n
d
er
g
o
es
a
p
r
ep
r
o
ce
s
s
in
g
s
tep
th
at
in
clu
d
es
to
k
en
izatio
n
an
d
s
to
p
-
wo
r
d
r
em
o
v
al.
T
h
e
r
aw
d
ata
t
h
at
was
g
ath
e
r
e
d
was
p
r
e
-
p
r
o
ce
s
s
ed
u
s
in
g
p
a
d
d
in
g
an
d
to
k
e
n
izatio
n
.
Key
w
o
r
d
s
ar
e
th
en
p
u
lled
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Meta
h
eu
r
is
tic
o
p
timiz
a
tio
n
fo
r
s
a
r
ca
s
m
d
etec
tio
n
in
s
o
cia
l m
ed
ia
w
ith
…
(
Gee
ta
S
a
h
u
)
5033
f
r
o
m
ea
ch
d
o
m
ain
.
On
ce
t
o
k
e
n
izatio
n
is
ac
h
iev
ed
,
e
m
b
ed
d
i
n
g
s
u
ch
as Wo
r
d
2
v
ec
,
k
n
o
wn
as wo
r
k
to
v
ec
to
r
,
is
u
s
ed
,
f
o
llo
wed
b
y
ze
r
o
an
d
en
d
p
ad
d
i
n
g
,
a
n
d
p
ad
d
in
g
tech
n
i
q
u
es u
tili
ze
d
.
3
.
3
.
F
e
a
t
ure
ex
t
r
a
ct
io
n
I
n
s
u
p
er
v
is
ed
lear
n
in
g
,
f
ea
tu
r
e
ex
tr
ac
tio
n
co
m
es
af
ter
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
p
h
ase.
A
s
et
o
f
f
ea
tu
r
es
is
cr
ea
ted
f
r
o
m
t
h
e
d
ata.
Am
o
n
g
th
e
tr
aits
ar
e
p
s
y
ch
o
lo
g
ical,
l
in
g
u
is
tic,
em
o
tio
n
al,
p
r
ag
m
ati
c,
an
d
ex
ag
g
er
ated
tr
aits
.
Af
ter
p
r
e
-
p
r
o
ce
s
s
in
g
,
t
h
e
to
k
en
ized
w
o
r
d
s
u
n
d
er
g
o
f
ea
tu
r
e
ex
tr
ac
tio
n
u
s
in
g
th
e
C
h
i
-
s
q
u
ar
e,
m
u
tu
al
in
f
o
r
m
atio
n
,
in
f
o
r
m
atio
n
g
ain
,
an
d
s
y
m
m
etr
ical
u
n
ce
r
tain
ty
tech
n
iq
u
es.
3
.
4
.
E
v
a
lua
t
i
o
n o
f
perf
o
r
m
a
nce
utilizing
Wo
rd2
v
ec
em
bedd
ing
a
nd
ze
ro
,
end pa
dd
in
g
wit
h CU
-
G
WO
T
ec
h
n
iq
u
es
f
o
r
p
ad
d
in
g
an
d
em
b
ed
d
in
g
ar
e
ess
en
tial
-
e
m
b
ed
d
in
g
is
th
e
p
r
o
ce
s
s
o
f
r
e
p
r
esen
tin
g
wo
r
d
s
o
r
to
k
e
n
s
in
v
ec
to
r
s
p
ac
e
as
n
u
m
b
er
s
.
T
o
tr
an
s
f
o
r
m
tex
t
in
p
u
t
in
to
a
n
u
m
er
ical
f
o
r
m
at
s
u
itab
le
f
o
r
m
ac
h
in
e
lear
n
in
g
,
we
u
s
e
W
o
r
d
2
v
ec
,
Glo
v
e,
a
n
d
B
E
R
T
.
B
y
allo
win
g
p
r
e
-
tr
ain
e
d
em
b
ed
d
in
g
to
d
if
f
er
e
n
t
task
s
,
tr
an
s
f
er
lear
n
in
g
is
u
s
ed
to
im
p
r
o
v
e
g
en
er
aliza
tio
n
a
n
d
s
av
e
co
m
p
u
tatio
n
.
Den
s
e
v
ec
to
r
s
r
ed
u
ce
th
e
h
ig
h
-
d
im
e
n
s
io
n
al
ch
ar
ac
ter
te
x
t
d
ata,
o
p
tim
izin
g
,
in
c
r
ea
s
in
g
th
e
ef
f
icien
cy
o
f
c
o
m
p
lex
o
p
er
atio
n
s
.
Pad
d
in
g
t
ec
h
n
iq
u
es
h
elp
in
tr
u
n
ca
tio
n
lo
s
s
,
ef
f
icien
t
b
atch
p
r
o
ce
s
s
in
g
&
m
em
o
r
y
m
an
a
g
em
e
n
t.
T
h
e
ac
cu
r
ac
ies
ac
h
iev
ed
i
n
W
o
r
d
2
v
ec
,
Glo
v
e,
an
d
B
E
R
T
ar
e
9
2
%,
9
0
%
,
an
d
8
9
%,
r
esp
ec
tiv
ely
.
