I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
,
p
p
.
1474
~
1
4
8
4
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
1
6
i
3
.
pp
1
4
7
4
-
1
4
8
4
1474
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Integ
ra
ting BE
R
T fine
-
tuning a
nd
genetic
alg
o
rithm f
o
r
superio
r dep
ress
i
o
n det
ec
tion in so
cia
l media
Abd Alla
h Ao
ura
g
h
1
,
M
o
ha
m
ed
B
a
ha
j
1
,
F
o
ua
d T
o
ufik
2
1
M
I
ET
La
b
o
r
a
t
o
r
y
,
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
i
q
u
e
s,
H
a
ssa
n
1
st
U
n
i
v
e
r
si
t
y
,
S
e
t
t
a
t
,
M
o
r
o
c
c
o
2
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
s L
a
b
o
r
a
t
o
r
y
,
H
i
g
h
e
r
S
c
h
o
o
l
o
f
Te
c
h
n
o
l
o
g
y
,
M
o
h
a
mm
e
d
V
U
n
i
v
e
r
si
t
y
,
S
a
l
e
,
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
Au
g
2
9
,
2
0
2
4
R
ev
is
ed
Feb
2
1
,
2
0
2
6
Acc
ep
ted
Ma
r
1
6
,
2
0
2
6
Early
d
e
tec
ti
o
n
o
f
d
e
p
re
ss
io
n
is
c
ru
c
ial
fo
r
m
in
imiz
in
g
i
ts
a
d
v
e
rse
e
ffe
c
ts
o
n
m
e
n
tal
a
n
d
p
h
y
sic
a
l
h
e
a
lt
h
.
R
e
c
e
n
t
a
d
v
a
n
c
e
m
e
n
ts
in
n
a
tu
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
fa
c
il
it
a
te
th
e
larg
e
-
sc
a
le
a
n
a
ly
sis
o
f
so
c
ial
m
e
d
ia
tex
ts
to
id
e
n
t
if
y
d
e
p
re
ss
iv
e
ten
d
e
n
c
ies
.
Ou
r
st
u
d
y
in
tro
d
u
c
e
s
a
n
o
v
e
l
a
p
p
r
o
a
c
h
b
y
i
n
teg
ra
ti
n
g
a
g
e
n
e
ti
c
a
lg
o
rit
h
m
fo
r
h
y
p
e
r
p
a
ra
m
e
ter
tu
n
i
n
g
,
o
p
ti
m
izi
n
g
th
e
c
las
sifica
ti
o
n
p
e
rfo
rm
a
n
c
e
b
e
y
o
n
d
c
o
n
v
e
n
t
io
n
a
l
m
e
th
o
d
s.
We
p
r
o
v
i
d
e
a
c
o
m
p
re
h
e
n
siv
e
c
o
m
p
a
riso
n
o
f
v
e
c
to
riza
ti
o
n
tec
h
n
i
q
u
e
s,
in
c
l
u
d
i
n
g
term
fre
q
u
e
n
c
y
-
in
v
e
rse
d
o
c
u
m
e
n
t
fre
q
u
e
n
c
y
(T
F
-
IDF
)
,
Wo
rd
2
Ve
c
,
a
n
d
a
fi
n
e
-
tu
n
e
d
b
i
d
irec
ti
o
n
a
l
e
n
c
o
d
e
r
re
p
re
se
n
tatio
n
fro
m
t
ra
n
sfo
rm
e
rs
(
BERT
)
m
o
d
e
l
sp
e
c
ifi
c
a
ll
y
a
d
a
p
ted
t
o
o
u
r
d
a
tas
e
t.
Us
in
g
a
d
a
tas
e
t
o
f
7
,
7
3
1
e
n
tri
e
s,
we
imp
lem
e
n
ted
sta
n
d
a
rd
p
re
-
p
ro
c
e
ss
in
g
ste
p
s
su
c
h
a
s
sto
p
wo
rd
re
m
o
v
a
l
a
n
d
lem
m
a
ti
z
a
ti
o
n
b
e
fo
re
v
e
c
to
rizin
g
th
e
tex
t.
F
i
v
e
m
a
c
h
in
e
lea
rn
in
g
a
lg
o
r
it
h
m
s
—
d
e
c
isio
n
tree
,
lo
g
isti
c
re
g
re
ss
io
n
,
ra
n
d
o
m
f
o
re
st,
g
ra
d
ien
t
b
o
o
st
in
g
,
a
n
d
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
—
we
re
e
v
a
lu
a
ted
,
with
h
y
p
e
r
p
a
ra
m
e
ter
tu
n
in
g
p
e
rfo
rm
e
d
u
sin
g
a
g
e
n
e
ti
c
a
lg
o
ri
th
m
.
T
h
e
h
ig
h
e
st
a
c
c
u
ra
c
y
(9
5
.
9
9
%
)
a
n
d
F
1
-
sc
o
re
(9
5
.
9
1
%
)
we
re
a
c
h
iev
e
d
with
t
h
e
c
o
m
b
i
n
a
ti
o
n
o
f
fin
e
-
t
u
n
e
d
BERT
,
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
,
a
n
d
g
e
n
e
ti
c
a
lg
o
rit
h
m
o
p
ti
m
iza
ti
o
n
.
Th
is
st
u
d
y
d
e
m
o
n
s
trate
s
th
e
a
d
v
a
n
tag
e
s
o
f
i
n
teg
ra
ti
n
g
BER
T
fin
e
-
tu
n
in
g
with
g
e
n
e
ti
c
o
p
t
imiz
a
ti
o
n
,
o
u
t
p
e
rfo
rm
in
g
trad
i
ti
o
n
a
l
TF
-
ID
F
a
n
d
Wo
r
d
2
Ve
c
a
p
p
r
o
a
c
h
e
s
in
d
e
p
re
ss
io
n
d
e
tec
ti
o
n
.
K
ey
w
o
r
d
s
:
B
E
R
T
Dep
r
ess
io
n
d
etec
tio
n
Gen
etic
alg
o
r
ith
m
Ma
ch
in
e
lear
n
in
g
Natu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
Su
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
Vec
to
r
izatio
n
tech
n
iq
u
es
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
:
Ab
d
Allah
Ao
u
r
a
g
h
MI
E
T
L
ab
o
r
ato
r
y
,
Facu
lty
o
f
Scien
ce
s
an
d
T
ec
h
n
iq
u
es,
Hass
an
1
s
t U
n
iv
er
s
ity
Settat,
Mo
r
o
cc
o
E
m
ail: a
b
d
allah
ao
u
r
ag
h
@
g
m
a
il.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
Dep
r
ess
io
n
is
a
wid
esp
r
ea
d
m
en
tal
illn
ess
af
f
ec
tin
g
m
illi
o
n
s
o
f
p
e
o
p
le
wo
r
l
d
wid
e,
ch
a
r
ac
ter
ized
b
y
p
er
s
is
ten
t
f
ee
lin
g
s
o
f
s
ad
n
ess
,
lo
s
s
o
f
in
ter
est,
an
d
d
ec
r
ea
s
ed
en
er
g
y
[
1
]
.
Acc
o
r
d
in
g
to
th
e
W
o
r
ld
Hea
lt
h
Or
g
an
izatio
n
(
W
HO)
,
5
%
o
f
th
e
g
lo
b
al
p
o
p
u
latio
n
cu
r
r
e
n
tly
s
u
f
f
er
s
f
r
o
m
d
e
p
r
ess
io
n
,
wh
ich
,
in
its
m
o
s
t
s
ev
er
e
f
o
r
m
s
,
ca
n
lead
to
s
u
ic
id
e
[
2
]
.
Ar
o
u
n
d
7
0
0
,
0
0
0
p
eo
p
le
wo
r
ld
wid
e
t
o
o
k
th
eir
o
wn
l
iv
es,
a
n
u
m
b
er
th
at
is
s
tead
ily
in
cr
ea
s
in
g
,
u
n
d
er
s
co
r
in
g
th
e
g
r
av
ity
o
f
d
ep
r
ess
io
n
as
a
p
u
b
lic
h
ea
lth
is
s
u
e
[
1
]
,
[
2
]
.
T
h
e
co
n
s
eq
u
en
ce
s
o
f
th
is
d
is
ea
s
e
in
clu
d
e
s
ig
n
if
ica
n
t
h
a
r
m
to
q
u
a
lity
o
f
life
,
s
o
cial
r
elatio
n
s
h
ip
s
,
an
d
a
d
im
i
n
is
h
ed
ca
p
ac
ity
to
p
er
f
o
r
m
d
aily
a
ctiv
ities
[
2
]
,
[
3
]
.
T
r
ea
tm
e
n
t
o
p
tio
n
s
en
co
m
p
ass
p
h
ar
m
a
co
lo
g
ical
th
er
ap
ies,
p
s
y
ch
o
th
er
a
p
y
,
a
n
d
life
s
ty
le
in
ter
v
en
tio
n
s
[
3
]
.
Ho
wev
e
r
,
ch
allen
g
es
p
er
s
is
t,
s
u
ch
a
s
u
n
d
er
d
iag
n
o
s
is
,
ass
o
ciate
d
s
t
ig
m
a,
an
d
lim
ited
ac
ce
s
s
to
ca
r
e
in
s
o
m
e
r
eg
io
n
s
[
4
]
.
T
h
er
ef
o
r
e,
ea
r
ly
d
et
ec
tio
n
an
d
tim
ely
in
ter
v
en
tio
n
ar
e
cr
u
cial
t
o
m
i
tig
atin
g
th
e
ad
v
e
r
s
e
ef
f
ec
ts
o
f
d
ep
r
ess
io
n
an
d
r
e
d
u
cin
g
its
o
v
er
all
im
p
ac
t
o
n
p
u
b
lic
h
ea
lth
[
5
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
tin
g
B
E
R
T fin
e
-
tu
n
in
g
a
n
d
g
e
n
etic
a
lg
o
r
ith
m
fo
r
…
(
A
b
d
A
lla
h
A
o
u
r
a
g
h
)
1475
Ma
ch
in
e
lear
n
in
g
(
ML
)
tec
h
n
iq
u
es,
wh
e
n
in
teg
r
ated
with
ad
v
an
ce
m
e
n
ts
in
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P),
o
f
f
er
p
o
wer
f
u
l
t
o
o
ls
f
o
r
a
n
aly
zin
g
tex
tu
a
l
d
ata
a
n
d
p
r
e
d
ictin
g
m
ed
ical
co
n
d
itio
n
s
s
u
c
h
a
s
d
ep
r
ess
io
n
an
d
o
th
er
d
is
ea
s
es
[
6
]
,
[
7
]
.
B
y
lev
e
r
ag
in
g
NL
P,
we
ca
n
ex
am
in
e
a
n
d
in
te
r
p
r
et
th
e
s
u
b
tleties
o
f
lan
g
u
ag
e
i
n
s
o
cial
n
etwo
r
k
p
o
s
ts
,
f
ac
ilit
atin
g
th
e
ea
r
ly
d
etec
tio
n
o
f
d
ep
r
ess
iv
e
s
y
m
p
to
m
s
[
8
]
.
Un
lik
e
g
en
er
al
s
en
tim
en
t
an
aly
s
is
,
wh
ich
cl
ass
if
ies
tex
t
as
p
o
s
itiv
e,
n
eg
ativ
e,
o
r
n
eu
tr
al,
d
ep
r
ess
io
n
d
etec
tio
n
r
eq
u
ir
es
id
en
tify
in
g
lin
g
u
is
tic
p
atter
n
s
s
p
ec
if
ic
to
clin
ical
d
ep
r
ess
iv
e
s
tates,
s
u
ch
as
s
elf
-
r
ef
er
en
tial
ex
p
r
ess
io
n
s
an
d
co
g
n
itiv
e
d
is
to
r
tio
n
s
.
Sen
tim
e
n
t
an
aly
s
is
alo
n
e
m
ay
m
is
in
te
r
p
r
et
th
ese
cu
es,
m
a
k
in
g
it
n
e
ce
s
s
ar
y
to
d
e
v
elo
p
s
p
ec
ialized
m
o
d
els
tailo
r
ed
f
o
r
d
e
p
r
ess
io
n
d
etec
tio
n
[
6
]
,
[
8
]
.
Ad
v
an
ce
d
v
ec
to
r
izatio
n
te
ch
n
iq
u
es,
i
n
clu
d
in
g
ter
m
f
r
e
q
u
en
c
y
-
in
v
e
r
s
e
d
o
cu
m
en
t
f
r
e
q
u
en
c
y
(TF
-
I
DF
)
,
W
o
r
d
2
Vec
,
an
d
th
e
f
in
e
-
t
u
n
ed
b
id
ir
ec
tio
n
al
e
n
co
d
e
r
r
ep
r
esen
tatio
n
f
r
o
m
tr
an
s
f
o
r
m
er
s
(
B
E
R
T
)
m
o
d
el,
p
lay
a
cr
u
cial
r
o
le
in
ca
p
tu
r
in
g
th
e
co
n
tex
t
an
d
s
u
b
tle
em
o
tio
n
s
ex
p
r
ess
ed
in
tex
t
[
9
]
.
Mo
r
eo
v
er
,
h
y
p
er
p
a
r
am
ete
r
o
p
tim
izatio
n
th
r
o
u
g
h
g
e
n
et
ic
alg
o
r
ith
m
s
[
1
0
]
s
ig
n
if
ican
tly
en
h
an
ce
s
th
e
ac
c
u
r
ac
y
an
d
p
er
f
o
r
m
an
ce
o
f
p
r
e
d
ictiv
e
m
o
d
els.
An
o
th
er
k
ey
a
d
v
an
tag
e
o
f
NL
P
is
its
ca
p
ac
ity
to
co
n
tin
u
o
u
s
ly
m
o
n
ito
r
ch
an
g
es
in
an
in
d
i
v
id
u
al'
s
m
en
tal
s
tate
v
ia
r
ea
l
-
tim
e
tex
t
d
ata
a
n
aly
s
is
,
allo
win
g
f
o
r
r
a
p
id
r
esp
o
n
s
e
t
o
ev
o
lv
i
n
g
c
o
n
d
itio
n
s
an
d
tim
e
ly
s
u
p
p
o
r
t
[
1
1
]
.
B
y
in
te
g
r
atin
g
th
ese
tech
n
iq
u
es,
we
ca
n
n
o
t
o
n
ly
d
etec
t
d
ep
r
e
s
s
io
n
ea
r
lier
b
u
t
also
g
ain
d
e
ep
er
in
s
ig
h
ts
i
n
to
its
m
a
n
if
estatio
n
s
an
d
tr
ig
g
e
r
s
,
p
o
ten
tially
lead
in
g
to
m
o
r
e
ef
f
ec
tiv
e
p
r
ev
en
tio
n
an
d
tr
ea
tm
e
n
t stra
teg
ies.
I
n
th
is
c
o
n
tex
t,
Ar
ac
h
ch
i
g
e
et
a
l.
[
1
2
]
r
ev
iew
NL
P
a
n
d
ML
tech
n
iq
u
es
f
o
r
i
d
en
tify
in
g
d
ep
r
ess
io
n
in
o
n
lin
e
s
u
p
p
o
r
t
f
o
r
u
m
s
,
a
n
al
y
zin
g
2
9
ar
ticles
to
d
eter
m
in
e
th
e
m
o
s
t
ef
f
ec
tiv
e
a
n
d
s
ca
lab
le
f
ea
tu
r
e
co
m
b
in
atio
n
s
.
T
h
ei
r
s
tu
d
y
h
ig
h
lig
h
ts
ch
allen
g
es
s
u
ch
as
p
r
ac
tical
im
p
lem
en
tatio
n
a
n
d
eth
ical
is
s
u
es,
s
u
g
g
esti
n
g
f
u
t
u
r
e
r
esear
ch
to
r
ef
in
e
th
ese
a
p
p
r
o
ac
h
es
an
d
e
n
h
an
ce
t
h
eir
clin
ical
a
p
p
licatio
n
.
Glaz
et
a
l.
[
1
3
]
h
ig
h
lig
h
t
th
e
g
r
o
win
g
u
s
e
o
f
ML
an
d
NL
P
in
m
ed
ici
n
e,
n
o
tin
g
th
at
wh
ile
th
ese
m
o
d
e
ls
o
f
f
er
in
n
o
v
ativ
e
in
s
ig
h
ts
an
d
o
f
ten
v
alid
ate
e
x
is
tin
g
h
y
p
o
th
eses
,
th
e
y
ar
e
ty
p
ically
lim
ited
t
o
s
p
ec
if
ic
co
h
o
r
ts
,
lik
e
s
o
cial
m
ed
ia
u
s
er
s
,
wh
ich
m
ay
lim
it
th
eir
g
en
er
al
ap
p
licab
ilit
y
.
