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.
14
,
No
.
6
,
Dec
em
b
er
20
24
,
p
p
.
6
9
6
9
~
6
9
8
0
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
14
i
6
.
pp
6
9
6
9
-
6
9
8
0
6969
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Co
mpa
ra
tive a
na
ly
sis
of deep
S
ia
mes
e
mo
dels for
me
dica
l
repo
rts tex
t
sim
il
a
rity
Dia
n K
urnia
s
a
ri
1,
2
,
M
u
s
t
o
f
a
Usm
a
n
2
,
Wa
rso
no
2
,
F
a
v
o
rise
n Ro
s
y
k
ing
L
um
ba
nra
j
a
3
1
D
o
c
t
o
r
a
l
P
r
o
g
r
a
m
,
F
a
c
u
l
t
y
o
f
M
a
t
h
e
mat
i
c
s
a
n
d
N
a
t
u
r
a
l
S
c
i
e
n
c
e
s,
U
n
i
v
e
r
s
i
t
a
s
La
m
p
u
n
g
,
La
m
p
u
n
g
,
I
n
d
o
n
e
s
i
a
2
D
e
p
a
r
t
me
n
t
o
f
M
a
t
h
e
m
a
t
i
c
s
,
F
a
c
u
l
t
y
o
f
M
a
t
h
e
mat
i
c
s
a
n
d
N
a
t
u
r
a
l
S
c
i
e
n
c
e
s,
U
n
i
v
e
r
s
i
t
a
s L
a
mp
u
n
g
,
L
a
m
p
u
n
g
,
I
n
d
o
n
e
si
a
3
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
M
a
t
h
e
ma
t
i
c
s
a
n
d
N
a
t
u
r
a
l
S
c
i
e
n
c
e
s,
U
n
i
v
e
r
si
t
a
s
La
m
p
u
n
g
,
La
m
p
u
n
g
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
2
7
,
2
0
2
4
R
ev
is
ed
J
u
l 2
5
,
2
0
2
4
Acc
ep
ted
Au
g
6
,
2
0
2
4
Ev
e
n
t
h
o
u
g
h
m
e
d
ica
l
re
p
o
r
ts h
a
v
e
b
e
e
n
d
i
g
it
ize
d
,
th
e
y
a
re
g
e
n
e
ra
ll
y
tex
t
d
a
ta
a
n
d
h
a
v
e
n
o
t
b
e
e
n
u
se
d
o
p
ti
m
a
ll
y
.
Ex
trac
ti
n
g
in
f
o
rm
a
ti
o
n
fro
m
th
e
se
re
p
o
rts
is
c
h
a
ll
e
n
g
in
g
d
u
e
to
th
e
ir
h
ig
h
v
o
l
u
m
e
a
n
d
u
n
str
u
c
tu
re
d
n
a
tu
re
.
An
a
ly
z
in
g
th
e
e
x
trac
ti
o
n
o
f
re
lev
a
n
t
a
n
d
h
i
g
h
-
q
u
a
li
t
y
i
n
fo
rm
a
ti
o
n
c
a
n
b
e
a
c
h
iev
e
d
b
y
m
e
a
su
rin
g
se
m
a
n
ti
c
tex
tu
a
l
sim
il
a
rit
y
(S
TS
)
.
Co
n
se
q
u
e
n
tl
y
,
th
e
p
r
ima
ry
a
im
o
f
th
is
st
u
d
y
is
to
d
e
v
e
lo
p
a
n
d
e
v
a
lu
a
te
th
e
p
e
rfo
rm
a
n
c
e
o
f
f
o
u
r
m
o
d
e
ls:
S
iam
e
se
M
a
n
h
a
tt
a
n
c
o
n
v
o
l
u
ti
o
n
n
e
u
ra
l
n
e
two
rk
(
CNN
)
,
S
iam
e
se
M
a
n
h
a
tt
a
n
lo
n
g
sh
o
rt
-
term
m
e
m
o
ry
(
LS
TM
)
,
S
iam
e
se
M
a
n
h
a
tt
a
n
h
y
b
rid
CN
N
-
LS
TM
,
a
n
d
S
iam
e
se
M
a
n
h
a
tt
a
n
h
y
b
rid
LS
TM
-
CNN
,
i
n
d
e
term
in
i
n
g
S
T
S
b
e
twe
e
n
se
n
ten
c
e
p
a
irs
in
m
e
d
ica
l
re
p
o
rts
.