T
h
e
p
o
s
itiv
e
Me
tr
ic
-
F_
m
ea
s
u
r
e
ev
alu
ate
d
o
v
er
W
o
r
d
2
v
ec
,
Glo
v
e,
a
n
d
B
E
R
T
is
9
3
%,
9
1
%,
a
n
d
9
0
%,
r
esp
ec
tiv
ely
.
T
h
e
n
e
g
ativ
e
m
etr
ic
-
FP
R
v
alu
es
f
o
r
W
o
r
d
2
v
ec
,
Glo
v
e,
an
d
B
E
R
T
ar
e
0
.
1
1
,
0
.
0
9
5
,
an
d
0
.
1
4
f
o
r
ze
r
o
p
a
d
d
in
g
.
Fig
u
r
e
4
th
e
p
er
f
o
r
m
an
ce
,
an
d
Fig
u
r
e
5
p
r
esen
ts
ch
ar
ac
ter
s
am
p
les,
with
Fig
u
r
e
5
(
a
)
s
h
o
win
g
th
e
ch
ar
ac
ter
s
p
r
i
o
r
t
o
p
r
ep
r
o
ce
s
s
in
g
,
an
d
Fig
u
r
es 5
(
b
)
to
5
(
k
)
s
h
o
win
g
th
e
c
h
ar
ac
t
er
s
f
o
llo
win
g
p
r
ep
r
o
ce
s
s
in
g
.
3
.
5
.
E
v
a
lua
t
i
o
n o
f
perf
o
r
m
a
nce
o
n t
he
o
ptim
a
l f
e
a
t
ures
a
nd
weig
ht
s
elec
t
io
n
T
h
e
ef
f
ec
tiv
en
ess
o
f
C
U
-
GW
O
(
E
HO+
GW
O)
wo
r
k
s
with
v
ar
io
u
s
tr
ain
in
g
d
ata
,
u
tili
zin
g
im
b
alan
ce
d
an
d
b
alan
ce
d
d
atasets
f
o
r
s
ar
ca
s
m
d
etec
tio
n
b
ased
o
n
th
e
b
est
f
ea
tu
r
es,
a
n
d
weig
h
t
s
ele
ctio
n
is
p
e
r
f
o
r
m
ed
[
2
7
]
.
T
h
e
m
o
d
el
p
er
f
o
r
m
s
b
etter
in
s
ar
ca
s
m
d
etec
tio
n
,
ac
h
ie
v
ed
a
n
ac
c
u
r
ac
y
r
ate
o
f
9
1
%,
9
1
.
2
%,
an
d
9
2
.
7
5
%,
r
esp
ec
tiv
ely
.
T
ab
le
3
d
ep
icts
t
h
e
C
U
-
GW
O
co
m
p
ar
is
o
n
with
an
d
with
o
u
t
f
ea
tu
r
e
a
n
d
weig
h
t o
p
tim
izatio
n
s
.
T
ab
le
3
.
C
U
-
GW
O
co
m
p
ar
is
o
n
with
an
d
with
o
u
t f
ea
tu
r
e
an
d
weig
h
t o
p
tim
izatio
n
s
M
e
a
su
r
e
s
CU
-
G
W
O
w
i
t
h
o
u
t
f
e
a
t
u
r
e
s
s
e
l
e
c
t
i
o
n
CU
-
G
W
O
w
i
t
h
o
u
t
w
e
i
g
h
t
o
p
t
i
m
i
z
a
t
i
o
n
CU
-
G
W
O
w
i
t
h
o
u
t
f
e
a
t
u
r
e
sel
e
c
t
i
o
n
o
r
w
e
i
g
h
t
o
p
t
i
m
i
z
a
t
i
o
n
CU
-
G
W
O
w
i
t
h
w
e
i
g
h
t
o
p
t
i
m
i
z
a
t
i
o
n
a
n
d
f
e
a
t
u
r
e
sel
e
c
t
i
o
n
A
c
c
u
r
a
c
y
0
.
8
5
3
0
.
8
7
9
0
.
8
2
6
0
8
7
0
.
9
2
7
5
3
6
M
a
t
t
h
e
w
’
s
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
f
i
c
i
e
n
t
(
M
C
C
)
0
.
8
0
4
0
.
8
2
8
0
.
6
4
9
3
7
5
0
.
8
5
0
9
6
5
F
a
l
se
d
i
sc
o
v
e
r
y
r
a
t
e
(
F
D
R
)
0
.
1
4
0
0
.
1
1
4
0
.
1
5
1
7
8
6
0
.
0
7
3
1
7
1
S
e
n
s
i
t
i
v
i
t
y
0
.
8
5
4
0
.
8
8
0
0
.
8
3
3
3
3
3
0
.
9
5
0
F
a
l
se
p
o
si
t
i
v
e
r
a
t
e
(
FPR
)
0
.
1
4
8
0
.
1
2
2
0
.
1
8
2
7
9
6
0
.
1
0
3
4
4
8
N
e
g
a
t
i
v
e
p
r
e
d
i
c
t
i
v
e
v
a
l
u
e
(
N
P
V
)
0
.
8
4
6
0
.
8
7
1
0
.
8
0
0
0
.
9
2
8
5
7
1
S
p
e
c
i
f
i
c
i
t
y
0
.
8
5
2
0
.
8
7
8
0
.
8
1
7
2
0
4
0
.
8
9
6
5
5
2
Th
e
R
a
n
d
i
n
d
e
x
0
.