T
h
eir
s
u
r
v
ey
em
p
h
asizes
th
e
p
o
ten
tial
o
f
NL
P
to
ex
p
lo
r
e
o
t
h
er
wis
e
in
ac
ce
s
s
ib
l
e
d
ata
b
u
t
s
tr
ess
es
th
e
n
ee
d
f
o
r
ca
r
ef
u
l
eth
ical
ev
alu
atio
n
b
ef
o
r
e
in
te
g
r
atin
g
th
ese
tech
n
iq
u
es
in
to
m
en
tal
h
ea
lth
ca
r
e.
J
ain
et
a
l.
[
1
4
]
ap
p
lied
ML
an
d
NL
P
to
p
r
ed
ict
d
ep
r
ess
iv
e
p
o
s
ts
o
n
R
ed
d
it.
T
h
ey
an
aly
ze
d
co
m
m
e
n
ts
an
d
p
o
s
ts
r
elate
d
to
s
u
icid
al
id
ea
tio
n
u
s
in
g
alg
o
r
ith
m
s
s
u
ch
as
n
aiv
e
B
ay
es,
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVM)
,
lo
g
is
tic
r
eg
r
ess
io
n
,
an
d
r
an
d
o
m
f
o
r
est.
T
h
e
s
tu
d
y
a
ch
i
ev
ed
an
ac
cu
r
ac
y
o
f
7
7
.
1
2
%
a
n
d
a
n
F1
-
s
co
r
e
o
f
7
7
%
with
SVM,
h
ig
h
lig
h
tin
g
t
h
e
ef
f
ec
tiv
e
n
ess
o
f
th
ese
tech
n
iq
u
es
in
id
e
n
tify
in
g
at
-
r
is
k
in
d
iv
id
u
als
o
n
o
n
lin
e
p
latf
o
r
m
s
.
Saif
u
llah
et
a
l.
[
1
5
]
i
n
v
esti
g
ated
a
n
x
iety
d
ete
ctio
n
b
y
an
al
y
zin
g
ap
p
r
o
x
im
ately
4
,
8
6
2
co
m
m
e
n
ts
f
r
o
m
Yo
u
T
u
b
e.
T
h
ey
ap
p
li
ed
d
ata
m
in
i
n
g
a
n
d
m
ac
h
in
e
l
ea
r
n
in
g
al
g
o
r
ith
m
s
,
test
in
g
s
ix
class
if
ier
s
.
T
h
e
b
e
s
t
p
er
f
o
r
m
an
ce
was
ac
h
iev
ed
b
y
th
e
R
an
d
o
m
Fo
r
est
m
o
d
el,
wh
ich
attain
e
d
a
n
ac
cu
r
ac
y
o
f
8
4
.
9
9
%,
d
em
o
n
s
tr
atin
g
th
e
ef
f
ec
tiv
e
n
ess
o
f
m
a
ch
in
e
lear
n
in
g
tec
h
n
iq
u
es
in
id
en
tify
in
g
e
m
o
tio
n
s
f
r
o
m
o
n
lin
e
in
ter
ac
tio
n
s
.
Ko
u
r
et
a
l.
[
1
6
]
in
v
esti
g
ated
p
r
ed
ictin
g
u
s
er
s
'
m
en
tal
s
ta
tes
b
y
class
if
y
in
g
d
ep
r
ess
ed
an
d
n
o
n
-
d
ep
r
ess
ed
in
d
iv
id
u
als
f
r
o
m
T
witter
d
ata.
T
h
ey
u
s
ed
a
h
y
b
r
id
m
o
d
el
co
m
b
in
in
g
a
co
n
v
o
lu
ti
o
n
al
n
eu
r
a
l
n
etwo
r
k
(
C
NN)
an
d
a
b
id
i
r
ec
tio
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
n
etwo
r
k
(
B
iLST
M)
,
ac
h
iev
i
n
g
an
ac
c
u
r
ac
y
o
f
9
4
.
2
8
%
o
n
d
ep
r
ess
io
n
-
r
elate
d
twee
ts
.
T
h
is
m
o
d
el
o
u
tp
e
r
f
o
r
m
ed
o
th
e
r
a
p
p
r
o
ac
h
es
in
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
.
C
h
er
ed
d
y
et
a
l.
[
1
7
]
h
ig
h
lig
h
ted
th
e
p
o
ten
tial
o
f
NL
P
f
o
r
d
etec
tin
g
d
e
p
r
ess
io
n
th
r
o
u
g
h
twee
t
an
aly
s
is
o
n
s
o
cial
n
etwo
r
k
s
.
T
h
ey
n
o
ted
th
e
n
ee
d
f
o
r
m
o
r
e
r
o
b
u
s
t
d
at
a
to
e
n
h
an
ce
ac
cu
r
ac
y
an
d
r
ec
all.
T
h
eir
s
tu
d
y
,
em
p
lo
y
in
g
class
if
ier
s
lik
e
n
ai
v
e
B
ay
es,
SVM,
an
d
lo
g
is
tic
r
eg
r
ess
io
n
to
ca
teg
o
r
ize
twee
ts
in
to
p
o
s
itiv
e
an
d
n
eg
ativ
e
s
en
tim
en
ts
,
f
o
u
n
d
th
at
SVM
ac
h
iev
ed
th
e
h
ig
h
est
p
er
f
o
r
m
an
ce
with
8
5
%
ac
c
u
r
a
cy
,
8
9
%
p
r
ec
is
io
n
,
an
d
6
3
%
r
ec
all
an
d
F1
-
s
co
r
e
.
C
h
en
et
a
l.
[
1
8
]
u
tili
ze
d
N
L
P
to
d
etec
t
d
e
p
r
ess
io
n
in
u
n
s
tr
u
ctu
r
ed
m
ed
ical
r
ec
o
r
d
s
.
An
al
y
zin
g
2
2
,
3
5
5
M
an
d
ar
in
r
ec
o
r
d
s
,
th
ey
e
m
p
lo
y
ed
a
B
E
R
T
m
o
d
el
co
m
b
in
ed
with
C
NNs.
T
h
eir
s
tu
d
y
s
h
o
wed
s
tr
o
n
g
p
e
r
f
o
r
m
a
n
ce
with
g
en
er
al
(
AUC
9
3
%)
an
d
civ
ilian
(
AUC
9
1
%)
m
o
d
els,
b
u
t
th
e
m
ilit
ar
y
s
am
p
les
u
n
d
er
p
er
f
o
r
m
ed
(
AU
C
7
9
%),
with
t
h
e
m
ilit
ar
y
-
s
p
e
cif
ic
m
o
d
el
ac
h
iev
in
g
a
b
etter
AUC
o
f
8
2
%.
T
h
e
f
in
d
in
g
s
v
alid
ate
th
e
u
s
e
o
f
d
ee
p
lear
n
in
g
tec
h
n
iq
u
es
f
o
r
d
ep
r
ess
io
n
s
cr
ee
n
in
g
wh
ile
als
o
em
p
h
asizin
g
th
e
im
p
o
r
tan
ce
o
f
c
o
n
s
id
er
in
g
s
p
e
cif
ic
co
n
tex
ts
,
s
u
ch
as
m
ilit
ar
y
s
tatu
s
.
Su
b
r
am
an
ia
n
et
a
l.
[
1
9
]
in
v
esti
g
ated
th
e
r
o
le
o
f
s
o
cial
m
ed
ia
in
th
e
s
p
r
ea
d
o
f
h
o
s
tile
an
d
to
x
ic
co
n
ten
t,
in
clu
d
in
g
h
ate
s
p
ee
ch
an
d
ab
u
s
iv
e
lan
g
u
ag
e
.
T
h
eir
s
tu
d
y
em
p
lo
y
ed
tex
tu
al
r
ep
r
esen
tatio
n
an
d
en
co
d
in
g
t
ec
h
n
iq
u
es
lik
e
b
i
-
g
r
am
s
,
tr
i
-
g
r
am
s
,
an
d
Fas
tTe
x
t
to
ca
p
tu
r
e
s
em
an
tic
an
d
s
y
n
ta
ctic
in
f
o
r
m
atio
n
.
Ma
ch
in
e
an
d
d
ee
p
lear
n
in
g
m
eth
o
d
s
,
in
cl
u
d
in
g
C
NN,
B
E
R
T
,
an
d
SVM,
wer
e
u
s
ed
to
class
if
y
s
o
cial
m
ed
ia
p
o
s
ts
as
eit
h
er
in
d
icativ
e
o
f
d
ep
r
ess
io
n
o
r
n
o
t.
T
h
e
r
esu
lts
s
h
o
wed
th
at
SVM
ac
h
iev
ed
a
n
ac
cu
r
ac
y
o
f
8
0
% in
id
en
tif
y
in
g
d
ep
r
ess
io
n
s
y
m
p
to
m
s
.
Ou
r
r
ese
ar
ch
b
u
il
d
s
o
n
e
x
is
ti
n
g
s
t
u
d
ies
o
f
d
ep
r
ess
io
n
d
e
tec
ti
o
n
f
r
o
m
s
o
ci
al
n
e
tw
o
r
k
p
o
s
ts
,
ad
d
r
ess
i
n
g
s
ev
er
al
g
a
p
s
i
d
en
ti
f
ie
d
i
n
th
e
lit
er
at
u
r
e
.
W
h
ile
p
r
e
v
i
o
u
s
ap
p
r
o
ac
h
es
h
av
e
o
f
te
n
r
el
ie
d
o
n
s
t
an
d
a
r
d
M
L
tec
h
n
i
q
u
es
a
n
d
co
n
v
e
n
ti
o
n
a
l
NL
P
m
o
d
els
wi
th
li
m
it
e
d
h
y
p
er
p
a
r
a
m
e
te
r
o
p
ti
m
iz
ati
o
n
,
o
u
r
s
tu
d
y
i
n
t
r
o
d
u
ce
s
a
m
o
r
e
i
n
n
o
v
a
ti
v
e
an
d
in
te
g
r
at
ed
a
p
p
r
o
a
ch
.
W
e
e
n
h
a
n
c
e
m
o
d
el
p
er
f
o
r
m
a
n
ce
t
h
r
o
u
g
h
ad
v
a
n
c
e
d
te
c
h
n
iq
u
es,
in
c
lu
d
i
n
g
f
i
n
e
-
t
u
n
e
d
B
E
R
T
a
n
d
g
e
n
etic
a
lg
o
r
it
h
m
-
b
as
e
d
h
y
p
er
p
a
r
a
m
e
te
r
o
p
ti
m
iz
ati
o
n
,
s
e
tti
n
g
o
u
r
wo
r
k
a
p
a
r
t
f
r
o
m
ea
r
li
er
m
et
h
o
d
s
.
W
e
c
o
m
p
ar
e
d
v
a
r
i
o
u
s
te
x
t
v
ec
to
r
i
za
t
io
n
m
e
th
o
d
s
,
i
n
cl
u
d
in
g
T
F
-
I
D
F,
W
o
r
d
2
V
ec
,
an
d
a
B
E
R
T
m
o
d
el
s
p
e
ci
f
ic
all
y
t
u
n
e
d
t
o
o
u
r
d
a
tase
t,
e
n
h
a
n
ci
n
g
th
e
c
ap
tu
r
e
o
f
li
n
g
u
is
ti
c
n
u
a
n
c
e
s
in
o
n
li
n
e
f
o
r
u
m
s
.
Ad
d
it
io
n
a
ll
y
,
o
u
r
s
t
u
d
y
is
n
o
t
ab
l
e
f
o
r
in
c
o
r
p
o
r
ati
n
g
a
g
e
n
et
ic
al
g
o
r
it
h
m
f
o
r
h
y
p
e
r
p
a
r
a
m
e
ter
o
p
ti
m
iz
ati
o
n
,
a
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
4
7
4
-
1
4
8
4
1476
tec
h
n
i
q
u
e
o
f
t
e
n
o
v
e
r
l
o
o
k
e
d
o
r
a
d
d
r
ess
e
d
t
h
r
o
u
g
h
s
im
p
l
er
g
r
i
d
s
ea
r
ch
m
et
h
o
d
s
i
n
o
t
h
er
s
t
u
d
ies
.
T
h
is
ap
p
r
o
ac
h
all
o
ws
f
o
r
a
m
o
r
e
t
h
o
r
o
u
g
h
e
x
p
l
o
r
ati
o
n
o
f
th
e
h
y
p
e
r
p
a
r
a
m
ete
r
s
p
ac
e
,
r
es
u
lti
n
g
i
n
n
o
ta
b
l
e
i
m
p
r
o
v
e
m
e
n
ts
i
n
p
r
e
d
ic
ti
v
e
p
e
r
f
o
r
m
a
n
c
e.
Ad
d
i
t
io
n
al
ly
,
o
u
r
s
tu
d
y
ev
al
u
at
es
a
v
a
r
i
et
y
o
f
m
a
ch
in
e
l
ea
r
n
i
n
g
al
g
o
r
i
th
m
s
,
i
n
cl
u
d
i
n
g
d
e
cisi
o
n
tr
ee
,
l
o
g
is
ti
c
r
eg
r
ess
i
o
n
,
r
a
n
d
o
m
f
o
r
es
t,
g
r
a
d
ie
n
t
b
o
o
s
ti
n
g
,
a
n
d
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
i
n
e,
en
s
u
r
i
n
g
a
co
m
p
r
eh
e
n
s
i
v
e
a
n
d
r
i
g
o
r
o
u
s
as
s
ess
m
e
n
t
o
f
t
h
e
m
o
s
t
ef
f
e
cti
v
e
m
et
h
o
d
s
f
o
r
cl
ass
i
f
y
i
n
g
R
e
d
d
it
p
o
s
ts
.
T
h
e
f
o
llo
win
g
s
ec
tio
n
s
ar
e
o
r
g
an
ized
as
f
o
llo
ws:
s
ec
tio
n
2
d
escr
ib
es
th
e
m
ater
ials
an
d
m
eth
o
d
s
em
p
lo
y
ed
in
th
e
s
tu
d
y
.
Secti
o
n
3
p
r
esen
ts
an
d
an
aly
ze
s
th
e
r
esu
lts
,
h
ig
h
lig
h
tin
g
th
e
im
p
ac
t
o
f
th
e
ap
p
lied
tech
n
iq
u
es.
Sectio
n
4
ex
p
lain
s
th
e
lim
itatio
n
s
o
f
th
e
s
tu
d
y
.
Fin
ally
,
s
ec
tio
n
5
s
u
m
m
ar
izes
t
h
e
k
e
y
c
o
n
clu
s
io
n
s
an
d
o
f
f
er
s
s
u
g
g
esti
o
n
s
f
o
r
f
u
t
u
r
e
r
esear
ch
.
2.
M
AT
E
R
I
AL
S AN
D
M
E
T
H
O
DS
2
.
1
.
P
ro
po
s
ed
m
et
ho
do
lo
g
y
Fo
r
o
u
r
s
tu
d
y
o
n
d
ep
r
ess
io
n
d
etec
tio
n
f
r
o
m
R
ed
d
it
p
o
s
ts
,
we
u
tili
ze
d
a
d
ataset
co
m
p
r
is
in
g
7
,
7
3
1
en
tr
ies.
W
e
ad
o
p
ted
a
r
ig
o
r
o
u
s
m
eth
o
d
o
lo
g
y
in
co
r
p
o
r
ati
n
g
NL
P
an
d
ML
tech
n
iq
u
es
to
e
n
s
u
r
e
a
s
y
s
tem
atic
wo
r
k
f
lo
w.
T
h
e
p
r
o
ce
s
s
co
m
m
en
ce
d
with
th
e
ap
p
licatio
n
o
f
tr
ad
itio
n
al
NL
P
tech
n
iq
u
es,
in
clu
d
in
g
s
to
p
wo
r
d
r
em
o
v
al
an
d
lem
m
atiza
tio
n
,
to
p
r
ep
ar
e
th
e
tex
t
u
al
d
ata
f
o
r
d
etailed
an
aly
s
is
.
Nex
t,
we
v
ec
to
r
ized
th
e
tex
ts
u
s
in
g
th
r
ee
d
is
tin
ct
tech
n
iq
u
e
s
:
T
F
-
I
DF,
W
o
r
d
2
Vec
,
an
d
a
f
in
e
-
tu
n
ed
B
E
R
T
m
o
d
el
s
p
ec
if
ically
ad
ap
ted
to
o
u
r
d
ataset.
T
h
is
co
m
p
ar
is
o
n
allo
wed
u
s
to
as
s
es
s
th
e
ef
f
ec
tiv
en
ess
o
f
ea
ch
v
ec
to
r
izatio
n
m
eth
o
d
in
th
e
co
n
tex
t
o
f
d
ep
r
ess
io
n
d
etec
ti
o
n
.
All
m
o
d
els
wer
e
ev
alu
ate
d
u
s
in
g
5
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
to
en
s
u
r
e
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
en
t.
W
e
th
en
ap
p
lied
f
iv
e
m
ac
h
in
e
lear
n
in
g
al
g
o
r
ith
m
s
:
d
ec
is
io
n
tr
ee
,
lo
g
is
tic
r
eg
r
ess
io
n
,
r
an
d
o
m
f
o
r
est,
g
r
ad
ien
t
b
o
o
s
tin
g
,
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e.