P
e
rfo
rm
a
n
c
e
c
o
m
p
a
riso
n
s
we
re
c
o
n
d
u
c
ted
u
sin
g
C
o
sin
e
S
imilarit
y
a
n
d
wo
r
d
m
o
v
e
r'
s
d
istan
c
e
(W
M
D)
m
e
th
o
d
s
.
Th
e
re
su
lt
s
in
d
ica
te
th
a
t
t
h
e
S
iam
e
se
M
a
n
h
a
tt
a
n
h
y
b
ri
d
LS
T
M
-
C
NN
m
o
d
e
l
o
u
t
p
e
rfo
rm
s
th
e
o
th
e
r
m
o
d
e
ls,
wi
th
a
sim
il
a
rit
y
sc
o
re
o
f
1
f
o
r
e
a
c
h
se
n
ten
c
e
p
a
ir,
sig
n
if
y
in
g
i
d
e
n
ti
c
a
l
se
m
a
n
ti
c
m
e
a
n
in
g
.
K
ey
w
o
r
d
s
:
B
io
m
ed
ical
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
B
io
W
o
r
d
Vec
Hy
b
r
id
L
STM
-
C
NN
Me
d
ical
r
ep
o
r
t
Sem
an
tic
tex
t similar
ity
Siam
ese
Ma
n
h
attan
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
:
Mu
s
to
f
a
Usma
n
Dep
ar
tm
en
t o
f
Ma
th
em
atics,
Facu
lty
o
f
Ma
th
e
m
atics a
n
d
Na
tu
r
al
Scien
ce
s
,
Un
iv
er
s
itas
L
am
p
u
n
g
Pro
f
.
Su
m
an
tr
i Bro
j
o
n
eg
o
r
o
s
t
r
ee
t
No
.
1
,
Ged
o
n
g
Me
n
en
g
,
B
an
d
ar
L
am
p
u
n
g
,
L
am
p
u
n
g
,
I
n
d
o
n
esia
E
m
ail:
u
s
m
an
_
alf
h
a@
y
ah
o
o
.
c
o
m
1.
I
NT
RO
D
UCT
I
O
N
Ov
er
th
e
p
ast
d
ec
ad
e,
th
er
e
h
as
b
ee
n
ab
u
n
d
an
t
tex
tu
al
d
ata
a
cr
o
s
s
v
ar
io
u
s
f
ield
s
,
in
clu
d
in
g
b
io
m
ed
ical
r
esear
ch
[
1
]
.
Ho
wev
er
,
th
ese
v
alu
ab
le
r
eso
u
r
ce
s
h
av
e
o
f
ten
g
o
n
e
u
n
d
e
r
u
tili
ze
d
.
Fo
r
in
s
ta
n
ce
,
m
u
ch
m
ed
ical
r
ep
o
r
t
te
x
t
d
ata
r
e
m
ain
s
ar
c
h
iv
ed
a
n
d
u
n
ex
p
l
o
r
ed
,
ev
en
af
ter
d
ig
itizatio
n
.
T
h
is
d
ata'
s
s
h
ee
r
v
o
l
u
m
e
an
d
u
n
s
tr
u
ctu
r
ed
n
atu
r
e
m
ak
e
e
x
tr
ac
tin
g
r
elev
an
t,
h
ig
h
-
q
u
ality
i
n
f
o
r
m
atio
n
d
if
f
icu
lt.
On
e
p
r
ac
tical
ap
p
r
o
ac
h
t
o
ex
tr
ac
tin
g
r
elev
an
t,
h
ig
h
-
q
u
ality
in
f
o
r
m
atio
n
f
r
o
m
m
ed
ical
r
e
p
o
r
t
tex
ts
in
v
o
lv
es
m
ea
s
u
r
in
g
s
em
an
tic
tex
tu
al
s
im
ilar
ity
(
STS)
.