8
7
8
0
.
9
0
4
0
.
9
0
8
8
9
3
0
.
9
6
3
0
8
7
P
r
e
c
i
s
i
o
n
0
.
8
6
0
0
.
8
8
6
0
.
8
4
8
2
1
4
0
.
9
2
6
8
2
9
F
a
l
se
n
e
g
a
t
i
v
e
r
a
t
e
(
F
N
R
)
0
.
1
4
6
0
.
1
2
0
0
.
1
6
6
6
6
7
0
.
0
5
0
F
-
mea
su
r
e
0
.
8
5
7
0
.
8
8
3
0
.
8
4
0
7
0
8
0
.
9
3
8
2
7
2
Fig
u
r
e
4
.
T
h
e
g
r
a
p
h
in
d
icate
s
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
test
ed
an
d
p
er
f
o
r
m
a
n
ce
is
ac
h
iev
ed
in
ter
m
s
o
f
ac
cu
r
ac
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
2
7
-
5
0
3
7
5034
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
(
g
)
(
h
)
(
i)
(
j)
(
k
)
Fig
u
r
e
5
.
C
h
ar
ac
ter
s
am
p
les we
r
e
tak
en
f
r
o
m
(
a)
ac
cu
r
ac
y
,
(
b
)
p
r
ec
is
io
n
,
(
c)
s
en
s
itiv
ity
,
(
d
)
s
p
ec
if
icity
,
(
e)
f_
m
ea
s
u
r
e,
(
f
)
r
an
d
_
in
d
ex
,
(
g
)
MCC
,
(
h
)
NPV,
(
i)
FP
R
,
(
j)
FNR
,
an
d
(
k
)
FDR
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Meta
h
eu
r
is
tic
o
p
timiz
a
tio
n
fo
r
s
a
r
ca
s
m
d
etec
tio
n
in
s
o
cia
l m
ed
ia
w
ith
…
(
Gee
ta
S
a
h
u
)
5035
3
.
6
.
Co
nfusi
o
n
m
a
t
rix
A
tab
le
u
s
ed
to
ass
ess
a
class
i
f
icatio
n
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
i
s
ca
lled
a
c
o
n
f
u
s
io
n
m
at
r
ix
.
T
o
d
escr
ib
e
th
e
ca
teg
o
r
izatio
n
f
in
d
i
n
g
s
,
it
co
n
tr
asts
th
e
ac
tu
al
clas
s
lab
els
with
th
e
p
r
ed
icted
class
l
ab
els.
A
co
n
f
u
s
io
n
m
atr
ix
is
u
s
ed
to
ass
e
s
s
h
o
w
well
a
clas
s
if
icatio
n
m
o
d
el
p
er
f
o
r
m
s
.
T
h
e
ac
tu
al
an
d
a
n
ticip
ated
v
alu
es
in
th
e
m
atr
ix
ar
e
co
m
p
ar
ed
b
y
t
h
e
m
o
d
el.
T
ab
le
4
s
h
o
ws th
e
ev
alu
atio
n
o
f
c
o
n
f
u
s
io
n
m
atr
ix
.
T
ab
le
4
.
C
o
n
f
u
s
io
n
m
atr
ix
v
al
u
es
M
o
d
e
l
TN
FP
FN
TP
W
o
r
d
2
V
e
c
–
Z
e
r
o
p
a
d
d
i
n
g
87
7
10
1
0
3
G
l
o
V
e
–
Ze
r
o
p
a
d
d
i
n
g
81
13
14
99
B
ER
T
–
Z
e
r
o
p
a
d
d
i
n
g
87
6
12
1
0
2
W
o
r
d
2
V
e
c
–
E
n
d
p
a
d
d
i
n
g
(
s
e
t
1
)
82
11
5
1
0
9
W
o
r
d
2
V
e
c
–
E
n
d
p
a
d
d
i
n
g
(
s
e
t
2
)
82
11
9
1
0
5
W
o
r
d
2
V
e
c
–
E
n
d
p
a
d
d
i
n
g
(
s
e
t
3
)
82
11
10
1
0
4
3
.
7
.
P
er
f
o
r
m
a
nce
ev
a
lua
t
io
n wit
h Wo
rd2
v
ec
em
bedd
ing
a
nd
ze
ro
,
end pa
dd
ing
A
b
a
l
an
c
ed
d
a
t
a
s
e
t
p
r
o
v
i
d
es
m
o
r
e
ac
c
u
r
a
t
e
r
e
s
u
l
t
s
o
f
0
.
9
2
2
1
co
m
p
a
r
ed
to
a
n
im
b
a
l
an
c
ed
d
a
t
a
s
e
t
0
.
8
9
8
5
.
F
D
R
i
n
i
m
b
al
a
n
c
e
d
a
t
a
s
e
t
i
s
0
.
1
2
9
0
3
a
n
d
b
a
l
a
n
c
e
d
d
a
t
a
s
e
t
i
s
0
.
0
8
8
0
0
.
T
a
b
l
e
5
s
h
o
w
s
t
h
e
e
v
a
l
u
a
t
i
o
n
.