T
o
en
h
a
n
ce
th
e
p
er
f
o
r
m
an
ce
o
f
th
ese
m
o
d
els,
we
em
p
l
o
y
ed
a
g
en
etic
alg
o
r
ith
m
f
o
r
h
y
p
er
p
ar
am
eter
o
p
tim
izatio
n
.
T
h
is
s
tr
u
ctu
r
ed
ap
p
r
o
ac
h
aim
s
to
m
ax
im
ize
th
e
ac
c
u
r
a
cy
o
f
d
e
p
r
ess
io
n
d
etec
tio
n
th
r
o
u
g
h
a
co
m
p
a
r
ativ
e
e
v
alu
atio
n
o
f
v
ec
to
r
izatio
n
tech
n
iq
u
es a
n
d
co
m
p
r
eh
en
s
iv
e
m
o
d
el
o
p
tim
izatio
n
.
T
h
e
s
tu
d
y
'
s
f
r
am
ewo
r
k
is
illu
s
tr
ated
in
Fig
u
r
e
1
.
Fig
u
r
e
1
.
Ov
e
r
v
iew
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
2
.
2
.
Da
t
a
s
et
T
h
e
d
ataset
u
s
ed
in
o
u
r
s
tu
d
y
,
av
ailab
le
o
n
Kag
g
le
[
2
0
]
,
co
n
s
is
ts
o
f
7
,
7
3
1
R
ed
d
it
p
o
s
ts
,
with
3
,
9
0
0
ca
teg
o
r
ized
as
d
ep
r
ess
iv
e
an
d
3
,
8
3
1
as
n
o
n
-
d
ep
r
ess
iv
e.
E
ac
h
p
o
s
t
h
as
an
av
er
ag
e
len
g
th
o
f
3
6
8
ch
ar
ac
te
r
s
.
T
h
is
d
ataset
is
p
ar
ticu
lar
ly
ad
v
an
tag
eo
u
s
f
o
r
d
e
p
r
ess
io
n
d
et
ec
tio
n
,
as
R
ed
d
it
p
o
s
ts
o
f
ten
f
ea
tu
r
e
d
ir
ec
t
an
d
s
p
o
n
tan
eo
u
s
ex
p
r
ess
io
n
s
o
f
u
s
er
s
'
em
o
tio
n
s
,
o
f
f
er
in
g
v
alu
a
b
le
in
s
ig
h
ts
f
o
r
id
en
tify
in
g
s
ig
n
s
o
f
d
ep
r
ess
io
n
.
T
h
e
d
ataset
is
well
b
alan
ce
d
,
with
5
0
.
4
%
o
f
p
o
s
ts
ca
teg
o
r
i
ze
d
as
d
ep
r
ess
iv
e
an
d
4
9
.
6
%
as
n
o
n
-
d
e
p
r
ess
iv
e,
elim
in
atin
g
th
e
n
ee
d
f
o
r
r
esam
p
lin
g
tech
n
iq
u
es
to
ad
d
r
ess
cl
ass
im
b
alan
ce
an
d
co
n
tr
ib
u
tin
g
to
a
m
o
r
e
r
eliab
le
ev
alu
atio
n
o
f
class
if
icatio
n
m
o
d
els.
Giv
en
th
is
b
alan
ce
,
n
o
d
ata
au
g
m
e
n
tatio
n
tech
n
iq
u
e
s
wer
e
ap
p
lied
,
as
ar
tific
ial
tr
an
s
f
o
r
m
atio
n
s
,
s
u
c
h
as
p
ar
ap
h
r
asin
g
o
r
b
ac
k
-
tr
an
s
latio
n
,
co
u
ld
in
tr
o
d
u
ce
b
i
ases
o
r
d
is
to
r
t
th
e
s
u
b
tle
lin
g
u
is
tic
m
ar
k
er
s
cr
itical
f
o
r
ac
cu
r
ate
class
if
icatio
n
.
Sin
ce
d
ep
r
ess
io
n
-
r
elate
d
l
an
g
u
ag
e
is
h
ig
h
l
y
n
u
an
ce
d
,
p
r
eser
v
in
g
th
e
a
u
th
en
ticity
o
f
u
s
er
-
g
e
n
er
ated
c
o
n
ten
t
en
s
u
r
es
th
at
th
e
m
o
d
el
lear
n
s
f
r
o
m
r
ea
l
d
ep
r
ess
iv
e
an
d
n
o
n
-
d
ep
r
ess
iv
e
tex
t
s
am
p
les,
en
h
an
cin
g
class
if
icatio
n
r
eliab
ilit
y
.
Ho
wev
e
r
,
it
is
im
p
o
r
tan
t
to
n
o
te
th
at
th
e
d
ataset
co
n
s
is
ts
e
x
clu
s
iv
ely
o
f
E
n
g
lis
h
-
lan
g
u
ag
e
p
o
s
ts
an
d
d
o
es
n
o
t
p
r
o
v
id
e
d
em
o
g
r
ap
h
ic
d
etails
s
u
ch
as
ag
e
o
r
s
o
cio
-
ec
o
n
o
m
ic
b
ac
k
g
r
o
u
n
d
,
wh
ich
m
ay
li
m
it
th
e
g
en
er
aliza
b
ilit
y
o
f
th
e
f
in
d
in
g
s
to
m
o
r
e
d
iv
er
s
e
p
o
p
u
latio
n
s
.
2
.
3
.
P
re
pro
ce
s
s
ing
t
ec
hn
iqu
es
Data
p
r
ep
r
o
ce
s
s
in
g
is
es
s
en
tial
in
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
,
en
s
u
r
in
g
th
at
tex
tu
al
d
ata
is
s
tr
u
ctu
r
ed
f
o
r
ef
f
ec
tiv
e
an
al
y
s
is
.
I
n
o
u
r
s
tu
d
y
,
we
ap
p
lied
s
tan
d
ar
d
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
,
in
clu
d
in
g
s
to
p
wo
r
d
r
em
o
v
al
an
d
lem
m
atiza
tio
n
,
to
r
ef
in
e
tex
tu
al
f
ea
tu
r
es a
n
d
en
h
an
ce
class
if
icatio
n
p
er
f
o
r
m
an
ce
[
2
1
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
tin
g
B
E
R
T fin
e
-
tu
n
in
g
a
n
d
g
e
n
etic
a
lg
o
r
ith
m
fo
r
…
(
A
b
d
A
lla
h
A
o
u
r
a
g
h
)
1477
2
.
3
.
1
.
Sto
p wo
rds
Sto
p
wo
r
d
r
e
m
o
v
al
elim
in
ate
s
f
r
eq
u
en
tly
o
cc
u
r
r
i
n
g
wo
r
d
s
(
e.
g
.
,
“th
e
,
”
“a
n
d
,
”
“in
”)
th
at
co
n
tr
ib
u
te
litt
le
to
th
e
s
em
an
tic
m
ea
n
in
g
o
f
a
s
en
ten
ce
.
T
h
is
s
tep
r
ed
u
ce
s
d
ata
d
im
en
s
io
n
ality
,
r
em
o
v
es
n
o
is
e,
an
d
im
p
r
o
v
es
co
m
p
u
tatio
n
al
ef
f
ici
en
cy
.
B
y
d
is
ca
r
d
in
g
n
o
n
-
in
f
o
r
m
ativ
e
wo
r
d
s
,
th
e
m
o
d
el
f
o
c
u
s
es
o
n
k
ey
ter
m
s
th
at
co
n
tr
ib
u
te
t
o
th
e
class
if
icatio
n
task
[
2
1
]
.
2
.
3
.
2
.
L
em
ma
t
iza
t
io
n
L
em
m
atiza
tio
n
s
tan
d
ar
d
izes
wo
r
d
s
b
y
co
n
v
e
r
tin
g
th
em
to
th
eir
b
ase
o
r
d
ictio
n
ar
y
f
o
r
m
wh
ile
p
r
eser
v
in
g
g
r
am
m
atica
l
m
ea
n
in
g
.
Un
lik
e
s
tem
m
in
g
,
wh
ich
tr
im
s
wo
r
d
s
to
th
eir
r
o
o
t
(
"r
u
n
n
in
g
,
"
"r
an
,
"
an
d
"r
u
n
s
"
to
"r
u
n
")
,
lem
m
atiza
ti
o
n
co
n
s
id
er
s
lin
g
u
is
tic
co
n
tex
t,
en
s
u
r
in
g
a
cc
u
r
ate
wo
r
d
tr
an
s
f
o
r
m
atio
n
s
.
T
h
is
p
r
o
ce
s
s
m
in
im
izes
r
ed
u
n
d
an
t
v
ar
iatio
n
s
in
tex
tu
al
d
ata,
lea
d
in
g
to
m
o
r
e
co
n
s
is
ten
t
an
d
m
ea
n
in
g
f
u
l
f
ea
tu
r
es
f
o
r
class
if
icatio
n
m
o
d
els
[
2
1
]
.
2
.
4
.
Vec
t
o
riza
t
io
n t
ec
hn
iqu
es
I
n
NL
P,
v
ec
to
r
izatio
n
is
cr
u
cial
f
o
r
co
n
v
er
tin
g
tex
t
in
to
n
u
m
er
ical
d
ata
th
at
ML
alg
o
r
ith
m
s
ca
n
p
r
o
ce
s
s
.
T
h
ese
m
eth
o
d
s
tr
an
s
f
o
r
m
wo
r
d
s
in
to
v
ec
to
r
s
,
ca
p
t
u
r
in
g
s
em
an
tic
a
n
d
c
o
n
tex
tu
a
l
r
elatio
n
s
h
ip
s
,
an
d
o
f
f
er
s
ig
n
if
ica
n
t
ad
v
a
n
tag
es
in
ac
cu
r
ac
y
an
d
f
lex
ib
ilit
y
f
o
r
tex
t
class
if
icatio
n
[
2
2
]
,
[
2
3
]
.
I
n
o
u
r
s
tu
d
y
,
we
s
elec
ted
th
e
f
o
llo
win
g
v
ec
to
r
i
za
tio
n
m
eth
o
d
s
f
o
r
th
eir
ef
f
icien
cy
in
ca
p
tu
r
in
g
co
m
p
lex
p
atter
n
s
in
lar
g
e
s
o
cial
m
ed
ia
d
atasets
an
d
th
eir
wid
es
p
r
ea
d
p
o
p
u
la
r
ity
.
2
.
4
.
1
.
T
F
-
I
DF
TF
-
I
DF
is
a
v
ec
to
r
izatio
n
tec
h
n
iq
u
e
in
NL
P
th
at
ass
ess
e
s
a
wo
r
d
'
s
s
ig
n
if
ican
ce
with
in
a
d
o
cu
m
en
t
r
elativ
e
to
a
lar
g
er
co
r
p
u
s
.
T
er
m
f
r
eq
u
en
cy
(
T
F)
ca
lcu
lates
h
o
w
f
r
eq
u
en
tly
a
te
r
m
ap
p
e
ar
s
in
a
d
o
cu
m
en
t,
wh
ile
in
v
er
s
e
d
o
cu
m
e
n
t
f
r
eq
u
en
cy
(
I
DF)
d
o
wn
weig
h
ts
ter
m
s
th
at
ar
e
co
m
m
o
n
ac
r
o
s
s
m
an
y
d
o
c
u
m
en
ts
.
B
y
b
alan
cin
g
lo
ca
l
f
r
eq
u
en
cy
with
g
lo
b
al
r
a
r
ity
,
T
F
-
I
DF
en
h
an
ce
s
th
e
ac
cu
r
ac
y
o
f
tex
t
u
al
r
ep
r
esen
tatio
n
s
,
m
ak
in
g
it e
f
f
ec
tiv
e
f
o
r
h
ig
h
lig
h
tin
g
im
p
o
r
tan
t w
o
r
d
s
in
co
n
t
ex
t
[
2
2
]
,
[
2
3
]
.
2
.
4
.
2
.
Wo
rd2
Vec
W
o
r
d
2
Vec
is
a
wo
r
d
v
ec
to
r
i
za
tio
n
tech
n
iq
u
e
th
at
r
e
p
r
esen
ts
wo
r
d
s
as
d
en
s
e
v
ec
to
r
s
i
n
a
h
ig
h
-
d
im
en
s
io
n
al
s
p
ac
e.
Dev
elo
p
e
d
b
y
Go
o
g
le,
th
is
m
eth
o
d
lear
n
s
th
ese
v
ec
to
r
r
ep
r
esen
tatio
n
s
b
y
an
al
y
zin
g
wo
r
d
co
-
o
cc
u
r
r
en
ce
s
with
in
a
tex
t
co
r
p
u
s
,
en
s
u
r
in
g
th
at
wo
r
d
s
a
p
p
ea
r
in
g
in
s
im
ilar
co
n
tex
ts
h
av
e
clo
s
ely
alig
n
e
d
v
ec
to
r
s
.
T
h
is
ap
p
r
o
ac
h
ef
f
ec
tiv
ely
ca
p
tu
r
es
s
em
an
tic
an
d
s
y
n
tactic
r
elatio
n
s
h
ip
s
b
etw
ee
n
wo
r
d
s
,
wh
er
e
s
im
ilar
m
ea
n
in
g
s
o
r
co
n
tex
tu
al
u
s
es
r
esu
lt
in
s
im
ilar
v
ec
t
o
r
p
o
s
itio
n
s
.
B
y
r
ed
u
cin
g
d
im
en
s
io
n
ality
wh
ile
p
r
eser
v
in
g
th
ese
r
elatio
n
s
h
ip
s
,
W
o
r
d
2
Vec
en
h
an
ce
s
th
e
q
u
al
ity
o
f
tex
t
u
al
an
aly
s
is
[
2
2
]
,
[
2
3
]
.
I
n
o
u
r
s
tu
d
y
,
we
ex
p
er
im
en
ted
with
v
ar
i
o
u
s
v
e
cto
r
d
im
en
s
io
n
s
a
n
d
f
o
u
n
d
th
at
a
d
im
en
s
io
n
o
f
5
0
0
y
ield
ed
th
e
b
est
r
esu
lts
f
o
r
d
etec
tin
g
d
ep
r
ess
io
n
in
th
e
p
o
s
t
-
co
n
tex
t.
2
.
4
.
3
.
F
ine
-
t
un
ed
B
E
R
T
B
E
R
T
i
s
a
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
m
o
d
el
th
at
lev
er
ag
e
s
a
b
id
ir
ec
tio
n
al
tr
an
s
f
o
r
m
er
ar
ch
itectu
r
e
to
g
r
asp
th
e
co
n
tex
t
o
f
wo
r
d
s
with
in
s
en
ten
ce
s
.
B
y
an
aly
zin
g
tex
t
in
b
o
th
d
ir
ec
tio
n
s
,
B
E
R
T
ca
p
tu
r
es
r
ich
er
co
n
tex
tu
al
in
f
o
r
m
atio
n
th
an
tr
ad
itio
n
al
m
o
d
els.
I
t
is
p
r
e
-
tr
ai
n
ed
o
n
v
ast
am
o
u
n
ts
o
f
tex
t
a
n
d
ca
n
b
e
f
in
e
-
t
u
n
ed
f
o
r
s
p
ec
if
ic
task
s
b
y
ad
d
in
g
tailo
r
ed
o
u
tp
u
t
lay
er
s
.
I
n
o
u
r
s
tu
d
y
,
we
f
in
e
-
tu
n
e
d
B
E
R
T
b
y
in
co
r
p
o
r
atin
g
ad
d
itio
n
al
o
u
tp
u
t
lay
e
r
s
s
p
ec
if
ically
d
esig
n
ed
f
o
r
o
u
r
p
o
s
t
-
class
if
icatio
n
task
.
T
h
is
m
o
d
e
l
was
r
e
-
tr
ain
ed
o
n
o
u
r
R
ed
d
it
d
ataset
to
en
h
an
c
e
its
ab
ilit
y
to
ca
p
tu
r
e
th
e
n
u
an
ce
s
o
f
o
u
r
co
r
p
u
s
,
u
ltima
tely
o
p
tim
izin
g
its
p
er
f
o
r
m
an
ce
f
o
r
d
e
p
r
ess
io
n
d
e
tectio
n
[
2
2
]
,
[
2
4
]
.
2
.
5
.
M
a
chine
lea
rni
ng
a
lg
o
rit
hm
s
Ma
ch
in
e
lear
n
in
g
alg
o
r
ith
m
s
ar
e
cr
u
cial
f
o
r
class
if
y
in
g
d
at
a
in
n
atu
r
al
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
.
T
h
e
y
f
ac
ilit
ate
th
e
id
en
tific
atio
n
an
d
m
o
d
elin
g
o
f
in
tr
icate
r
elatio
n
s
h
ip
s
b
etwe
en
in
p
u
t
f
ea
tu
r
es
an
d
tar
g
et
ca
teg
o
r
ies
u
s
in
g
tr
ain
in
g
d
ata
.