STS
is
a
f
o
u
n
d
atio
n
al
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P)
task
th
at
as
s
ess
e
s
h
o
w
clo
s
ely
two
s
en
ten
ce
s
co
n
v
ey
th
e
s
am
e
m
ea
n
in
g
[
2
]
.
I
n
b
io
m
e
d
ical
NL
P
(
B
io
NL
P),
ST
S
p
lay
s
a
p
iv
o
tal
r
o
le
in
e
n
s
u
r
in
g
th
e
ac
c
u
r
ate
in
ter
p
r
etatio
n
an
d
r
etr
iev
al
o
f
in
f
o
r
m
atio
n
f
r
o
m
b
io
m
e
d
ical
d
o
cu
m
e
n
ts
[
3
]
.
T
h
at
is
d
u
e
to
i
ts
n
u
m
er
o
u
s
d
ir
ec
t
ap
p
licatio
n
s
in
in
f
o
r
m
atio
n
e
x
tr
ac
tio
n
,
s
u
ch
as
in
b
io
m
ed
ical
s
en
ten
ce
s
ea
r
ch
an
d
class
if
icatio
n
,
as
well
as
in
d
ir
ec
t
ap
p
li
ca
tio
n
s
lik
e
b
io
m
ed
ical
q
u
esti
o
n
an
s
wer
in
g
a
n
d
d
o
cu
m
e
n
t
lab
ellin
g
[
4
]
.
T
h
e
s
ig
n
if
ican
ce
o
f
STS
in
th
ese
ap
p
licatio
n
s
is
h
eig
h
ten
ed
b
y
th
e
f
ac
t
th
at
m
an
y
b
io
m
ed
ical
ter
m
s
ca
n
h
av
e
d
if
f
er
en
t m
ea
n
i
n
g
s
d
ep
e
n
d
in
g
o
n
th
e
tex
t'
s
co
n
tex
t
[
5
]
.
R
esear
ch
er
s
ty
p
ically
em
p
lo
y
th
r
ee
m
ain
ap
p
r
o
ac
h
es
to
ca
lcu
late
STS:
C
o
r
p
u
s
-
b
ased
,
k
n
o
wled
g
e
-
b
ased
,
an
d
s
tr
in
g
-
b
ased
m
eth
o
d
s
.
R
esear
ch
b
y
Su
n
ilk
u
m
a
r
an
d
Sh
aji
[
6
]
in
d
icate
s
th
at
th
e
C
o
r
p
u
s
-
b
ased
ap
p
r
o
ac
h
,
i
n
p
ar
ticu
lar
,
is
wid
ely
ad
o
p
ted
an
d
s
h
o
ws
p
r
o
m
is
in
g
r
esu
lts
.
T
h
e
C
o
r
p
u
s
-
b
ased
m
eth
o
d
c
o
m
p
r
is
es
two
s
tatis
t
ically
d
is
tin
ct
ap
p
r
o
ac
h
es.
T
h
e
f
ir
s
t
ap
p
r
o
ac
h
u
tili
ze
s
tr
ad
itio
n
al
s
tatis
tica
l
an
aly
s
is
tech
n
iq
u
es
lik
e
laten
t
s
em
an
tic
an
aly
s
is
(
L
S
A)
,
wh
ich
co
u
n
ts
wo
r
d
f
r
eq
u
en
cies
in
th
e
tex
t
b
u
t
d
o
es
n
o
t
d
elv
e
d
ee
p
l
y
in
to
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.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
9
6
9
-
6
9
8
0
6970
s
em
an
tic
in
f
o
r
m
atio
n
ex
t
r
ac
tio
n
.
On
th
e
o
th
e
r
h
an
d
,
th
e
s
ec
o
n
d
ap
p
r
o
ac
h
e
m
p
lo
y
s
d
ee
p
lear
n
in
g
(
DL
)
to
g
en
er
ate
wo
r
d
em
b
e
d
d
in
g
s
th
at
ca
p
tu
r
e
co
n
tex
tu
al
m
ea
n
in
g
s
in
tex
t,
ass
ess
in
g
s
im
ilar
it
y
u
s
in
g
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
s
(
ML
Ps
)
to
d
eter
m
in
e
lab
el
s
im
ilar
ities
[
7
]
.