B
en
ef
its
o
f
C
U
-
G
W
O
v
e
r
s
u
s
a
l
te
r
n
a
t
i
v
e
o
p
t
im
i
z
a
t
i
o
n
m
e
t
h
o
d
s
:
ex
p
lo
r
a
t
io
n
a
n
d
ex
p
lo
i
t
a
t
i
o
n
e
n
s
u
r
e
t
h
at
t
h
e
o
p
t
i
m
i
ze
r
s
e
x
p
lo
r
e
a
l
ar
g
er
s
o
lu
t
i
o
n
s
p
a
c
e
f
o
r
f
e
a
tu
r
e
w
e
i
g
h
t
s
an
d
m
o
d
e
l
p
a
r
am
e
t
er
a
d
j
u
s
t
m
en
t
,
p
r
e
v
en
t
i
n
g
p
r
em
at
u
r
e
c
o
n
v
er
g
en
c
e
.
M
u
l
t
im
o
d
al
f
e
a
tu
r
e
o
p
t
im
i
z
a
t
io
n
te
c
h
n
i
q
u
e
i
s
ex
c
e
l
l
en
t
f
o
r
c
o
n
d
u
c
t
in
g
ef
f
ec
t
i
v
e
s
e
ar
c
h
e
s
in
i
n
t
r
i
ca
t
e
m
u
l
t
im
o
d
a
l
f
ea
t
u
r
e
s
p
a
c
e
s
.
B
y
o
p
t
i
m
iz
i
n
g
b
o
th
f
e
a
tu
r
e
s
e
l
e
c
t
io
n
a
n
d
d
e
c
i
s
io
n
f
u
s
io
n
p
r
o
c
ed
u
r
e
s
,
i
t
o
f
f
e
r
s
im
p
r
o
v
e
d
f
i
t
n
es
s
e
v
a
l
u
a
t
io
n
.
A
l
g
o
r
i
th
m
s
s
u
c
h
a
s
d
if
f
er
e
n
t
i
a
l
e
v
o
lu
t
i
o
n
a
n
d
p
a
r
t
i
c
le
s
w
ar
m
o
p
t
i
m
iz
a
t
i
o
n
b
e
co
m
e
t
r
a
p
p
e
d
i
n
lo
c
a
l
o
p
t
i
m
a
a
n
d
a
r
e
u
n
a
b
le
to
m
a
i
n
ta
i
n
e
q
u
i
l
ib
r
iu
m
ac
r
o
s
s
m
o
d
a
l
i
t
i
e
s
.
T
ab
le
5
.
E
v
alu
atio
n
o
f
p
e
r
f
o
r
m
an
ce
u
tili
zin
g
W
o
r
d
2
v
ec
em
b
ed
d
in
g
an
d
ze
r
o
,
e
n
d
p
ad
d
in
g
M
e
t
h
o
d
A
c
c
u
r
a
c
y
S
e
n
s
i
t
i
v
i
t
y
S
p
e
c
i
f
i
c
i
t
y
P
r
e
c
i
s
i
o
n
F
N
R
F
mea
s
u
r
e
R
a
n
d
i
n
d
e
x
M
C
C
FPR
N
P
V
F
D
R
I
mb
a
l
a
n
c
e
d
a
t
a
0
.
8
9
8
0
.
9
5
5
0
.
8
2
9
0
.
8
7
0
0
.
0
4
4
0
.
9
1
1
0
.
9
4
7
0
.
7
9
8
0
.
1
7
0
0
.
9
3
9
0
.
1
2
9
B
a
l
a
n
c
e
d
d
a
t
a
0
.
9
2
2
0
.
9
5
1
0
.
8
8
7
0
.
9
1
1
0
.
0
4
8
0
.
9
3
0
0
.
9
6
0
0
.
8
4
4
0
.
1
1
2
0
.
9
3
8
0
.
0
8
8
4.
CO
NCLU
SI
O
N
A
f
i
v
e
-
p
h
a
s
e
m
o
d
e
l
f
o
r
s
a
r
ca
s
m
d
e
t
e
c
ti
o
n
i
s
p
r
o
p
o
s
ed
.
P
r
e
-
p
r
o
c
e
s
s
i
n
g
i
s
to
th
e
t
ex
t
in
p
u
t
,
w
h
i
ch
i
n
c
l
u
d
e
s
to
k
en
i
z
a
t
io
n
,
em
b
ed
d
i
n
g
,
an
d
p
ad
d
in
g
.
E
m
b
e
d
d
i
n
g
in
c
l
u
d
e
s
W
o
r
d
2
v
ec
,
Gl
o
v
e
,
an
d
B
E
R
T
,
f
o
l
l
o
we
d
b
y
p
a
d
d
in
g
.
C
U
-
GW
O
w
a
s
u
t
i
l
i
z
ed
.
N
N,
S
V
M
,
R
F
,
a
n
d
D
C
N
N
w
e
r
e
c
l
a
s
s
i
f
ie
r
s
u
s
ed
in
en
s
e
m
b
l
e
a
p
p
r
o
a
ch
e
s
.
T
h
e
D
C
N
N
r
e
s
u
l
t
s
i
n
d
i
c
a
t
ed
an
a
c
cu
r
a
c
y
o
f
0
.
9
2
2
.
A
b
a
l
a
n
ce
d
d
a
t
a
s
e
t
y
ie
l
d
s
m
o
r
e
ac
c
u
r
a
t
e
r
e
s
u
l
t
s
th
a
n
u
n
b
a
l
a
n
ce
d
.