Ou
r
alg
o
r
ith
m
s
elec
tio
n
is
b
ased
o
n
th
eir
s
tr
o
n
g
r
e
p
u
tatio
n
an
d
wid
esp
r
ea
d
ad
o
p
tio
n
with
in
th
e
s
cien
tific
co
m
m
u
n
ity
f
o
r
tex
t
class
if
icatio
n
task
s
.
T
h
ese
alg
o
r
ith
m
s
a
r
e
k
n
o
wn
f
o
r
th
ei
r
ef
f
ec
tiv
en
ess
ac
r
o
s
s
d
iv
er
s
e
co
n
tex
ts
an
d
th
eir
ca
p
a
city
to
m
an
ag
e
co
m
p
lex
d
atasets
,
m
ak
in
g
th
em
id
ea
l
f
o
r
o
u
r
s
tu
d
y
[
2
5
]
–
[
2
7
]
.
I
n
o
u
r
r
esear
ch
,
we
em
p
l
o
y
ed
t
h
e
f
o
llo
win
g
tech
n
iq
u
es:
2
.
5
.
1
.
Dec
is
io
n
t
ree
T
h
e
d
ec
is
io
n
tr
ee
(
DT
)
alg
o
r
ith
m
is
a
class
if
icatio
n
m
o
d
el
t
h
at
s
p
lits
d
ata
in
to
s
u
b
s
ets
b
ased
o
n
th
e
v
alu
es
o
f
s
p
ec
if
ic
f
ea
tu
r
es.
E
ac
h
in
ter
n
al
n
o
d
e
r
ep
r
esen
ts
a
d
ec
is
io
n
b
ased
o
n
a
f
ea
tu
r
e,
wh
ile
ea
ch
leaf
n
o
d
e
co
r
r
esp
o
n
d
s
to
a
tar
g
et
class
.
T
h
is
alg
o
r
ith
m
is
h
ig
h
ly
v
alu
ed
f
o
r
its
s
tr
aig
h
tf
o
r
war
d
in
ter
p
r
etab
ilit
y
an
d
ca
p
ac
ity
to
ca
p
tu
r
e
co
m
p
lex
r
elatio
n
s
h
ip
s
b
etwe
en
f
ea
tu
r
e
s
.
It
i
s
clea
r
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
m
ak
es
it
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
f
o
r
class
if
icatio
n
task
s
th
at
r
eq
u
ir
e
t
r
an
s
p
ar
en
t a
n
d
e
f
f
icien
t r
esu
lts
[
2
5
]
–
[
2
7
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
4
7
4
-
1
4
8
4
1478
2
.
5
.
2
.
L
o
g
is
t
ic
r
eg
re
s
s
io
n
L
o
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
is
a
class
if
icatio
n
alg
o
r
ith
m
th
at
m
o
d
els
t
h
e
p
r
o
b
a
b
ilit
y
o
f
a
n
in
s
tan
ce
b
elo
n
g
in
g
to
a
p
a
r
ticu
lar
clas
s
u
s
in
g
a
lo
g
is
tic
f
u
n
ctio
n
.
T
h
is
m
o
d
el
is
h
i
g
h
ly
r
eg
ar
d
ed
f
o
r
its
s
im
p
licity
,
s
p
ee
d
,
an
d
ef
f
ec
tiv
en
ess
in
b
i
n
ar
y
class
if
icatio
n
task
s
,
d
eliv
er
in
g
r
esu
lts
th
at
ar
e
ea
s
y
to
i
n
ter
p
r
et.
M
o
r
eo
v
er
,
its
ab
ilit
y
to
o
u
tp
u
t
p
r
o
b
a
b
ilit
ies
alo
n
g
s
id
e
class
if
icatio
n
s
all
o
ws
f
o
r
a
m
o
r
e
n
u
an
ce
d
ass
ess
m
en
t
o
f
p
r
e
d
ictio
n
u
n
ce
r
tain
ties
,
en
h
a
n
cin
g
d
ec
is
io
n
-
m
ak
in
g
in
v
ar
io
u
s
ap
p
licatio
n
s
[
2
5
]
–
[
2
7
]
.
2
.
5
.
3
.
Ra
nd
o
m
f
o
re
s
t
R
an
d
o
m
f
o
r
est
(
R
F)
is
a
clas
s
if
icatio
n
alg
o
r
ith
m
th
at
en
h
a
n
ce
s
p
r
ed
ictio
n
ac
cu
r
ac
y
an
d
r
o
b
u
s
tn
ess
b
y
co
m
b
in
in
g
m
u
ltip
le
d
ec
is
io
n
tr
ee
s
.
E
ac
h
tr
ee
in
th
e
f
o
r
est
is
co
n
s
tr
u
cted
u
s
in
g
a
r
an
d
o
m
s
u
b
s
et
o
f
th
e
d
ata
an
d
f
ea
tu
r
es,
an
d
th
e
f
in
al
p
r
e
d
ictio
n
is
o
b
tain
ed
b
y
ag
g
r
e
g
a
tin
g
th
e
p
r
ed
ictio
n
s
o
f
all
th
e
tr
ee
s
.
T
h
is
en
s
em
b
le
ap
p
r
o
ac
h
m
itig
ates
th
e
r
is
k
o
f
o
v
e
r
f
itti
n
g
a
n
d
im
p
r
o
v
es
o
v
er
all
p
er
f
o
r
m
a
n
ce
b
y
lev
er
ag
in
g
th
e
d
iv
er
s
ity
am
o
n
g
t
h
e
in
d
iv
i
d
u
al
tr
ee
s
,
m
ak
in
g
it a
p
o
wer
f
u
l t
o
o
l f
o
r
co
m
p
lex
class
if
icatio
n
task
s
[
2
5
]
–
[
2
7
]
.
2
.
5
.
4
.
G
ra
dient
b
o
o
s
t
ing
Gr
ad
ien
t
b
o
o
s
tin
g
(
GB
)
is
a
class
if
icatio
n
alg
o
r
ith
m
th
at
c
o
n
s
tr
u
cts
a
r
o
b
u
s
t
p
r
ed
ictiv
e
m
o
d
el
b
y
s
eq
u
en
tially
co
m
b
in
in
g
m
u
lti
p
le
wea
k
lea
r
n
er
s
,
t
y
p
ically
d
ec
is
io
n
tr
ee
s
.
T
h
e
alg
o
r
ith
m
wo
r
k
s
iter
ativ
el
y
,
wh
er
e
ea
ch
n
ew
tr
ee
is
tr
ain
e
d
to
co
r
r
ec
t
th
e
er
r
o
r
s
m
ad
e
b
y
th
e
p
r
ev
io
u
s
tr
ee
s
b
y
f
o
cu
s
i
n
g
o
n
th
e
r
esid
u
als.
T
h
i
s
p
r
o
c
e
s
s
o
f
i
te
r
a
t
i
v
e
r
e
f
i
n
e
m
e
n
t
r
e
d
u
c
e
s
b
ia
s
a
n
d
e
n
h
an
c
e
s
t
h
e
o
v
e
r
a
ll
a
c
c
u
r
a
c
y
o
f
t
h
e
m
o
d
e
l
,
m
a
k
i
n
g
g
r
a
d
i
e
n
t
b
o
o
s
t
i
n
g
p
a
r
t
ic
u
l
a
r
l
y
ef
f
e
c
t
i
v
e
i
n
h
a
n
d
l
i
n
g
c
o
m
p
l
e
x
cla
s
s
i
f
i
c
at
i
o
n
t
as
k
s
wi
t
h
h
i
g
h
p
r
ec
i
s
i
o
n
[
2
5
]
–
[
2
7
]
.
2
.
5
.
5
.
Su
pp
o
rt
v
ec
t
o
r
m
a
chi
ne
T
h
e
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
is
a
p
o
wer
f
u
l
class
if
icatio
n
alg
o
r
ith
m
th
at
aim
s
to
f
in
d
th
e
o
p
tim
al
h
y
p
e
r
p
lan
e
th
at
s
ep
ar
ates
d
if
f
er
en
t
class
es
o
f
d
ata
with
th
e
lar
g
est
m
ar
g
in
.
I
t
ex
ce
ls
with
co
m
p
lex
,
h
ig
h
-
d
im
e
n
s
io
n
al
d
atasets
an
d
u
tili
ze
s
th
e
k
er
n
el
tr
ick
to
h
an
d
le
n
o
n
-
lin
ea
r
p
r
o
b
lem
s
b
y
p
r
o
jectin
g
d
ata
i
n
to
a
h
ig
h
er
-
d
im
en
s
io
n
al
s
p
ac
e,
th
er
eb
y
m
ak
i
n
g
th
em
lin
ea
r
ly
s
ep
ar
ab
le.
T
h
is
ab
ilit
y
to
h
an
d
le
n
o
n
-
lin
ea
r
ity
ef
f
ec
tiv
ely
en
h
an
ce
s
its
p
er
f
o
r
m
an
ce
ac
r
o
s
s
d
iv
er
s
e
class
if
icatio
n
task
s
[
2
5
]
–
[
2
7
]
.
2
.
6
.
H
y
perpa
ra
m
et
er
o
ptim
iza
t
io
n us
ing
g
enet
ic
a
lg
o
rit
hm
s
Hy
p
er
p
ar
a
m
eter
o
p
tim
izatio
n
is
ess
en
tial
f
o
r
en
h
an
cin
g
m
ac
h
in
e
lear
n
in
g
m
o
d
el
p
er
f
o
r
m
an
ce
b
y
f
in
e
-
tu
n
in
g
alg
o
r
ith
m
p
ar
a
m
eter
s
to
m
a
x
im
ize
ac
cu
r
ac
y
an
d
g
en
e
r
aliza
tio
n
[
2
8
]
.
I
n
th
is
s
t
u
d
y
,
we
em
p
lo
y
ed
th
e
Hy
p
ONI
C
lib
r
a
r
y
,
wh
ich
i
m
p
lem
en
ts
a
g
en
etic
alg
o
r
ith
m
(
GA)
in
s
p
ir
ed
b
y
n
at
u
r
al
s
e
lectio
n
to
ef
f
icien
tly
ex
p
lo
r
e
co
m
p
le
x
,
h
i
g
h
-
d
im
en
s
io
n
al
s
ea
r
ch
s
p
ac
es.
Un
lik
e
Gr
id
Sear
ch
,
wh
ich
ex
h
a
u
s
tiv
ely
ev
alu
ates
all
p
o
s
s
ib
le
p
ar
am
eter
c
o
m
b
in
ati
o
n
s
an
d
b
ec
o
m
es
c
o
m
p
u
tatio
n
ally
ex
p
e
n
s
iv
e
in
h
ig
h
-
d
im
e
n
s
io
n
al
s
ettin
g
s
,
o
r
B
ay
esian
o
p
tim
izatio
n
,
wh
ich
r
elies
o
n
p
r
o
b
ab
ilis
tic
m
o
d
elin
g
an
d
m
ay
s
tr
u
g
g
le
with
h
ig
h
ly
n
o
n
-
lin
ea
r
s
ea
r
ch
s
p
ac
es,
GA
iter
ativ
el
y
r
ef
i
n
es
co
n
f
ig
u
r
atio
n
s
th
r
o
u
g
h
ev
o
lu
tio
n
ar
y
o
p
er
atio
n
s
s
u
ch
as
s
elec
tio
n
,
cr
o
s
s
o
v
er
,
an
d
m
u
tatio
n
,
allo
win
g
f
o
r
ad
ap
tiv
e
e
x
p
lo
r
atio
n
an
d
th
e
av
o
id
an
ce
o
f
lo
ca
l
m
in
im
a
[
2
9
]
.
T
h
e
GA
was
co
n
f
ig
u
r
ed
with
a
p
o
p
u
la
tio
n
s
ize
o
f
1
0
a
n
d
was
ex
ec
u
ted
o
v
er
5
g
en
er
ati
o
n
s
to
b
ala
n
ce
co
m
p
u
tatio
n
al
ef
f
icien
cy
an
d
m
o
d
el
o
p
tim
iz
atio
n
.
A
m
u
tatio
n
r
ate
o
f
0
.
1
was
ap
p
lied
to
in
tr
o
d
u
ce
co
n
tr
o
lled
v
ar
ia
b
ilit
y
,
p
r
ev
en
tin
g
p
r
em
atu
r
e
co
n
v
er
g
en
ce
to
s
u
b
o
p
tim
al
s
o
lu
tio
n
s
,
wh
ile
a
cr
o
s
s
o
v
er
r
ate
o
f
0
.
9
f
ac
ilit
ated
th
e
co
m
b
in
atio
n
o
f
h
i
g
h
-
p
e
r
f
o
r
m
in
g
c
o
n
f
ig
u
r
atio
n
s
t
o
en
h
a
n
ce
ex
p
lo
r
atio
n
.
T
o
en
s
u
r
e
r
o
b
u
s
t
s
elec
tio
n
,
a
to
u
r
n
am
e
n
t
s
elec
tio
n
s
tr
ateg
y
was
em
p
lo
y
ed
,
wh
er
e
th
e
b
est
-
p
er
f
o
r
m
in
g
in
d
iv
id
u
als
wer
e
r
etain
ed
f
o
r
t
h
e
n
ex
t
g
en
er
atio
n
,
an
d
elitis
m
was
en
ab
led
to
p
r
eser
v
e
t
o
p
s
o
lu
tio
n
s
ac
r
o
s
s
iter
atio
n
s
.
B
y
lev
er
ag
in
g
ev
o
lu
tio
n
ar
y
p
r
in
cip
les,
GA
en
ab
les
a
m
o
r
e
e
f
f
icien
t
an
d
f
lex
i
b
le
h
y
p
er
p
a
r
am
eter
s
ea
r
ch
,
r
e
d
u
cin
g
co
m
p
u
tatio
n
al
c
o
s
ts
wh
ile
im
p
r
o
v
in
g
m
o
d
el
p
er
f
o
r
m
an
ce
an
d
s
tab
ilit
y
[
1
0
]
,
[
3
0
]
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
W
e
im
p
lem
en
ted
o
u
r
ex
p
er
i
m
en
tal
f
r
am
ew
o
r
k
o
n
Go
o
g
l
e
C
o
lab
'
s
clo
u
d
p
latf
o
r
m
,
le
v
er
ag
in
g
its
16
GB
GPU
ac
ce
ler
atio
n
to
h
an
d
le
th
e
c
o
m
p
u
tatio
n
al
wo
r
k
lo
ad
.
T
h
e
tech
n
ical
s
tack
u
tili
ze
d
Py
t
h
o
n
alo
n
g
s
id
e
ess
en
tial
m
ac
h
in
e
lear
n
in
g
lib
r
ar
ies:
Pan
d
as
f
o
r
d
ata
m
an
ip
u
latio
n
,
Scik
it
-
lear
n
f
o
r
tr
ad
itio
n
al
alg
o
r
ith
m
s
,
T
e
n
s
o
r
Flo
w
f
o
r
n
eu
r
al
n
etwo
r
k
im
p
lem
e
n
tatio
n
,
n
atu
r
al
lan
g
u
ag
e
to
o
lk
it
(
NL
T
K)
f
o
r
tex
t
p
r
ep
r
o
ce
s
s
in
g
,
Gen
s
im
f
o
r
wo
r
d
em
b
e
d
d
in
g
s
,
T
r
an
s
f
o
r
m
er
s
f
o
r
B
E
R
T
-
b
ased
ar
ch
itectu
r
es,
an
d
Hy
p
ONI
C
f
o
r
h
y
p
er
p
ar
am
eter
o
p
tim
izatio
n
.
W
h
ile
tr
an
s
f
o
r
m
er
m
o
d
e
ls
lik
e
B
E
R
T
ty
p
ically
d
em
an
d
s
ig
n
if
ican
t
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
,
o
u
r
d
ataset
o
f
7
,
7
3
1
p
o
s
ts
allo
we
d
f
o
r
e
f
f
icien
t
f
in
e
-
t
u
n
in
g
,
co
m
p
letin
g
tr
ain
in
g
in
p
r
ac
tical
tim
ef
r
am
es
with
o
u
t
r
eq
u
ir
in
g
s
p
ec
ialized
h
ar
d
war
e.
Fo
r
c
o
m
p
r
e
h
en
s
iv
e
e
v
alu
at
io
n
,
we
e
m
p
lo
y
ed
m
u
ltip
le
p
er
f
o
r
m
a
n
ce
m
etr
ics:
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
-
s
co
r
e
[
3
1
]
,
[
3
2
]
,
wh
ich
co
llectiv
ely
ass
ess
d
if
f
er
en
t
asp
ec
ts
o
f
m
o
d
el
ef
f
ec
tiv
en
ess
.