W
ell
-
k
n
o
wn
wo
r
d
em
b
ed
d
in
g
m
o
d
els
lik
e
W
o
r
d
2
Vec
,
Glo
Ve,
an
d
Fas
tTe
x
t
ar
e
co
m
m
o
n
ly
em
p
lo
y
ed
f
o
r
STS
a
n
d
tr
ain
ed
o
n
g
en
er
al
-
d
o
m
ain
co
r
p
o
r
a.
Ho
wev
e
r
,
ap
p
ly
in
g
C
o
r
p
u
s
-
b
ased
STS
m
eth
o
d
s
in
B
io
NL
P
r
esear
ch
n
ec
ess
itates s
p
ec
ializ
ed
ap
p
r
o
ac
h
es,
s
u
ch
as
u
s
in
g
d
o
m
ain
-
s
p
ec
if
ic
b
io
m
e
d
ical
co
r
p
o
r
a
o
r
b
io
m
ed
ical
k
n
o
wled
g
e
s
o
u
r
ce
s
.
Desp
ite
th
ese
ef
f
o
r
ts
,
th
eir
ad
o
p
tio
n
w
ith
in
th
e
b
io
m
ed
ical
d
o
m
ain
r
em
ain
s
lim
ited
[
8
]
.
Fu
r
th
er
m
o
r
e
,
th
e
an
n
o
tatio
n
p
r
o
ce
s
s
in
b
io
m
ed
ical
r
esear
ch
d
em
an
d
s
th
e
ex
p
e
r
tis
e
o
f
m
e
d
ical
p
r
o
f
ess
io
n
als
[
7
]
,
[
8
]
,
lea
d
in
g
to
a
s
ca
r
city
o
f
lab
elled
d
atasets
an
d
h
in
d
er
i
n
g
p
r
o
g
r
ess
in
b
io
m
ed
ical
ST
S.
Me
asu
r
in
g
th
e
C
o
s
in
e
s
im
ilar
ity
(
C
S)
an
d
wo
r
d
m
o
v
er
'
s
d
is
tan
ce
(
W
MD
)
b
etwe
en
two
s
en
ten
ce
s
is
wid
ely
r
ec
o
g
n
ized
as
th
e
s
im
p
lest
an
d
m
o
s
t
f
r
eq
u
en
tly
ap
p
li
ed
m
eth
o
d
f
o
r
ev
alu
atin
g
STS
in
u
n
lab
eled
d
ata.
Ho
wev
er
,
C
S
o
f
ten
ex
h
ib
its
lo
wer
p
er
f
o
r
m
an
ce
lev
els
[
9
]
–
[
1
2
]
.
C
o
n
v
er
s
ely
,
W
MD
m
eth
o
d
s
s
tr
u
g
g
le
to
d
if
f
er
en
tiate
s
en
ten
ce
s
th
at
s
h
a
r
e
id
en
tical
ter
m
s
b
u
t
p
o
s
s
ess
d
is
tin
ct
s
em
an
tic
m
ea
n
in
g
s
,
p
r
im
ar
ily
d
u
e
to
th
eir
d
is
r
eg
ar
d
f
o
r
wo
r
d
o
r
d
er
[
1
3
]
.
An
o
th
er
wid
ely
ad
o
p
ted
an
d
h
ig
h
ly
r
e
g
ar
d
ed
a
p
p
r
o
ac
h
in
t
h
is
d
o
m
ain
is
th
e
DL
m
o
d
el
u
tili
zin
g
th
e
Siam
ese
n
eu
r
al
n
etwo
r
k
ar
c
h
itectu
r
e
in
tr
o
d
u
ce
d
b
y
Mu
e
ller
an
d
T
h
y
ag
a
r
ajan
[
1
4
]
.
T
h
e
Siam
ese
m
o
d
el
ar
ch
itectu
r
e
in
v
o
lv
es
two
id
en
tical
n
eu
r
al
n
etwo
r
k
s
o
p
er
atin
g
in
p
ar
allel,
ex
tr
ac
tin
g
wo
r
d
r
ep
r
esen
tatio
n
s
f
r
o
m
in
p
u
t
v
ec
to
r
s
.