L
im
i
t
a
t
i
o
n
s
:
t
h
e
a
n
n
o
ta
t
i
o
n
p
r
o
c
ed
u
r
e
w
a
s
m
o
r
e
d
if
f
i
cu
l
t
b
e
c
au
s
e
an
n
o
t
a
t
o
r
s
h
a
d
t
o
r
e
l
y
o
n
ju
s
t
th
e
p
u
b
l
i
c
d
a
t
a
s
e
t
.
M
i
s
u
n
d
e
r
s
t
an
d
in
g
s
co
u
l
d
r
e
s
u
l
t
f
r
o
m
m
a
ch
i
n
e
l
ea
r
n
in
g
a
lg
o
r
i
t
h
m
s
'
i
n
a
b
i
l
i
ty
to
g
r
a
s
p
c
o
n
t
e
x
t
's
s
u
b
t
l
e
t
i
e
s
.
A
m
o
d
e
l
t
r
a
i
n
ed
i
n
o
n
e
cu
l
t
u
r
a
l
s
e
t
t
in
g
m
i
g
h
t
n
o
t
p
e
r
f
o
r
m
w
e
l
l
i
n
a
n
o
t
h
e
r
.
T
h
e
e
f
f
ic
ie
n
c
y
o
f
th
e
m
o
d
e
l
c
a
n
b
e
i
m
p
r
o
v
e
d
b
y
i
n
c
lu
d
in
g
v
is
u
a
l
co
m
m
o
n
s
en
s
e
i
n
f
o
r
m
a
t
io
n
f
r
o
m
i
m
a
g
e
c
a
p
t
i
o
n
s
a
n
d
b
a
c
k
d
r
o
p
d
e
s
c
r
ip
t
io
n
s
t
h
a
t
i
s
n
o
t
in
c
lu
d
ed
.
Du
e
t
o
t
h
e
w
i
d
e
an
d
d
i
v
er
s
e
n
a
tu
r
e
o
f
t
h
e
in
p
u
t
,
t
h
e
c
u
r
r
e
n
t
m
o
d
e
l
h
a
s
n
o
t
t
a
k
en
i
n
t
o
a
c
co
u
n
t
a
n
s
w
er
s
,
r
ev
i
e
w
r
a
n
k
in
g
s
y
s
t
e
m
s
,
o
r
d
i
a
l
o
g
.
Fu
t
u
r
e
r
e
s
e
ar
ch
c
an
l
o
o
k
in
t
o
d
ev
e
l
o
p
i
n
g
s
a
r
ca
s
m
r
e
c
o
g
n
i
ti
o
n
m
o
d
e
l
s
th
a
t
a
r
e
r
eg
i
o
n
-
s
p
e
c
i
f
ic
a
n
d
d
o
m
a
i
n
-
s
p
e
c
if
i
c
(
p
o
l
i
t
ic
s
o
r
en
t
e
r
ta
i
n
m
e
n
t)
.
I
t
c
a
n
b
e
e
x
p
an
d
e
d
to
m
u
l
t
i
m
ed
i
a
co
m
p
u
t
in
g
ap
p
l
i
ca
t
i
o
n
s
a
s
r
e
a
l
-
t
im
e
v
o
ic
e
in
t
e
r
p
r
e
ta
t
i
o
n
,
d
e
e
p
f
a
k
e
v
id
eo
d
e
t
e
ct
i
o
n
,
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
io
n
,
an
d
th
e
in
t
er
n
e
t
o
f
m
u
l
t
i
m
ed
i
a
th
i
n
g
s
.
U
s
e
s
o
f
la
r
g
e
l
an
g
u
ag
e
m
o
d
e
l
s
(
L
L
Ms
)
f
o
r
s
ar
c
a
s
m
d
e
t
e
c
t
io
n
w
i
l
l
b
e
i
n
v
e
s
t
i
g
a
t
ed
i
n
f
u
t
u
r
e
s
t
u
d
i
e
s
.
I
m
ag
e
s
,
au
d
io
,
v
i
d
eo
,
e
m
o
t
i
co
n
s
,
an
d
m
em
e
s
c
an
a
l
l
b
e
u
s
e
d
t
o
i
d
e
n
ti
f
y
s
a
r
c
a
s
m
.
F
u
t
u
r
e
s
t
u
d
i
e
s
c
an
a
l
s
o
l
o
o
k
i
n
to
t
h
e
f
i
n
an
c
i
a
l
m
a
r
k
e
t
s
,
d
ep
r
e
s
s
i
o
n
,
d
e
f
am
a
t
i
o
n
,
g
o
v
er
n
m
e
n
t
i
s
s
u
e
s
,
c
y
b
e
r
s
t
a
lk
i
n
g
,
a
n
d
c
y
b
e
r
b
u
l
ly
i
n
g
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
5
0
2
7
-
5
0
3
7
5036
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
Gee
ta
Sah
u
✓
✓
✓
✓
✓
✓
✓
✓
Ma
n
o
j H
u
d
n
u
r
k
a
r
✓
✓
✓
✓
✓
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
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
a
ta
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
t
h
is
s
tu
d
y
a
r
e
av
ailab
le
i
n
G
itHu
b
at
h
ttp
s
://g
ith
u
b
.
co
m
/s
o
u
jan
y
ap
o
r
ia/MU
StA
R
D/b
lo
b
/m
aster
/d
ata/sar
ca
s
m
_
d
ata.
js
o
n
.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
K
a
l
a
i
v
a
n
i
a
n
d
D
.