T
h
e
co
m
p
lete
ex
p
er
im
en
tal
r
esu
lts
,
p
r
esen
ted
in
T
ab
les
1
-
6
,
s
y
s
tem
atica
lly
co
m
p
ar
e
t
h
e
i
m
p
ac
t
o
f
v
a
r
io
u
s
v
ec
to
r
izatio
n
ap
p
r
o
ac
h
es
(
T
F
-
I
DF,
W
o
r
d
2
Vec
,
B
E
R
T
)
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
tin
g
B
E
R
T fin
e
-
tu
n
in
g
a
n
d
g
e
n
etic
a
lg
o
r
ith
m
fo
r
…
(
A
b
d
A
lla
h
A
o
u
r
a
g
h
)
1479
g
en
etic
alg
o
r
ith
m
o
p
tim
izatio
n
ac
r
o
s
s
o
u
r
f
iv
e
m
ac
h
i
n
e
lear
n
in
g
ar
ch
itectu
r
es.
T
h
is
b
alan
ce
d
ap
p
r
o
ac
h
b
etwe
en
co
m
p
u
tatio
n
al
ef
f
icien
cy
an
d
m
eth
o
d
o
lo
g
ical
r
ig
o
r
d
e
m
o
n
s
tr
ates
th
at
o
u
r
s
o
lu
tio
n
ca
n
b
e
r
ea
lis
tically
d
ep
lo
y
ed
o
n
s
tan
d
ar
d
cl
o
u
d
in
f
r
astru
ctu
r
e,
m
a
k
in
g
it
ac
ce
s
s
ib
le
f
o
r
p
r
ac
tica
l
ap
p
licatio
n
s
wh
ile
m
ain
tain
in
g
r
o
b
u
s
t p
er
f
o
r
m
a
n
ce
s
tan
d
ar
d
s
.
T
ab
le
1
.
T
F
-
I
DF
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
9
0
.
6
9
%
9
0
.
9
1
%
9
0
.
3
1
%
9
0
.
6
1
%
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
(
L
R
)
9
4
.
9
6
%
9
8
.
0
6
%
9
2
.
6
7
%
9
5
.
2
9
%
R
a
n
d
o
m f
o
r
e
s
t
(
R
F
)
8
5
.
4
6
%
9
6
.
3
9
%
7
3
.
3
0
%
8
3
.
2
7
%
G
r
a
d
i
e
n
t
b
o
o
t
i
n
g
(
G
B
)
9
3
.
3
4
%
9
8
.
8
6
%
9
0
.
7
1
%
9
4
.
6
1
%
S
u
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
(
S
V
M
)
7
5
.
9
5
%
6
8
.
2
2
%
9
6
.
0
7
%
7
9
.
7
8
%
T
ab
le
2
.
W
o
r
d
2
Vec
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
8
2
.
2
9
%
8
1
.
4
5
%
8
3
.
9
0
%
8
2
.
6
6
%
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
(
L
R
)
8
6
.
3
0
%
8
1
.
3
6
%
9
3
.
7
2
%
8
7
.
1
0
%
R
a
n
d
o
m f
o
r
e
s
t
(
R
F
)
8
4
.
6
8
%
8
4
.
2
6
%
8
3
.
3
8
%
8
3
.
8
2
%
G
r
a
d
i
e
n
t
b
o
o
t
i
n
g
(
G
B
)
8
8
.
6
2
%
8
9
.
5
7
%
8
9
.
9
2
%
8
9
.
7
5
%
S
u
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
(
S
V
M
)
8
7
.
5
9
%
8
2
.
9
5
%
9
4
.
2
4
%
8
8
.
2
4
%
T
ab
le
3
.
Fin
e
-
tu
n
ed
B
E
R
T
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
8
6
.
8
8
%
8
6
.
8
6
%
8
6
.
5
2
%
8
6
.
6
9
%
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
(
L
R
)
9
5
.
4
1
%
9
6
.
3
9
%
9
4
.
2
4
%
9
5
.
3
0
%
R
a
n
d
o
m f
o
r
e
s
t
(
R
F
)
8
8
.
6
9
%
9
6
.
6
7
%
7
9
.
8
4
%
8
7
.
4
6
%
G
r
a
d
i
e
n
t
b
o
o
t
i
n
g
(
G
B
)
9
4
.
4
4
%
9
7
.
0
8
%
9
1
.
4
9
%
9
4
.
2
0
%
S
u
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
(
S
V
M
)
9
5
.
0
9
%
9
5
.
8
7
%
9
4
.
1
1
%
9
4
.
9
8
%
T
ab
le
4
.
T
F
-
I
DF +
Gen
etic
al
g
o
r
ith
m
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
9
2
.
3
1
%
9
5
.
1
0
%
8
9
.
0
1
%
9
1
.
9
5
%
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
(
L
R
)
9
5
.
4
8
%
9
5
.
2
5
%
9
4
.
5
0
%
9
4
.
8
8
%
R
a
n
d
o
m f
o
r
e
s
t
(
R
F
)
9
5
.
6
0
%
9
8
.
1
5
%
7
6
.
3
1
%
8
5
.
8
6
%
G
r
a
d
i
e
n
t
b
o
o
t
i
n
g
(
G
B
)
9
4
.
5
7
%
9
6
.
2
0
%
9
2
.
6
7
%
9
4
.
4
0
%
S
u
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
(
S
V
M
)
9
5
.
4
1
%
9
4
.
8
5
%
9
3
.
9
8
%
9
4
.
4
1
%
T
ab
le
5
.
W
o
r
d
2
Vec
+
Gen
etic
alg
o
r
ith
m
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
8
2
.
6
1
%
8
1
.
0
9
%
8
3
.
6
4
%
8
2
.
3
5
%
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
(
L
R
)
8
9
.
7
9
%
8
5
.
5
6
%
9
5
.
4
2
%
9
0
.
2
2
%
R
a
n
d
o
m f
o
r
e
s
t
(
R
F
)
8
6
.
1
7
%
8
5
.
9
0
%
8
6
.
1
3
%
8
6
.
0
1
%
G
r
a
d
i
e
n
t
b
o
o
t
i
n
g
(
G
B
)
9
0
.
5
0
%
9
0
.
5
4
%
9
0
.
1
8
%
9
0
.
3
6
%
S
u
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
(
S
V
M
)
9
0
.
8
9
%
9
1
.
5
9
%
8
9
.
7
9
%
9
0
.
6
8
%
T
ab
le
6
.
Fin
e
-
tu
n
ed
B
E
R
T
+
Gen
etic
alg
o
r
ith
m
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
D
e
c
i
s
i
o
n
t
r
e
e
(
D
T)
8
7
.
2
7
%
8
7
.
5
5
%
8
6
.
5
2
%
8
7
.
0
3
%
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
(
L
R
)
9
5
.
6
0
%
9
6
.
5
2
%
9
4
.
5
0
%
9
5
.
5
0
%
R
a
n
d
o
m f
o
r
e
s
t
(
R
F
)
8
9
.
4
6
%
9
6
.
8
8
%
8
1
.
2
8
%
8
8
.
4
0
%
G
r
a
d
i
e
n
t
b
o
o
t
i
n
g
(
G
B
)
9
4
.
9
6
%
9
7
.
6
4
%
9
2
.
0
2
%
9
4
.
7
4
%
S
u
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
(
S
V
M
)
9
5
.
9
9
%
9
6
.
6
8
%
9
5
.
1
6
%
9
5
.
9
1
%
Fo
r
th
e
d
ataset
v
ec
to
r
ize
d
u
s
i
n
g
T
F
-
I
DF,
as
s
h
o
wn
in
T
a
b
l
e
1
,
lo
g
is
tic
r
eg
r
ess
io
n
d
eliv
er
ed
th
e
b
est
o
v
er
all
p
er
f
o
r
m
an
c
e
with
an
a
cc
u
r
ac
y
o
f
9
4
.
9
6
%
an
d
an
F1
-
s
co
r
e
o
f
9
5
.
2
9
%.
Gr
ad
ien
t
b
o
o
s
tin
g
ac
h
iev
ed
th
e
h
ig
h
est
p
r
ec
is
io
n
at
9
8
.
8
6
%,
wh
ile
SVM
ex
ce
lled
in
r
ec
all
with
9
6
.
0
7
%.
Fo
r
t
h
e
W
o
r
d
2
Vec
v
ec
to
r
izatio
n
,
d
etailed
in
T
ab
le
2
,
g
r
a
d
ien
t
b
o
o
s
tin
g
s
h
o
wed
s
tr
o
n
g
p
er
f
o
r
m
an
ce
with
an
ac
cu
r
ac
y
o
f
8
8
.
6
2
%,
p
r
ec
is
io
n
o
f
8
9
.
5
7
%,
an
d
an
F1
-
s
co
r
e
o
f
8
9
.
7
5
%.
SVM
ag
ain
d
em
o
n
s
tr
a
ted
s
u
p
er
io
r
r
ec
all
with
a
s
co
r
e
o
f
9
4
.
2
4
%.
L
astl
y
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
4
7
4
-
1
4
8
4
1480
w
i
t
h
t
h
e
f
i
n
e
-
t
u
n
e
d
B
E
R
T
m
o
d
e
l
,
a
s
p
r
e
s
e
n
te
d
i
n
T
a
b
l
e
3
,
l
o
g
i
s
t
i
c
r
e
g
r
es
s
i
o
n
l
e
d
w
i
t
h
a
n
a
c
c
u
r
a
c
y
o
f
9
5
.
4
1
%
,
r
e
c
a
l
l
o
f
9
4
.
2
4
%
,
a
n
d
a
n
F
1
-
s
co
r
e
o
f
9
5
.
3
0
%
,
w
h
i
l
e
g
r
a
d
i
e
n
t
b
o
o
s
t
i
n
g
a
c
h
i
e
v
e
d
t
h
e
h
i
g
h
es
t
p
r
e
c
i
s
i
o
n
a
t
9
7
.
0
8
%
.
Fo
llo
win
g
h
y
p
er
p
ar
am
eter
o
p
tim
izatio
n
with
th
e
g
en
e
tic
alg
o
r
ith
m
,
s
ig
n
if
ican
t
p
er
f
o
r
m
a
n
ce
im
p
r
o
v
em
e
n
ts
wer
e
o
b
s
er
v
e
d
f
o
r
th
e
T
F
-
I
DF
m
o
d
el
(
T
ab
l
e
4
)
,
with
a
n
o
tab
le
1
9
%
en
h
an
ce
m
en
t
f
o
r
SVM
(
C
:
6
.
4
2
,
g
am
m
a:
'
s
ca
le
'
,
k
er
n
el:
'
lin
ea
r
'
,
d
eg
r
ee
:
4
)
.
R
an
d
o
m
f
o
r
est
(
n
_
esti
m
ato
r
s
:
4
0
,
cr
iter
io
n
:
'
g
in
i'
,
m
in
_
s
am
p
les_
s
p
lit:
9
,
m
ax
_
d
ep
th
:
7
)
ac
h
iev
ed
th
e
h
i
g
h
est
ac
cu
r
ac
y
(
9
5
.
6
0
%)
an
d
p
r
ec
is
io
n
(
9
8
.
1
5
%),
wh
ile
lo
g
is
tic
r
eg
r
ess
io
n
ex
ce
lled
in
r
ec
all
(
9
4
.
5
0
%)
an
d
F1
-
s
co
r
e
(
9
4
.
8
8
%).
Fo
r
t
h
e
W
o
r
d
2
Ve
c
m
o
d
el
(
T
a
b
le
5
)
,
o
p
tim
izatio
n
r
esu
lted
in
an
a
v
er
ag
e
p
e
r
f
o
r
m
an
ce
b
o
o
s
t
o
f
2
%.
SVM
(
C
:
9
.
6
2
,
g
am
m
a:
'
au
to
'
,
k
er
n
el:
'
lin
ea
r
'
,
d
eg
r
ee
:
3
)
d
e
m
o
n
s
tr
ated
t
h
e
b
est
ac
cu
r
ac
y
(
9
0
.
8
9
%),
p
r
ec
is
io
n
(
9
1
.
5
9
%),
a
n
d
F1
-
s
co
r
e
(
9
0
.
6
8
%),
w
h
ile
lo
g
is
tic
r
eg
r
ess
io
n
(
p
en
alty
:
'
l2
'
,
C
:
8
7
0
.
1
5
,
m
ax
_
iter
:
1
8
0
,
s
o
lv
er
:
'
n
ewto
n
-
c
g
'
)
ac
h
iev
e
d
th
e
h
ig
h
est
r
ec
all
(
9
5
.
4
2
%).
W
ith
th
e
f
in
e
-
tu
n
ed
B
E
R
T
m
o
d
el
(
T
ab
le
6
)
,
h
y
p
er
p
ar
am
eter
o
p
tim
izatio
n
led
to
an
a
v
er
ag
e
im
p
r
o
v
em
e
n
t
o
f
1
%.
SVM
(
C
:
9
.
6
7
,
g
am
m
a:
'
au
to
'
,
k
er
n
e
l:
'
lin
ea
r
'
,
d
eg
r
ee
:
4
)
p
e
r
f
o
r
m
ed
b
est
in
ac
cu
r
ac
y
(
9
5
.
9
9
%),
r
ec
all
(
9
5
.
1
6
%),
an
d
F1
-
s
co
r
e
(
9
5
.
9
1
%),
wh
i
le
g
r
ad
ien
t
b
o
o
s
tin
g
(
lo
s
s
:
'
l
o
g
_
lo
s
s
'
,
cr
iter
io
n
:
'
f
r
ied
m
an
_
m
s
e'
,
n
_
esti
m
ato
r
s
:
7
0
,
m
a
x
_
d
e
p
th
:
7
,
m
in
_
s
a
m
p
les_
s
p
lit:
4
)
ac
h
iev
ed
th
e
h
ig
h
est
p
r
ec
is
io
n
(
9
7
.
6
4
%).
T
o
en
s
u
r
e
th
e
r
o
b
u
s
tn
ess
o
f
th
ese
r
esu
lts
,
we
e
m
p
lo
y
ed
k
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
,
wh
ich
m
itig
ates
p
er
f
o
r
m
an
ce
v
ar
iatio
n
s
b
y
ev
alu
atin
g
m
o
d
els
ac
r
o
s
s
m
u
ltip
le
tr
ain
in
g
an
d
test
in
g
s
p
lits
.
Giv
en
th
is
r
ig
o
r
o
u
s
v
alid
atio
n
a
p
p
r
o
ac
h
,
we
d
id
n
o
t
p
er
f
o
r
m
ad
d
itio
n
al
s
tatis
tic
al
s
ig
n
if
ican
ce
test
s
s
u
ch
as
ANOV
A
o
r
t
-
test
s
,
wh
ich
ar
e
m
o
r
e
s
u
ited
f
o
r
c
o
m
p
ar
in
g
m
o
d
els
tr
ain
ed
o
n
in
d
ep
en
d
e
n
t
d
atasets
r
ath
er
th
an
th
o
s
e
ass
es
s
ed
u
s
in
g
cr
o
s
s
-
v
alid
atio
n
.
Fig
u
r
e
2
p
r
o
v
id
es
a
g
r
ap
h
ical
o
v
e
r
v
iew
o
f
th
e
p
er
f
o
r
m
a
n
ce
o
f
v
ar
io
u
s
m
o
d
els,
o
f
f
e
r
in
g
a
clea
r
an
d
in
tu
itiv
e
v
is
u
aliza
tio
n
o
f
th
e
r
esu
lts
.
Fig
u
r
e
2
.
Me
tr
ics
an
aly
s
is
f
o
r
d
if
f
er
en
t a
lg
o
r
ith
m
s
T
h
e
r
esu
lts
o
f
o
u
r
s
tu
d
y
in
d
ic
ate
th
at
th
e
SVM
m
o
d
el,
wh
e
n
co
m
b
in
e
d
with
a
f
in
e
-
t
u
n
ed
B
E
R
T
an
d
o
p
tim
ized
u
s
in
g
a
g
en
etic
alg
o
r
ith
m
,
ac
h
iev
ed
th
e
h
ig
h
est
o
v
er
all
p
e
r
f
o
r
m
an
ce
.
I
t
attain
ed
an
ac
cu
r
ac
y
o
f
9
5
.
9
9
%,
a
r
ec
all
o
f
9
5
.
1
6
%,
an
d
an
F1
-
s
co
r
e
o
f
9
5
.
9
1
%.
T
h
is
ex
ce
p
tio
n
al
p
er
f
o
r
m
an
ce
o
f
th
e
SVM
is
attr
ib
u
ted
to
its
ca
p
ac
ity
to
o
p
tim
ize
th
e
m
ar
g
in
b
etwe
e
n
class
es
th
r
o
u
g
h
p
r
ec
is
e
d
e
cisi
o
n
h
y
p
er
p
la
n
es,
th
er
eb
y
en
h
an
cin
g
its
ac
cu
r
ac
y
in
class
if
y
in
g
p
o
s
itiv
e
ex
am
p
les.
T
h
e
f
in
e
-
tu
n
ed
B
E
R
T
m
o
d
el
was
p
iv
o
tal
in
ac
h
iev
in
g
th
ese
r
esu
lts
.
I
ts
ad
ap
tatio
n
to
o
u
r
d
ataset
allo
wed
it
to
ef
f
ec
tiv
ely
ca
p
tu
r
e
co
n
tex
tu
al
n
u
an
ce
s
an
d
co
m
p
lex
r
elatio
n
s
h
ip
s
in
R
ed
d
it
co
m
m
en
ts
,
lea
d
in
g
to
s
ig
n
if
ican
t
p
er
f
o
r
m
an
ce
im
p
r
o
v
e
m
en
ts
.
Ad
d
itio
n
ally
,
TF
-
I
DF
o
u
tp
er
f
o
r
m
ed
W
o
r
d
2
Vec
in
o
u
r
s
tu
d
y
.
T
h
is
is
d
u
e
to
T
F
-
I
DF's
ef
f
ec
tiv
en
ess
i
n
h
ig
h
lig
h
tin
g
ter
m
im
p
o
r
tan
ce
in
s
h
o
r
t,
s
p
ec
if
ic
t
ex
ts
lik
e
R
ed
d
it
c
o
m
m
en
ts
,
wh
er
ea
s
W
o
r
d
2
Vec
,
wh
ich
ex
ce
ls
in
ca
p
tu
r
in
g
d
ee
p
s
em
an
tic
r
elatio
n
s
h
ip
s
,
was
n
o
t
as
wel
l
-
s
u
ited
to
o
u
r
d
ataset.
T
h
e
lack
o
f
s
p
ec
if
ic
o
p
tim
izatio
n
f
o
r
W
o
r
d
2
Vec
lik
ely
co
n
s
tr
ain
ed
its
ab
ilit
y
to
ca
p
tu
r
e
th
e
ess
en
tial
co
n
tex
tu
al
n
u
an
ce
s
f
o
r
ac
c
u
r
ate
class
if
icatio
n
,
em
p
h
asizin
g
th
e
n
ee
d
f
o
r
v
ec
to
r
izatio
n
m
eth
o
d
s
to
b
e
tailo
r
ed
to
th
e
s
p
ec
if
ic
c
o
n
tex
t
o
f
th
e
d
ata.
L
o
g
is
tic
r
eg
r
ess
io
n
also
d
eliv
er
ed
im
p
r
ess
iv
e
r
esu
lts
,
ac
h
iev
in
g
an
ac
cu
r
ac
y
o
f
9
5
.
6
0
%
an
d
a
n
F
1
-
s
co
r
e
o
f
9
5
.
5
0
%,
h
ig
h
lig
h
tin
g
its
ef
f
ec
tiv
en
ess
i
n
ca
p
tu
r
in
g
lin
ea
r
r
elatio
n
s
h
ip
s
.
Gr
ad
ien
t
b
o
o
s
tin
g
ex
ce
lled
in
p
r
ec
is
io
n
with
a
s
co
r
e
o
f
9
7
.
6
4
%,
s
h
o
wca
s
in
g
its
s
tr
en
g
th
in
ag
g
r
eg
atin
g
wea
k
tr
ee
s
to
co
r
r
ec
t
class
if
icatio
n
er
r
o
r
s
.
T
h
e
65
.
00
%
70.
00%
75.
00%
80.
00%
85.
00%
90.
00%
95.
00%
10
0
.
00
%
DT
LR
RF
GB
S
V
M
D
T
+
G
A
L
R
+
G
A
R
F
+
G
A
G
B
+
G
A
S
V
M
+
G
A
DT
LR
RF
GB
S
V
M
D
T
+
G
A
L
R
+
G
A
R
F
+
G
A
G
B
+
G
A
S
V
M
+
G
A
DT
LR
RF
GB
S
V
M
D
T
+
G
A
L
R
+
G
A
R
F
+
G
A
G
B
+
G
A
S
V
M
+
G
A
TF
-
I
D
F
W
O
R
D
2
V
E
C
F
I
N
E
-
T
U
N
E
D
B
E
R
T
a
c
c
u
ra
c
y
p
r
e
c
isio
n
re
c
a
ll
f1
-s
c
o
re
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
tin
g
B
E
R
T fin
e
-
tu
n
in
g
a
n
d
g
e
n
etic
a
lg
o
r
ith
m
fo
r
…
(
A
b
d
A
lla
h
A
o
u
r
a
g
h
)
1481
g
en
etic
alg
o
r
ith
m
p
lay
e
d
a
cr
u
cial
r
o
le
in
o
p
tim
izin
g
h
y
p
er
p
ar
am
eter
s
b
y
ef
f
icie
n
tly
ex
p
lo
r
in
g
p
ar
am
ete
r
s
p
ac
es
an
d
s
ig
n
if
ican
tly
en
h
an
cin
g
m
o
d
el
p
e
r
f
o
r
m
an
ce
.
T
h
is
m
eth
o
d
allo
wed
u
s
to
f
in
e
-
tu
n
e
p
ar
am
ete
r
s
ef
f
ec
tiv
ely
,
ac
h
ie
v
in
g
o
p
tim
al
co
n
f
ig
u
r
atio
n
s
an
d
th
e
b
est p
o
s
s
ib
le
r
esu
lts
.
C
o
m
p
ar
in
g
o
u
r
r
esu
lts
with
th
o
s
e
o
f
p
r
ev
i
o
u
s
s
tu
d
ies,
s
u
ch
as
[
1
2
]
,
[
1
3
]
,
we
f
i
n
d
s
im
ilar
ities
in
th
e
ap
p
licatio
n
o
f
m
ac
h
in
e
lear
n
in
g
tech
n
i
q
u
es
f
o
r
d
ep
r
ess
io
n
d
etec
tio
n
.
T
h
ese
s
tu
d
ies
also
h
ig
h
lig
h
t
th
e
ef
f
ec
tiv
en
ess
o
f
m
o
d
els lik
e
SVM
in
ac
h
iev
in
g
r
o
b
u
s
t r
esu
lts
.
Fo
r
in
s
tan
ce
,
[
1
4
]
r
e
p
o
r
ted
t
h
at
SVMs a
ch
iev
ed
an
ac
cu
r
ac
y
o
f
7
7
.
1
2
%
an
d
an
F1
-
s
co
r
e
o
f
7
7
%.
I
n
a
n
o
th
er
ca
s
e,
[
1
5
]
o
b
s
er
v
e
d
th
at
R
an
d
o
m
Fo
r
est
p
er
f
o
r
m
ed
well
with
an
ac
cu
r
ac
y
o
f
8
4
.
9
9
%
in
a
n
aly
zin
g
s
o
cial
n
etwo
r
k
p
o
s
ts
,
alth
o
u
g
h
th
eir
r
esu
lts
wer
e
lo
wer
th
an
th
o
s
e
o
b
tain
ed
i
n
o
u
r
s
tu
d
y
.
I
n
ad
d
itio
n
,
[
1
6
]
u
tili
ze
d
a
h
y
b
r
id
C
NN
an
d
B
iLST
M
m
o
d
el,
ac
h
iev
in
g
an
ac
cu
r
ac
y
o
f
9
4
.
2
8
%
o
n
d
ep
r
ess
io
n
-
r
elate
d
twee
ts
.
[
1
7
]
d
em
o
n
s
tr
ated
th
at
SVMs
co
u
ld
r
ea
ch
8
5
%
ac
cu
r
ac
y
o
n
twee
ts
u
s
in
g
v
ar
i
o
u
s
class
if
ier
s
.
Fu
r
th
er
m
o
r
e,
[
1
8
]
v
alid
ated
d
ee
p
lea
r
n
in
g
tech
n
iq
u
es
lik
e
B
E
R
T
,
ac
h
iev
in
g
an
AUC
o
f
9
3
%
f
o
r
d
ep
r
ess
io
n
d
etec
ti
o
n
in
m
ed
ical
r
ec
o
r
d
s
,
alth
o
u
g
h
th
eir
m
o
d
el'
s
p
er
f
o
r
m
an
ce
v
a
r
ied
d
e
p
en
d
i
n
g
o
n
th
e
c
o
n
tex
t.
Ou
r
s
tu
d
y
s
ets
its
elf
ap
ar
t
b
y
in
co
r
p
o
r
atin
g
a
g
en
etic
alg
o
r
i
th
m
f
o
r
h
y
p
er
p
a
r
am
eter
o
p
tim
izatio
n
,
a
m
eth
o
d
o
l
o
g
y
s
eld
o
m
ex
p
lo
r
ed
in
p
r
ev
io
u
s
r
esear
ch
o
n
d
ep
r
ess
io
n
d
etec
tio
n
.
B
y
f
in
e
-
tu
n
in
g
m
o
d
el
co
n
f
ig
u
r
atio
n
s
,
th
is
ap
p
r
o
ac
h
led
to
9
5
.
9
9
%
ac
cu
r
ac
y
,
9
5
.
1
6
%
r
ec
all,
an
d
a
9
5
.
9
1
%
F1
-
s
co
r
e
with
th
e
f
i
n
e
-
tu
n
ed
SVM
an
d
B
E
R
T
m
o
d
el.
Prio
r
s
tu
d
ies
p
r
im
ar
ily
r
elied
o
n
co
n
v
en
tio
n
al
tu
n
i
n
g
s
tr
ateg
ies
s
u
ch
as
g
r
id
s
ea
r
ch
an
d
r
an
d
o
m
s
ea
r
c
h
,
wh
ich
lack
t
h
e
ad
a
p
tiv
e
ca
p
ab
ilit
ies
o
f
g
en
etic
o
p
tim
izat
io
n
[
1
7
]
,
[
1
9
]
.
T
h
e
in
teg
r
atio
n
o
f
B
E
R
T
f
in
e
-
t
u
n
in
g
with
ev
o
l
u
tio
n
ar
y
o
p
tim
izatio
n
s
ig
n
if
ican
tly
en
h
a
n
ce
s
class
if
icat
io
n
ef
f
ec
tiv
en
ess
,
m
ak
in
g
it
a
p
r
o
m
is
in
g
av
en
u
e
f
o
r
r
e
al
-
wo
r
ld
d
ep
lo
y
m
en
t.
B
ey
o
n
d
m
eth
o
d
o
lo
g
ical
ad
v
an
ce
m
e
n
ts
,
o
u
r
ap
p
r
o
ac
h
is
h
ig
h
ly
ad
ap
tab
le
f
o
r
in
teg
r
atio
n
in
to
m
en
tal
h
ea
lth
p
latf
o
r
m
s
an
d
s
o
cial
m
ed
ia
m
o
n
ito
r
in
g
to
o
ls
aim
ed
at
ea
r
ly
d
ep
r
ess
io
n
d
etec
tio
n
.
Po
ten
tial
im
p
lem
en
tatio
n
s
i
n
clu
d
e
API
-
b
ased
in
teg
r
atio
n
with
h
ea
lth
ap
p
lic
atio
n
s
lik
e
T
alk
s
p
ac
e
an
d
W
o
eb
o
t,
em
b
ed
d
in
g
with
in
elec
tr
o
n
ic
h
ea
lth
r
ec
o
r
d
s
y
s
tem
s
,
o
r
au
to
m
atin
g
p
r
elim
in
ar
y
s
cr
ee
n
in
g
s
in
clin
ical
wo
r
k
f
lo
ws.
Fu
tu
r
e
r
esear
ch
co
u
ld
ex
ten
d
th
is
wo
r
k
b
y
i
n
co
r
p
o
r
atin
g
m
u
ltimo
d
al
an
aly
s
is
—
s
u
ch
as
s
p
ee
ch
s
en
tim
en
t
d
etec
tio
n
an
d
f
ac
ial
ex
p
r
ess
io
n
r
ec
o
g
n
itio
n
—
to
im
p
r
o
v
e
d
iag
n
o
s
tic
ac
cu
r
ac
y
.
E
th
ical
c
o
n
s
id
er
atio
n
s
,
in
clu
d
i
n
g
d
ata
p
r
iv
ac
y
,
in
f
o
r
m
ed
co
n
s
en
t,
an
d
alg
o
r
ith
m
ic
tr
an
s
p
ar
en
cy
,
r
em
ai
n
ce
n
tr
al
to
e
n
s
u
r
in
g
co
m
p
lian
ce
with
r
eg
u
lato
r
y
f
r
a
m
ewo
r
k
s
an
d
m
in
im
izin
g
b
ias
in
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
.
T
o
s
y
s
tem
atica
lly
co
m
p
ar
e
ex
is
tin
g
a
p
p
r
o
ac
h
es
with
o
u
r
m
eth
o
d
o
l
o
g
y
,
T
ab
le
7
s
u
m
m
ar
izes
k
ey
s
tu
d
ies
in
d
e
p
r
ess
io
n
d
etec
tio
n
f
r
o
m
o
n
lin
e
co
n
ten
t,
d
etailin
g
th
eir
d
atasets
,
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es,
an
d
p
e
r
f
o
r
m
an
c
e
m
etr
ics.
T
h
is
s
y
n
th
esis
h
ig
h
lig
h
ts
b
o
th
th
e
p
r
o
g
r
ess
io
n
o
f
tech
n
ical
ap
p
r
o
ac
h
es
an
d
th
e
p
er
s
is
ten
t
g
ap
s
o
u
r
s
tu
d
y
a
d
d
r
ess
es,
p
ar
ticu
la
r
ly
th
r
o
u
g
h
g
en
etic
alg
o
r
ith
m
o
p
tim
izatio
n
an
d
B
E
R
T
f
in
e
-
tu
n
in
g
.
T
ab
le
7
.
R
elate
d
wo
r
k
co
m
p
a
r
is
o
n
S
t
u
d
y
D
a
t
a
s
e
t
s
o
u
r
c
e
S
a
mp
l
e
si
z
e
F
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
M
o
d
e
l
(
s)
u
se
d
B
e
st
p
e
r
f
o
r
ma
n
c
e
met
r
i
c
s
K
e
y
l
i
m
i
t
a
t
i
o
n
s
A
r
a
c
h
c
h
i
g
e
e
t
a
l
.
[
1
2
]
R
e
d
d
i
t
/
F
a
c
e
b
o
o
k
/
Tw
i
t
t
e
r
f
o
r
u
ms
2
9
st
u
d
i
e
s
S
e
ma
n
t
i
c
p
a
t
t
e
r
n
s
+
b
e
h
a
v
i
o
r
a
l
mark
e
r
s
V
a
r
i
o
u
s
M
L
a
l
g
o
r
i
t
h
ms
(
S
V
M
,
L
R
,
R
F
)
q
u
a
l
i
t
a
t
i
v
e
a
n
a
l
y
si
s
C
u
l
t
u
r
a
l
/
l
a
n
g
u
a
g
e
b
a
r
r
i
e
r
s,
s
e
l
f
-
d
i
a
g
n
o
si
s
e
r
r
o
r
s,
u
ser
a
n
o
n
y
m
i
t
y
G
l
a
z
e
t
a
l
.
[
1
3
]
M
u
l
t
i
-
p
l
a
t
f
o
r
m
f
o
r
u
ms,
me
d
i
c
a
l
D
B
s,
s
o
c
i
a
l
me
d
i
a
5
8
st
u
d
i
e
s
TF
-
I
D
F
,
L
I
W
C
,
n
-
g
r
a
m
s
V
a
r
i
o
u
s
M
L
a
l
g
o
r
i
t
h
ms
(
S
V
M
,
L
R
,
R
F
)
r
e
v
i
e
w
M
e
t
h
o
d
h
e
t
e
r
o
g
e
n
e
i
t
y
,
C
o
h
o
r
t
b
i
a
s,
Et
h
i
c
a
l
l
i
mi
t
s
Jai
n
e
t
a
l
.
[
1
4
]
R
e
d
d
i
t
6
0
,
0
0
0
r
e
c
o
r
d
s
B
a
g
-
of
-
W
o
r
d
s,
TF
-
I
D
F
S
V
M
,
N
B
,
L
R
,
RF
A
c
c
:
7
7
.
1
2
%,
F
1
:
7
7
%
Li
mi
t
e
d
p
l
a
t
f
o
r
m
d
i
v
e
r
s
i
t
y
S
a
i
f
u
l
l
a
h
e
t
a
l
.
[
1
5
]
Y
o
u
T
u
b
e
4
,
8
6
2
c
o
mm
e
n
t
s
C
o
u
n
t
-
V
e
c
t
o
r
i
z
a
t
i
o
n
,
TF
-
I
D
F
R
F
,
L
R
,
D
T,
S
V
M
,
K
N
N
,
NB
A
c
c
:
8
4
.
9
9
%
N
o
d
e
e
p
l
e
a
r
n
i
n
g
,
l
i
m
i
t
e
d
p
l
a
t
f
o
r
m
d
i
v
e
r
s
i
t
y
K
o
u
r
e
t
a
l
.
[
1
6
]
Tw
i
t
t
e
r
2
,
5
5
8
,
5
3
0
4
r
e
c
o
r
d
s
C
N
N
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
+
B
i
LST
M
seq
u
e
n
c
i
n
g
C
N
N
-
B
i
LS
TM
A
c
c
:
9
4
.
2
8
%
S
e
q
-
l
e
n
g
t
h
i
ss
u
e
s
,
P
l
a
t
f
o
r
m
b
i
a
s,
M
i
ss
e
d
c
o
m
o
r
b
i
d
i
t
i
e
s
C
h
e
r
e
d
d
y
e
t
a
l
.