T
h
e
n
mo
z
h
i
,
“
S
a
r
c
a
sm
i
d
e
n
t
i
f
i
c
a
t
i
o
n
a
n
d
d
e
t
e
c
t
i
o
n
i
n
c
o
n
v
e
r
s
i
o
n
c
o
n
t
e
x
t
u
s
i
n
g
B
ER
T,
”
i
n
P
ro
c
e
e
d
i
n
g
s o
f
t
h
e
S
e
c
o
n
d
W
o
rks
h
o
p
o
n
Fi
g
u
ra
t
i
v
e
L
a
n
g
u
a
g
e
Pr
o
c
e
ssi
n
g
,
2
0
2
0
,
p
p
.
72
–
76
,
doi
:
1
0
.
1
8
6
5
3
/
v
1
/
P
1
7
.
[
2
]
Y
.
D
u
,
T.
Li
,
M
.
S
.
P
a
t
h
a
n
,
H
.
K
.
Te
k
l
e
h
a
i
m
a
n
o
t
,
a
n
d
Z
.
Y
a
n
g
,
“
A
n
e
f
f
e
c
t
i
v
e
sarc
a
sm
d
e
t
e
c
t
i
o
n
a
p
p
r
o
a
c
h
b
a
se
d
o
n
s
e
n
t
i
me
n
t
a
l
c
o
n
t
e
x
t
a
n
d
i
n
d
i
v
i
d
u
a
l
e
x
p
r
e
ss
i
o
n
h
a
b
i
t
s,
”
C
o
g
n
i
t
i
v
e
C
o
m
p
u
t
a
t
i
o
n
,
v
o
l
.
1
4
,
p
p
.
7
8
–
9
0
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
2
5
5
9
-
0
2
1
-
0
9
8
3
2
-
x.
[
3
]
R
.
J
u
s
t
o
,
T
.
C
o
r
c
o
r
a
n
,
S
.
M
.
Lu
k
i
n
,
M
.
W
a
l
k
e
r
,
a
n
d
M
.
I
.
T
o
r
r
e
s,
“
E
x
t
r
a
c
t
i
n
g
r
e
l
e
v
a
n
t
k
n
o
w
l
e
d
g
e
f
o
r
t
h
e
d
e
t
e
c
t
i
o
n
o
f
sar
c
a
sm
a
n
d
n
a
s
t
i
n
e
ss
i
n
t
h
e
s
o
c
i
a
l
w
e
b
,
”
K
n
o
w
l
e
d
g
e
-
B
a
se
d
S
y
s
t
e
m
s
,
v
o
l
.
6
9
,
p
p
.
1
2
4
–
1
3
3
,
2
0
1
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
k
n
o
sy
s
.
2
0
1
4
.
0
5
.
0
2
1
.
[
4
]
A.
-
M
.
I
d
d
r
i
su
,
S
.
M
e
n
sa
h
,
F
.
B
o
a
f
o
,
G
.
R
.
Y
e
l
u
r
i
p
a
t
i
,
a
n
d
P
.
K
u
d
j
o
,
“
A
sen
t
i
m
e
n
t
a
n
a
l
y
s
i
s
f
r
a
m
e
w
o
r
k
t
o
c
l
a
s
si
f
y
i
n
s
t
a
n
c
e
s
o
f
sarca
s
t
i
c
s
e
n
t
i
m
e
n
t
s
w
i
t
h
i
n
t
h
e
a
v
i
a
t
i
o
n
se
c
t
o
r
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
Ma
n
a
g
e
m
e
n
t
D
a
t
a
I
n
si
g
h
t
s
,
v
o
l
.
3
,
n
o
.
2
,
N
o
v
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
j
i
me
i
.
2
0
2
3
.
1
0
0
1
8
0
.
[
5
]
A
.
K
u
mar,
S
.
D
i
k
sh
i
t
,
a
n
d
V
.
H
.
C
.
A
l
b
u
q
u
e
r
q
u
e
,
“
Ex
p
l
a
i
n
a
b
l
e
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
f
o
r
sar
c
a
sm
d
e
t
e
c
t
i
o
n
i
n
d
i
a
l
o
g
u
e
s
,
”
W
i
re
l
e
ss
C
o
m
m
u
n
i
c
a
t
i
o
n
s
a
n
d
M
o
b
i
l
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
0
2
1
,
n
o
.
1
,
J
u
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
1
/
2
9
3
9
3
3
4
.
[
6
]
R
.
K
u
m
a
r
a
n
d
J.
K
a
u
r
,
“
R
a
n
d
o
m
f
o
r
e
s
t
-
b
a
s
e
d
sa
r
c
a
s
t
i
c
t
w
e
e
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
u
s
i
n
g
m
u
l
t
i
p
l
e
f
e
a
t
u
r
e
c
o
l
l
e
c
t
i
o
n
,
”
i
n
M
u
l
t
i
m
e
d
i
a
Bi
g
D
a
t
a
C
o
m
p
u
t
i
n
g
f
o
r
I
o
T
A
p
p
l
i
c
a
t
i
o
n
s
,
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
v
o
l
.
1
6
3
,
S
i
n
g
a
p
o
r
e
:
S
p
r
i
n
g
e
r
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
981
-
13
-
8759
-
3_5.
[
7
]
V
.
G
o
v
i
n
d
a
n
a
n
d
V
.
B
a
l
a
k
r
i
s
h
n
a
n
,
“
A
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
i
n
a
n
a
l
y
si
n
g
t
h
e
e
f
f
e
c
t
o
f
h
y
p
e
r
b
o
l
e
s
u
s
i
n
g
n
e
g
a
t
i
v
e
se
n
t
i
m
e
n
t
t
w
e
e
t
s
f
o
r
s
a
r
c
a
sm
d
e
t
e
c
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
K
i
n
g
S
a
u
d
U
n
i
v
e
rsi
t
y
-
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
3
4
,
n
o
.
8
,
p
p
.
5
1
1
0
–
5
1
2
0
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
k
su
c
i
.
2
0
2
2
.
0
1
.
0
0
8
.
[
8
]
K
.
N
i
mal
a
,
R
.
Je
b
a
k
u
mar,
a
n
d
M
.
S
a
r
a
v
a
n
a
n
,
“
S
e
n
t
i
m
e
n
t
t
o
p
i
c
s
a
r
c
a
s
m
mi
x
t
u
r
e
mo
d
e
l
t
o
d
i
s
t
i
n
g
u
i
sh
s
a
r
c
a
s
m
p
r
e
v
a
l
e
n
t
t
o
p
i
c
s
b
a
s
e
d
o
n
t
h
e
s
e
n
t
i
me
n
t
b
e
a
r
i
n
g
w
o
r
d
s
i
n
t
h
e
t
w
e
e
t
s
,
”
J
o
u
rn
a
l
o
f
Am
b
i
e
n
t
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
H
u
m
a
n
i
ze
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
2
,
p
p
.
6
8
0
1
–
6
8
1
0
,
Ju
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
2
6
5
2
-
0
2
0
-
0
2
3
1
5
-
1.
[
9
]
Y
.
Y
.
Ta
n
,
C
.
C
h
o
w
,
J.
K
a
n
e
s
a
n
,
J.
H
.
C
h
u
a
h
,
a
n
d
Y
.
L
i
m,
“
S
e
n
t
i
m
e
n
t
a
n
a
l
y
si
s
a
n
d
s
a
r
c
a
sm
d
e
t
e
c
t
i
o
n
u
s
i
n
g
d
e
e
p
mu
l
t
i
t
a
s
k
l
e
a
r
n
i
n
g
,
”
W
i
re
l
e
ss
Pe
rs
o
n
a
l
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
1
2
9
,
p
p
.
2
2
1
3
–
2
2
3
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
2
7
7
-
0
2
3
-
1
0
2
3
5
-
4.
[
1
0
]
C
.
I
.
Ek
e
,
A
.
A
.
N
o
r
ma
n
,
a
n
d
L.
S
h
u
i
b
,
“
C
o
n
t
e
x
t
-
b
a
se
d
f
e
a
t
u
r
e
t
e
c
h
n
i
q
u
e
f
o
r
sarca
s
m
i
d
e
n
t
i
f
i
c
a
t
i
o
n
i
n
b
e
n
c
h
mar
k
d
a
t
a
set
s
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
B
E
R
T
m
o
d
e
l
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
4
8
5
0
1
–
4
8
5
1
8
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
6
8
3
2
3
.
[
1
1
]
R
.
A
.
B
a
g
a
t
e
a
n
d
R
.
S
u
g
u
n
a
,
“
S
a
r
c
a
sm
d
e
t
e
c
t
i
o
n
o
f
t
w
e
e
t
s
w
i
t
h
o
u
t
#
sar
c
a
sm:
d
a
t
a
sci
e
n
c
e
a
p
p
r
o
a
c
h
,
”
I
n
d
o
n
e
si
a
n
J
o
u
rn
a
l
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
3
,
n
o
.
2
,
p
p
.
9
9
3
–
1
0
0
1
,
A
u
g
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s.
v
2
3
.
i
2
.
p
p
9
9
3
-
1
0
0
1
.
[
1
2
]
A
.
A
g
a
r
w
a
l
,
A
.
K
.
Jh
a
,
A
.
J
a
i
sw
a
l
,
a
n
d
V
.
K
u
mar,
“
I
r
o
n
y
d
e
t
e
c
t
i
o
n
u
s
i
n
g
t
r
a
n
sf
o
r
mers
,
”
i
n
2
0
2
0
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
a
n
d
D
a
t
a
S
c
i
e
n
c
e
(
C
D
S
)
,
S
t
a
n
f
o
r
d
,
C
A
,
U
S
A
:
I
EEE,
2
0
2
0
,
p
p
.
1
6
5
-
1
6
8
,
d
o
i
:
1
0
.
1
1
0
9
/
C
D
S
4
9
7
0
3
.
2
0
2
0
.
0
0
0
4
0
.
[
1
3
]
K
.
S
u
n
d
a
r
a
r
a
j
a
n
a
n
d
A
.