[
1
7
]
Tw
i
t
t
e
r
3
,
0
0
0
p
o
s
t
s
TF
-
I
D
F
S
V
M
,
N
B
A
c
c
:
8
5
%
Li
mi
t
e
d
p
l
a
t
f
o
r
m
d
i
v
e
r
s
i
t
y
C
h
e
n
e
t
a
l
.
[
1
8
]
M
e
d
i
c
a
l
r
e
c
o
r
d
s
(
M
a
n
d
a
r
i
n
)
2
2
,
3
5
5
r
e
c
o
r
d
s
B
ER
T
e
mb
e
d
d
i
n
g
s
B
ER
T
-
C
N
N
A
U
C
:
9
3
%
M
i
l
i
t
a
r
y
/
c
i
v
i
l
i
a
n
p
e
r
f
o
r
m
a
n
c
e
g
a
p
s
S
u
b
r
a
ma
n
i
a
n
e
t
a
l
.
[
1
9
]
M
u
l
t
i
-
S
o
c
i
a
l
med
i
a
p
l
a
t
f
o
r
ms
-
F
a
st
T
e
x
t
,
n
-
g
r
a
ms
S
V
M
,
B
ER
T,
C
N
N
A
c
c
:
8
0
%
N
o
se
v
e
r
i
t
y
g
r
a
d
e
s,
La
b
e
l
b
i
a
s
O
u
r
S
t
u
d
y
R
e
d
d
i
t
7
,
7
3
1
p
o
s
t
s
B
ER
T,
W
o
r
d
2
V
e
c
,
TF
-
I
D
F
S
V
M
(
G
A
-
o
p
t
i
m
i
z
e
d
)
+
B
ER
T
A
c
c
:
9
5
.
9
9
%
P
l
a
t
f
o
r
m
b
i
a
s,
mi
ssi
n
g
d
e
m
o
g
r
a
p
h
i
c
a
n
a
l
y
si
s
,
sar
c
a
sm
/
sel
f
-
d
e
p
r
e
c
a
t
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
4
7
4
-
1
4
8
4
1482
4.
L
I
M
I
T
AT
I
O
NS
W
h
ile
o
u
r
s
tu
d
y
ad
v
a
n
ce
s
d
e
p
r
ess
io
n
d
etec
tio
n
in
s
o
cial
m
ed
ia
p
o
s
ts
,
s
ev
er
al
lim
itatio
n
s
m
u
s
t
b
e
ac
k
n
o
wled
g
e
d
.
T
h
e
d
ataset,
c
o
n
s
is
tin
g
s
o
lely
o
f
R
ed
d
it
p
o
s
ts
,
m
ay
lim
it
g
en
er
aliza
b
ilit
y
to
o
th
er
p
latf
o
r
m
s
,
an
d
th
e
ab
s
en
ce
o
f
d
em
o
g
r
ap
h
ic
attr
ib
u
tes
lik
e
ag
e
an
d
g
en
d
er
r
estricts
b
ias
an
aly
s
is
.
Ad
d
itio
n
ally
,
lin
g
u
is
tic
am
b
ig
u
ity
r
em
ain
s
a
ch
allen
g
e,
as
s
ar
ca
s
tic
o
r
s
elf
-
d
ep
r
ec
atin
g
s
tatem
en
ts
m
ay
b
e
m
is
cl
ass
if
ied
,
lead
in
g
to
f
alse
p
o
s
itiv
es,
wh
ile
f
al
s
e
n
eg
ativ
es
co
u
ld
o
v
er
lo
o
k
in
d
i
v
id
u
als
in
n
ee
d
.
T
o
ad
d
r
ess
t
h
ese
r
is
k
s
,
r
ef
in
in
g
class
if
icatio
n
th
r
esh
o
ld
s
,
in
co
r
p
o
r
atin
g
h
u
m
an
o
v
er
s
ig
h
t,
a
n
d
p
o
s
itio
n
in
g
th
ese
m
o
d
els
as
s
u
p
p
o
r
tiv
e
to
o
ls
r
ath
er
th
an
s
tan
d
alo
n
e
d
iag
n
o
s
tic
s
o
lu
tio
n
s
ar
e
cr
u
cial
s
tep
s
.
Fro
m
a
c
o
m
p
u
tatio
n
al
p
er
s
p
ec
tiv
e,
tr
an
s
f
o
r
m
e
r
-
b
ased
m
o
d
els
ty
p
ically
r
eq
u
i
r
e
s
ig
n
if
ican
t
r
eso
u
r
ce
s
.
Ho
wev
er
,
o
u
r
ex
p
er
im
en
ts
o
n
Go
o
g
le
C
o
l
ab
with
a
16
GB
GPU
d
em
o
n
s
tr
ated
th
a
t
f
in
e
-
tu
n
i
n
g
B
E
R
T
o
n
th
is
d
at
aset
was
f
ea
s
ib
le
with
in
a
r
ea
s
o
n
ab
le
tim
e
f
r
am
e,
s
u
g
g
esti
n
g
th
at
clo
u
d
-
b
ased
d
ep
lo
y
m
en
t
is
a
v
iab
le
o
p
tio
n
.
Fu
tu
r
e
s
tu
d
ies
s
h
o
u
l
d
f
u
r
th
er
ex
p
lo
r
e
ef
f
icien
c
y
tr
ad
e
-
o
f
f
s
,
p
ar
ticu
lar
ly
wh
e
n
s
ca
lin
g
to
lar
g
er
d
atasets
o
r
r
ea
l
-
tim
e
a
p
p
licatio
n
s
.
E
n
h
an
ci
n
g
m
o
d
el
in
ter
p
r
etab
ilit
y
is
an
o
th
er
i
m
p
o
r
tan
t
d
ir
ec
tio
n
.
T
ec
h
n
iq
u
es
s
u
ch
as
SHA
P
an
d
L
I
ME
co
u
ld
p
r
o
v
id
e
tr
an
s
p
ar
en
t
e
x
p
lan
atio
n
s
f
o
r
p
r
ed
ictio
n
s
,
wh
ich
is
esp
ec
ially
r
elev
an
t
as
f
u
tu
r
e
r
esear
c
h
in
teg
r
ates
ad
d
itio
n
al
u
s
er
attr
ib
u
tes
to
im
p
r
o
v
e
cla
s
s
if
icatio
n
ac
cu
r
ac
y
.
M
o
r
eo
v
er
,
ex
ten
d
in
g
th
is
f
r
am
ewo
r
k
b
ey
o
n
d
d
ep
r
ess
io
n
d
etec
tio
n
to
o
th
er
m
en
tal
h
ea
lth
co
n
d
itio
n
s
,
s
u
ch
as
a
n
x
iety
o
r
b
i
p
o
lar
d
is
o
r
d
er
,
co
u
ld
b
r
o
ad
e
n
its
ap
p
licab
ilit
y
.
Fin
ally
,
wh
ile
we
o
p
tim
ized
B
E
R
T
u
s
in
g
a
g
en
etic
alg
o
r
ith
m
,
f
u
tu
r
e
r
esear
c
h
co
u
ld
in
v
esti
g
ate
alter
n
ativ
e
tr
an
s
f
o
r
m
er
ar
c
h
itectu
r
es
s
u
ch
as
R
o
B
E
R
T
a
o
r
GPT,
as
we
ll
as
co
m
p
ar
e
GA
with
B
ay
esian
o
p
tim
izatio
n
o
r
g
r
id
s
ea
r
ch
f
o
r
h
y
p
er
p
a
r
am
eter
tu
n
in
g
.
Ad
d
itio
n
ally
,
v
alid
atin
g
th
e
m
o
d
el
o
n
clin
ically
an
n
o
tated
d
atasets
an
d
in
co
r
p
o
r
atin
g
m
u
ltimo
d
al
d
ata
—
s
u
ch
as
s
p
ee
ch
o
r
b
e
h
av
io
r
al
cu
es
—
co
u
ld
f
u
r
th
e
r
s
tr
en
g
th
en
its
ef
f
ec
tiv
en
ess
in
r
ea
l
-
wo
r
ld
m
e
n
tal
h
ea
lth
ass
ess
m
en
t.
5.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
d
em
o
n
s
tr
ates
th
e
ef
f
ec
tiv
en
ess
o
f
m
ac
h
in
e
lear
n
in
g
tec
h
n
iq
u
es,
p
ar
ticu
lar
l
y
SVM
with
f
in
e
-
tu
n
e
d
B
E
R
T
,
in
d
etec
tin
g
d
ep
r
ess
iv
e
co
n
ten
t
in
o
n
li
n
e
co
m
m
en
ts
.
B
y
in
c
o
r
p
o
r
ati
n
g
h
y
p
er
p
ar
am
eter
o
p
tim
izatio
n
th
r
o
u
g
h
g
en
etic
alg
o
r
ith
m
s
,
we
ac
h
iev
ed
im
p
r
ess
iv
e
r
esu
lts
:
9
5
.
9
9
%
ac
cu
r
a
cy
,
9
5
.
1
6
%
r
ec
all,
an
d
9
5
.
9
1
%
F1
-
s
co
r
e.
T
h
is
a
d
v
an
ce
d
ap
p
r
o
ac
h
s
ig
n
if
ica
n
tly
en
h
an
ce
s
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
co
m
p
ar
e
d
to
p
r
ev
io
u
s
r
esear
ch
,
h
ig
h
lig
h
tin
g
h
o
w
s
o
p
h
is
ticated
m
o
d
el
tu
n
in
g
an
d
B
E
R
T
’
s
co
n
tex
tu
al
u
n
d
er
s
tan
d
i
n
g
ca
n
ad
v
an
ce
to
o
ls
f
o
r
d
etec
tin
g
d
ep
r
ess
iv
e
d
is
o
r
d
er
s
.
T
h
e
i
m
p
licatio
n
s
o
f
o
u
r
r
esear
ch
ex
ten
d
b
e
y
o
n
d
th
e
im
m
ed
iate
f
in
d
in
g
s
.
I
n
te
g
r
ati
n
g
B
E
R
T
f
in
e
-
tu
n
i
n
g
with
g
e
n
etic
alg
o
r
ith
m
o
p
tim
izatio
n
estab
lis
h
es
a
r
o
b
u
s
t
f
r
am
ewo
r
k
f
o
r
d
e
v
elo
p
in
g
m
o
r
e
ac
cu
r
ate
an
d
r
eliab
le
p
r
e
d
ictiv
e
to
o
ls
.
T
h
is
a
p
p
r
o
ac
h
h
as
th
e
p
o
te
n
tial
to
r
ev
o
lu
tio
n
ize
t
h
e
d
iag
n
o
s
is
an
d
m
an
a
g
em
en
t
o
f
d
ep
r
ess
iv
e
d
is
o
r
d
er
s
.
Fu
tu
r
e
r
esear
c
h
s
h
o
u
ld
v
alid
ate
th
ese
r
esu
lts
ac
r
o
s
s
d
iv
er
s
e
d
atasets
an
d
clin
ical
s
ettin
g
s
to
en
s
u
r
e
b
r
o
ad
er
ap
p
licab
ilit
y
an
d
r
o
b
u
s
tn
ess
.
Ad
d
itio
n
ally
,
in
teg
r
atin
g
th
es
e
m
eth
o
d
s
with
elec
tr
o
n
ic
h
ea
lth
r
ec
o
r
d
s
an
d
r
ea
l
-
tim
e
p
atien
t
m
o
n
ito
r
i
n
g
s
y
s
tem
s
co
u
ld
f
u
r
t
h
er
e
n
h
an
ce
th
eir
u
tili
ty
.
Ultim
ately
,
o
u
r
s
tu
d
y
a
d
v
an
ce
s
p
r
ed
ictiv
e
m
ed
icin
e
b
y
p
r
o
v
id
in
g
a
p
r
ac
tical
ap
p
r
o
ac
h
to
d
etec
ti
n
g
d
ep
r
ess
iv
e
co
n
ten
t,
p
av
in
g
th
e
way
f
o
r
p
e
r
s
o
n
alize
d
tr
ea
t
m
en
t
s
tr
ateg
ies
an
d
im
p
r
o
v
e
d
p
atien
t m
a
n
ag
em
e
n
t,
with
s
ig
n
if
ican
t im
p
licatio
n
s
f
o
r
b
o
th
r
esear
ch
an
d
clin
ical
p
r
ac
tice.
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
.
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
Ab
d
Allah
Ao
u
r
a
g
h
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Mo
h
am
ed
B
ah
aj
✓
✓
✓
✓
✓
✓
✓
Fo
u
ad
T
o
u
f
ik
✓
✓
✓
✓
✓
✓
✓
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
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
tin
g
B
E
R
T fin
e
-
tu
n
in
g
a
n
d
g
e
n
etic
a
lg
o
r
ith
m
fo
r
…
(
A
b
d
A
lla
h
A
o
u
r
a
g
h
)
1483
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
ata
t
h
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
a
r
e
av
aila
b
le
f
r
o
m
th
e
c
o
r
r
esp
o
n
d
in
g
a
u
th
o
r
,
AA,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
M
.
M
a
j
e
t
a
l
.
,
“
Th
e
c
l
i
n
i
c
a
l
c
h
a
r
a
c
t
e
r
i
z
a
t
i
o
n
o
f
t
h
e
a
d
u
l
t
p
a
t
i
e
n
t
w
i
t
h
d
e
p
r
e
s
si
o
n
a
i
me
d
a
t
p
e
r
so
n
a
l
i
z
a
t
i
o
n
o
f
ma
n
a
g
e
m
e
n
t
,
”
Wo
r
l
d
Psy
c
h
i
a
t
r
y
,
v
o
l
.
1
9
,
n
o
.
3
,
p
p
.
2
6
9
–
2
9
3
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
2
/
w
p
s.
2
0
7
7
1
.
[
2
]
W
H
O
,
“
D
e
p
r
e
ssi
v
e
d
i
so
r
d
e
r
(
d
e
p
r
e
ss
i
o
n
)
,
”
T
h
e
EC
PH
E
n
c
y
c
l
o
p
e
d
i
a
o
f
P
syc
h
o
l
o
g
y
.
2
0
2
5
,
A
c
c
e
ss
e
d
:
A
u
g
.
2
6
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
w
w
w
.
w
h
o
.
i
n
t
/
n
e
w
s
-
r
o
o
m
/
f
a
c
t
-
sh
e
e
t
s/
d
e
t
a
i
l
/
d
e
p
r
e
ssi
o
n
.
[
3
]
L.
C
u
i
e
t
a
l
.
,
“
M
a
j
o
r
d
e
p
r
e
ss
i
v
e
d
i
s
o
r
d
e
r
:
h
y
p
o
t
h
e
s
i
s,
m
e
c
h
a
n
i
sm
,
p
r
e
v
e
n
t
i
o
n
a
n
d
t
r
e
a
t
m
e
n
t
,
”
S
i
g
n
a
l
T
ra
n
s
d
u
c
t
i
o
n
a
n
d
T
a
rg
e
t
e
d
T
h
e
r
a
p
y
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
1
–
3
2
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
3
9
2
-
0
2
4
-
0
1
7
3
8
-
y.
[
4
]
P
.
C
u
i
j
p
e
r
s,
A
.
S
t
r
i
n
g
a
r
i
s,
a
n
d
M
.
W
o
l
p
e
r
t
,
“
Tr
e
a
t
m
e
n
t
o
u
t
c
o
m
e
s
f
o
r
d
e
p
r
e
ss
i
o
n
:
c
h
a
l
l
e
n
g
e
s
a
n
d
o
p
p
o
r
t
u
n
i
t
i
e
s,
”
T
h
e
L
a
n
c
e
t
Psy
c
h
i
a
t
r
y
,
v
o
l
.
7
,
n
o
.
1
1
,
p
p
.
9
2
5
–
9
2
7
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
S
2
2
1
5
-
0
3
6
6
(
2
0
)
3
0
0
3
6
-
5.
[
5
]
K
.
A
.
M
c
L
a
u
g
h
l
i
n
,
“
Th
e
p
u
b
l
i
c
h
e
a
l
t
h
i
m
p
a
c
t
o
f
m
a
j
o
r
d
e
p
r
e
ssi
o
n
:
a
c
a
l
l
f
o
r
i
n
t
e
r
d
i
s
c
i
p
l
i
n
a
r
y
p
r
e
v
e
n
t
i
o
n
e
f
f
o
r
t
s,
”
Pr
e
v
e
n
t
i
o
n
S
c
i
e
n
c
e
,
v
o
l
.
1
2
,
n
o
.
4
,
p
p
.
3
6
1
–
3
7
1
,
2
0
1
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
1
2
1
-
0
1
1
-
0
2
3
1
-
8.
[
6
]
T.
Z
h
a
n
g
,
A
.
M
.
S
c
h
o
e
n
e
,
S
.
Ji
,
a
n
d
S
.