P
a
l
a
n
i
sam
y
,
“
M
u
l
t
i
-
r
u
l
e
b
a
s
e
d
e
n
se
mb
l
e
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
mo
d
e
l
f
o
r
sarc
a
sm
t
y
p
e
d
e
t
e
c
t
i
o
n
i
n
Tw
i
t
t
e
r
,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
N
e
u
r
o
sc
i
e
n
c
e
,
v
o
l
.
2
0
2
0
,
n
o
.
1
,
J
a
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
0
/
2
8
6
0
4
7
9
.
[
1
4
]
R
.
M
i
sr
a
a
n
d
P
.
A
r
o
r
a
,
“
S
a
r
c
a
s
m
d
e
t
e
c
t
i
o
n
u
si
n
g
n
e
w
s
h
e
a
d
l
i
n
e
s
d
a
t
a
se
t
,
”
AI
O
p
e
n
,
v
o
l
.
4
,
p
p
.
1
3
–
1
8
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
i
o
p
e
n
.
2
0
2
3
.
0
1
.
0
0
1
.
[
1
5
]
Y
.
Z
h
a
n
g
,
Y
.
Li
u
,
Q
.
Li
,
a
n
d
P
.
Ti
w
a
r
i
,
“
C
F
N
:
A
c
o
m
p
l
e
x
-
v
a
l
u
e
d
f
u
z
z
y
n
e
t
w
o
r
k
f
o
r
s
a
r
c
a
sm
d
e
t
e
c
t
i
o
n
i
n
c
o
n
v
e
r
sa
t
i
o
n
s,”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
F
u
zzy
S
y
st
e
m
s
,
v
o
l
.
2
9
,
n
o
.
1
2
,
p
p
.
3
6
9
6
–
3
7
1
0
,
A
p
r
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TFU
ZZ.
2
0
2
1
.
3
0
7
2
4
9
2
.
[
1
6
]
D
.
K
.
S
h
a
r
ma,
B
.
S
i
n
g
h
,
S
.
A
g
a
r
w
a
l
,
N
.
P
a
c
h
a
u
r
i
,
A
.
A
.
A
l
h
u
ssa
n
,
a
n
d
H
.
A
.
A
b
d
a
l
l
a
h
,
“
S
a
r
c
a
s
m
d
e
t
e
c
t
i
o
n
o
v
e
r
s
o
c
i
a
l
m
e
d
i
a
p
l
a
t
f
o
r
ms
u
si
n
g
h
y
b
r
i
d
e
n
se
mb
l
e
mo
d
e
l
w
i
t
h
f
u
z
z
y
l
o
g
i
c
,
”
E
l
e
c
t
r
o
n
i
c
s
,
v
o
l
.
1
2
,
n
o
.
4
,
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s1
2
0
4
0
9
3
7
.
[
1
7
]
I
.
A
.
F
a
r
h
a
a
n
d
W
.
M
a
g
d
y
,
“
F
r
o
m
A
r
a
b
i
c
se
n
t
i
m
e
n
t
a
n
a
l
y
s
i
s
t
o
s
a
r
c
a
sm
d
e
t
e
c
t
i
o
n
:
t
h
e
A
r
S
a
r
c
a
sm
d
a
t
a
s
e
t
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
4
t
h
Wo
r
k
sh
o
p
o
n
O
p
e
n
-
S
o
u
rc
e
Ar
a
b
i
c
C
o
rp
o
r
a
a
n
d
Pro
c
e
ss
i
n
g
T
o
o
l
s
,
M
a
r
se
i
l
l
e
,
F
r
a
n
c
e
:
E
u
r
o
p
e
a
n
La
n
g
u
a
g
e
R
e
so
u
r
c
e
A
sso
c
i
a
t
i
o
n
,
2
0
2
0
,
p
p
.
3
2
–
3
9
.
[
1
8
]
I
.
To
u
a
h
r
i
a
n
d
A
.
M
a
z
r
o
u
i
,
“
E
n
h
a
n
c
e
men
t
o
f
a
mu
l
t
i
-
d
i
a
l
e
c
t
a
l
se
n
t
i
me
n
t
a
n
a
l
y
s
i
s
sy
st
e
m
b
y
t
h
e
d
e
t
e
c
t
i
o
n
o
f
t
h
e
i
mp
l
i
e
d
s
a
r
c
a
s
t
i
c
f
e
a
t
u
r
e
s,
”
K
n
o
w
l
e
d
g
e
-
B
a
se
d
S
y
s
t
e
m
s
,
v
o
l
.
2
2
7
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
k
n
o
sy
s
.
2
0
2
1
.
1
0
7
2
3
2
.
[
1
9
]
A
.
O
n
a
n
a
n
d
M
.
A
.
To
c
o
g
l
u
,
“
A
t
e
r
m
w
e
i
g
h
t
e
d
n
e
u
r
a
l
l
a
n
g
u
a
g
e
m
o
d
e
l
a
n
d
st
a
c
k
e
d
b
i
d
i
r
e
c
t
i
o
n
a
l
LS
TM
b
a
s
e
d
f
r
a
m
e
w
o
r
k
f
o
r
sarca
sm
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
”
I
E
EE
A
c
c
e
ss
,
v
o
l
.
9
,
p
p
.
7
7
0
1
–
7
7
2
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
4
9
7
3
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.