A
n
a
n
i
a
d
o
u
,
“
N
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
a
p
p
l
i
e
d
t
o
me
n
t
a
l
i
l
l
n
e
ss
d
e
t
e
c
t
i
o
n
:
a
n
a
r
r
a
t
i
v
e
r
e
v
i
e
w
,
”
n
p
j
D
i
g
i
t
a
l
Me
d
i
c
i
n
e
,
v
o
l
.
5
,
n
o
.
1
,
p
p
.
1
–
1
3
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
7
4
6
-
0
2
2
-
0
0
5
8
9
-
7.
[
7
]
B
.
G
.
T
e
f
e
r
r
a
e
t
a
l
.
,
“
S
c
r
e
e
n
i
n
g
f
o
r
d
e
p
r
e
ss
i
o
n
u
si
n
g
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ss
i
n
g
:
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
I
n
t
e
r
a
c
t
i
v
e
J
o
u
rn
a
l
o
f
Me
d
i
c
a
l
R
e
se
a
rc
h
,
v
o
l
.
1
3
,
p
.
e
5
5
0
6
7
,
2
0
2
4
,
d
o
i
:
1
0
.
2
1
9
6
/
5
5
0
6
7
.
[
8
]
D
.
D
.
D
e
S
o
u
z
a
,
J
.
R
o
b
i
n
,
M
.
G
u
m
u
s,
a
n
d
A
.
Y
e
u
n
g
,
“
N
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ss
i
n
g
a
s
a
n
e
m
e
r
g
i
n
g
t
o
o
l
t
o
d
e
t
e
c
t
l
a
t
e
-
l
i
f
e
d
e
p
r
e
ss
i
o
n
,
”
Fr
o
n
t
i
e
rs
i
n
Ps
y
c
h
i
a
t
r
y
,
v
o
l
.
1
2
,
p
.
7
1
9
1
2
5
,
2
0
2
1
,
d
o
i
:
1
0
.
3
3
8
9
/
f
p
s
y
t
.
2
0
2
1
.
7
1
9
1
2
5
.
[
9
]
P
.
S
h
w
e
t
a
a
n
d
K
.
A
d
h
i
y
a
,
“
C
o
m
p
a
r
a
t
i
v
e
st
u
d
y
o
f
f
e
a
t
u
r
e
e
n
g
i
n
e
e
r
i
n
g
f
o
r
a
u
t
o
ma
t
e
d
s
h
o
r
t
a
n
sw
e
r
g
r
a
d
i
n
g
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
-
2
0
2
2
I
EEE
Wo
r
l
d
C
o
n
f
e
re
n
c
e
o
n
Ap
p
l
i
e
d
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
C
o
m
p
u
t
i
n
g
,
AI
C
2
0
2
2
,
2
0
2
2
,
p
p
.
5
9
4
–
5
9
7
,
d
o
i
:
1
0
.
1
1
0
9
/
A
I
C
5
5
0
3
6
.
2
0
2
2
.
9
8
4
8
8
5
1
.
[
1
0
]
D
.
L.
S
h
a
n
t
h
i
a
n
d
N
.
C
h
e
t
h
a
n
,
“
G
e
n
e
t
i
c
a
l
g
o
r
i
t
h
m
b
a
se
d
h
y
p
e
r
-
p
a
r
a
me
t
e
r
t
u
n
i
n
g
t
o
i
mp
r
o
v
e
t
h
e
p
e
r
f
o
r
ma
n
c
e
o
f
mac
h
i
n
e
l
e
a
r
n
i
n
g
mo
d
e
l
s
,
”
S
N
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
4
,
n
o
.
2
,
p
.
1
1
9
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s4
2
9
7
9
-
022
-
0
1
5
3
7
-
8.
[
1
1
]
R
.
J
a
i
n
,
R
.
S
.
R
a
i
,
S
.
J
a
i
n
,
R
.
A
h
l
u
w
a
l
i
a
,
a
n
d
J.
G
u
p
t
a
,
“
R
e
a
l
t
i
me
se
n
t
i
me
n
t
a
n
a
l
y
s
i
s o
f
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
u
s
i
n
g
m
u
l
t
i
m
e
d
i
a
i
n
p
u
t
,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
2
,
n
o
.
2
6
,
p
p
.
4
1
0
2
1
–
4
1
0
3
6
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
4
2
-
0
2
3
-
1
5
2
1
3
-
3.
[
1
2
]
I
.
A
.
N
.
A
r
a
c
h
c
h
i
g
e
,
P
.
S
a
n
d
a
n
a
p
i
t
c
h
a
i
,
a
n
d
R
.
W
e
e
r
a
s
i
n
g
h
e
,
“
I
n
v
e
s
t
i
g
a
t
i
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
&
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ss
i
n
g
t
e
c
h
n
i
q
u
e
s
a
p
p
l
i
e
d
f
o
r
p
r
e
d
i
c
t
i
n
g
d
e
p
r
e
ssi
o
n
d
i
so
r
d
e
r
f
r
o
m
o
n
l
i
n
e
su
p
p
o
r
t
f
o
r
u
ms
:
A
s
y
st
e
ma
t
i
c
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
I
n
f
o
rm
a
t
i
o
n
(
S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
1
2
,
n
o
.
1
1
,
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
i
n
f
o
1
2
1
1
0
4
4
4
.
[
1
3
]
A
.
Le
G
l
a
z
e
t
a
l
.
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
i
n
men
t
a
l
h
e
a
l
t
h
:
S
y
st
e
mat
i
c
r
e
v
i
e
w
,
”
J
o
u
r
n
a
l
o
f
Me
d
i
c
a
l
I
n
t
e
r
n
e
t
Re
se
a
r
c
h
,
v
o
l
.
2
3
,
n
o
.
5
,
2
0
2
1
,
d
o
i
:
1
0
.
2
1
9
6
/
1
5
7
0
8
.
[
1
4
]
P
.
Ja
i
n
,
K
.
R
.
S
r
i
n
i
v
a
s,
a
n
d
A
.
V
i
c
h
a
r
e
,
“
D
e
p
r
e
ss
i
o
n
a
n
d
s
u
i
c
i
d
e
a
n
a
l
y
si
s
u
si
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
N
LP
,
”
J
o
u
rn
a
l
o
f
Ph
y
si
c
s:
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s
,
v
o
l
.
2
1
6
1
,
n
o
.
1
,
p
.
1
2
0
3
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
2
1
6
1
/
1
/
0
1
2
0
3
4
.
[
1
5
]
S
.
S
a
i
f
u
l
l
a
h
,
Y
.
F
a
u
z
i
y
a
h
,
a
n
d
A
.
S
.
A
r
i
b
o
w
o
,
“
C
o
m
p
a
r
i
s
o
n
o
f
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
f
o
r
se
n
t
i
me
n
t
a
n
a
l
y
si
s
i
n
d
e
t
e
c
t
i
n
g
a
n
x
i
e
t
y
b
a
s
e
d
o
n
so
c
i
a
l
me
d
i
a
d
a
t
a
,
”
J
u
r
n
a
l
I
n
f
o
rm
a
t
i
k
a
,
v
o
l
.
1
5
,
n
o
.
1
,
p
.
4
5
,
2
0
2
1
,
d
o
i
:
1
0
.
2
6
5
5
5
/
j
i
f
o
.
v
1
5
i
1
.
a
2
0
1
1
1
.
[
1
6
]
H
.
K
o
u
r
a
n
d
M
.
K
.
G
u
p
t
a
,
“
A
n
h
y
b
r
i
d
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
f
o
r
d
e
p
r
e
ssi
o
n
p
r
e
d
i
c
t
i
o
n
f
r
o
m
u
ser
t
w
e
e
t
s
u
s
i
n
g
f
e
a
t
u
r
e
-
r
i
c
h
C
N
N
a
n
d
b
i
-
d
i
r
e
c
t
i
o
n
a
l
LS
TM
,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
1
,
n
o
.
1
7
,
p
p
.
2
3
6
4
9
–
2
3
6
8
5
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
0
4
2
-
022
-
1
2
6
4
8
-
y.
[
1
7
]
S
.
C
h
e
r
e
d
d
y
,
K
.
G
e
e
t
h
a
,
a
n
d
A
.
G
.
S
r
e
e
d
e
v
i
,
“
Tw
e
e
t
i
n
g
t
h
e
b
l
u
e
s:
l
e
v
e
r
a
g
i
n
g
N
LP
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
m
o
d
e
l
s
f
o
r
d
e
p
r
e
ss
i
o
n
d
e
t
e
c
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
2
n
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
d
v
a
n
c
e
m
e
n
t
i
n
C
o
m
p
u
t
a
t
i
o
n
a
n
d
C
o
m
p
u
t
e
r
T
e
c
h
n
o
l
o
g
i
e
s
,
I
n
C
A
C
C
T
2
0
2
4
,
2
0
2
4
,
p
p
.
8
7
5
–
8
8
0
,
d
o
i
:
1
0
.
1
1
0
9
/
I
n
C
A
C
C
T
6
1
5
9
8
.
2
0
2
4
.
1
0
5
5
0
9
6
1
.
[
1
8
]
T.
-
C
h
e
n
,
H
.
-
C
h
u
,
Y
.
-
T
a
i
,
a
n
d
S
.
-
Y
a
n
g
,
“
P
e
r
f
o
r
ma
n
c
e
s
o
f
d
e
p
r
e
ss
i
o
n
d
e
t
e
c
t
i
o
n
t
h
r
o
u
g
h
d
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ss
i
n
g
t
o
ma
n
d
a
r
i
n
C
h
i
n
e
se
m
e
d
i
c
a
l
r
e
c
o
r
d
s:
C
o
mp
a
r
i
s
o
n
b
e
t
w
e
e
n
c
i
v
i
l
i
a
n
a
n
d
mi
l
i
t
a
r
y
p
o
p
u
l
a
t
i
o
n
s
.
,
”
T
a
i
w
a
n
e
se
J
o
u
r
n
a
l
o
f
Psy
c
h
i
a
t
r
y
,
v
o
l
.
3
6
,
n
o
.
1
,
p
p
.
3
2
–
3
8
,
2
0
2
2
.
[
1
9
]
M
.
S
u
b
r
a
m
a
n
i
a
n
,
G
.
R
a
j
u
,
A
.
S
u
r
e
sh
k
u
m
a
r
,
C
.
A
n
b
a
r
a
s
u
,
K
.
S
.
V
a
d
i
v
e
l
,
a
n
d
P
.
S
.
N
a
n
d
h
i
n
i
,
“
F
r
o
m
w
o
r
d
s
t
o
e
mo
t
i
o
n
s:
i
d
e
n
t
i
f
y
i
n
g
d
e
p
r
e
ssi
o
n
t
h
r
o
u
g
h
so
c
i
a
l
me
d
i
a
i
n
si
g
h
t
s
,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
i
n
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
,
v
o
l
.
2
0
4
6
C
C
I
S
,
p
p
.
2
6
8
–
2
8
2
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
5
8
4
9
5
-
4
_
2
0
.
[
2
0
]
I
n
F
a
mo
u
s
C
o
d
e
r
,
“
D
e
p
r
e
ssi
o
n
:
R
e
d
d
i
t
d
a
t
a
se
t
(
c
l
e
a
n
e
d
)
”
K
a
g
g
l
e
,
2
0
2
4
,
A
c
c
e
sse
d
:
A
u
g
.
2
6
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
w
w
w
.
k
a
g
g
l
e
.
c
o
m
/
d
a
t
a
se
t
s/
i
n
f
a
mo
u
sc
o
d
e
r
/
d
e
p
r
e
ssi
o
n
-
r
e
d
d
i
t
-
c
l
e
a
n
e
d
.
[
2
1
]
M
.
S
i
i
n
o
,
I
.
Ti
n
n
i
r
e
l
l
o
,
a
n
d
M
.
L
a
C
a
sci
a
,
“
I
s
t
e
x
t
p
r
e
p
r
o
c
e
ss
i
n
g
st
i
l
l
w
o
r
t
h
t
h
e
t
i
m
e
?
A
c
o
m
p
a
r
a
t
i
v
e
s
u
r
v
e
y
o
n
t
h
e
i
n
f
l
u
e
n
c
e
o
f
p
o
p
u
l
a
r
p
r
e
p
r
o
c
e
ss
i
n
g
m
e
t
h
o
d
s
o
n
Tr
a
n
sf
o
r
mers
a
n
d
t
r
a
d
i
t
i
o
n
a
l
c
l
a
s
si
f
i
e
r
s,
”
I
n
f
o
rm
a
t
i
o
n
S
y
st
e
m
s
,
v
o
l
.
1
2
1
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
s.
2
0
2
3
.
1
0
2
3
4
2
.
[
2
2
]
J.
R
a
v
i
a
n
d
S
.
K
u
l
k
a
r
n
i
,
“
T
e
x
t
e
mb
e
d
d
i
n
g
t
e
c
h
n
i
q
u
e
s
f
o
r
e
f
f
i
c
i
e
n
t
c
l
u
st
e
r
i
n
g
o
f
t
w
i
t
t
e
r
d
a
t
a
,
”
Ev
o
l
u
t
i
o
n
a
r
y
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
6
,
n
o
.
5
,
p
p
.
1
6
6
7
–
1
6
7
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
2
0
6
5
-
023
-
0
0
8
2
5
-
3.
[
2
3
]
L.
J
i
a
x
i
n
,
“
T
e
x
t
v
e
c
t
o
r
i
z
a
t
i
o
n
i
n
s
e
n
t
i
m
e
n
t
a
n
a
l
y
si
s
:
A
c
o
mp
a
r
a
t
i
v
e
s
t
u
d
y
o
f
TF
-
I
D
F
a
n
d
W
o
r
d
2
V
e
c
f
r
o
m
A
maz
o
n
f
i
n
e
f
o
o
d
r
e
v
i
e
w
s,”
I
T
M
We
b
o
f
C
o
n
f
e
r
e
n
c
e
s
,
v
o
l
.
7
0
,
p
.
3
0
0
1
,
2
0
2
5
.
[
2
4
]
P
.
A
smi
a
n
d
M
.
S
.
S
a
n
a
j
,
“
T
o
x
i
c
s
p
e
e
c
h
c
l
a
s
si
f
i
c
a
t
i
o
n
v
i
a
d
e
e
p
l
e
a
r
n
i
n
g
u
si
n
g
c
o
mb
i
n
e
d
f
e
a
t
u
r
e
s
f
r
o
m
B
E
R
T
&
F
a
s
t
T
e
x
t
e
mb
e
d
d
i
n
g
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
En
g
i
n
e
e
ri
n
g
Re
s
e
a
r
c
h
&
T
e
c
h
n
o
l
o
g
y
(
I
J
ERT)
,
I
C
C
I
D
T
-
2
0
2
1
C
o
n
f
e
r
e
n
c
e
Pro
c
e
e
d
i
n
g
s
,
v
o
l
.
9
,
n
o
.
7
,
p
p
.
6
8
–
7
1
,
2
0
2
1
,
d
o
i
:
1
0
.
1
7
5
7
7
/
I
JER
TC
O
N
V
9
I
S
0
7
0
1
6
.
[
2
5
]
A
.
A
l
i
R
a
z
a
e
t
a
l
.
,
“
R
e
v
i
e
w
t
o
u
n
f
o
l
d
t
h
e
r
o
l
e
o
f
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
i
n
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
,
”
J
o
u
r
n
a
l
o
f
Po
l
i
c
y
Re
se
a
rc
h
,
v
o
l
.
9
,
n
o
.
4
,
p
p
.
1
5
2
–
1
6
2
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
6
1
5
0
6
/
0
2
.
0
0
1
3
6
.
[
2
6
]
S
.
U
d
d
i
n
,
A
.
K
h
a
n
,
M
.
E
.
H
o
ss
a
i
n
,
a
n
d
M
.
A
.
M
o
n
i
,
“
C
o
m
p
a
r
i
n
g
d
i
f
f
e
r
e
n
t
s
u
p
e
r
v
i
se
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
f
o
r
d
i
s
e
a
se
p
r
e
d
i
c
t
i
o
n
,
”
B
MC
Me
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s a
n
d
D
e
c
i
si
o
n
Ma
k
i
n
g
,
v
o
l
.
1
9
,
n
o
.
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
8
6
/
s
1
2
9
1
1
-
0
1
9
-
1
0
0
4
-
8.
[
2
7
]
T.
S
o
o
d
,
P
.
G
a
r
g
,
H
.
P
a
n
n
u
,
a
n
d
S
.
D
u
l
l
a
t
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
a
n
d
N
LP
f
o
r
a
n
x
i
e
t
y
a
n
d
d
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
u
s
i
n
g
Tw
i
t
t
e
r
d
a
t
a
,
”
i
n
2
0
2
5
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
g
n
i
t
i
v
e
C
o
m
p
u
t
i
n
g
i
n
E
n
g
i
n
e
e
ri
n
g
,
C
o
m
m
u
n
i
c
a
t
i
o
n
s,
S
c
i
e
n
c
e
s
a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.