T
h
e
f
i
n
al
o
u
tp
u
t
is
ty
p
ically
co
m
p
ar
ed
u
s
in
g
C
o
s
in
e
d
is
tan
ce
to
d
eter
m
in
e
STS
[
1
5
]
,
[
1
6
]
.
Su
b
s
eq
u
en
t
r
esear
ch
b
y
Sh
i
et
a
l.
[
1
7
]
in
d
icate
d
th
at
th
e
Ma
n
h
attan
s
im
ilar
ity
m
etr
ic
o
f
f
er
s
f
aster
co
n
v
er
g
en
ce
an
d
h
ig
h
er
ac
c
u
r
ac
y
t
h
an
o
th
e
r
m
etr
ics,
in
clu
d
in
g
C
o
s
in
e
d
is
tan
ce
.
Hen
ce
f
o
r
th
,
th
is
m
o
d
el
is
r
ef
er
r
ed
to
as th
e
Siam
ese
Ma
n
h
attan
.
T
h
e
is
s
u
e
tack
led
in
th
is
p
a
p
er
is
th
e
u
n
d
e
r
u
tili
za
tio
n
a
n
d
d
if
f
icu
lty
o
f
tr
a
n
s
f
o
r
m
in
g
lar
g
e
v
o
lu
m
es
o
f
u
n
s
tr
u
ctu
r
ed
m
e
d
ical
r
ep
o
r
t
t
ex
t
d
ata
in
to
ac
tio
n
ab
le
in
f
o
r
m
atio
n
.
T
h
e
p
r
o
p
o
s
ed
s
o
lu
tio
n
in
th
is
p
ap
er
is
to
im
p
lem
en
t
th
e
Siam
ese
Ma
n
h
attan
ar
ch
itectu
r
e
in
DL
to
as
s
ess
B
io
m
ed
ical
ST
S
u
s
in
g
u
n
lab
eled
leu
k
em
ia
m
ed
ical
r
ep
o
r
t
d
ata
an
d
a
C
o
r
p
u
s
-
b
ased
s
tr
ateg
y
with
b
io
m
ed
ical
d
o
m
ain
-
s
p
ec
if
ic
co
r
p
o
r
a
k
n
o
wn
as
B
io
W
o
r
d
Vec
.
T
h
e
m
o
d
el
in
c
o
r
p
o
r
ates
two
d
is
tin
ct
lab
ellin
g
p
r
o
ce
s
s
es
u
tili
zin
g
th
e
C
S
an
d
W
MD
m
eth
o
d
s
.
T
h
e
p
a
p
er
aim
s
t
o
d
e
m
o
n
s
tr
at
e
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
Sia
m
ese
Ma
n
h
attan
m
o
d
el
i
n
ac
c
u
r
ately
in
ter
p
r
etin
g
an
d
r
etr
iev
in
g
in
f
o
r
m
atio
n
f
r
o
m
b
io
m
ed
ical
d
o
cu
m
e
n
ts
,
ad
d
r
ess
in
g
th
e
p
r
o
b
lem
o
f
u
n
d
er
u
tili
ze
d
tex
tu
al
d
ata
in
th
e
b
io
m
e
d
ical
f
ield
.
T
h
e
s
tr
u
ctu
r
e
o
f
th
is
p
ap
er
is
as
f
o
llo
ws:
s
ec
tio
n
2
will
r
ev
iew
r
elate
d
wo
r
k
o
n
STS.
Sectio
n
3
will
d
etail
th
e
m
eth
o
d
o
lo
g
y
em
p
l
o
y
ed
in
th
is
s
tu
d
y
,
wh
ile
s
ec
tio
n
4
will
p
r
esen
t
th
e
r
esu
lts
an
d
d
is
cu
s
s
io
n
.
Fin
ally
,
s
ec
tio
n
5
will su
m
m
ar
ize
th
e
c
o
n
clu
s
io
n
s
d
r
awn
f
r
o
m
th
e
r
es
ea
r
ch
f
in
d
in
g
s
.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
Acc
o
r
d
in
g
to
d
e
So
u
za
[
1
8
]
,
th
e
Siam
ese
Ma
n
h
attan
ar
c
h
itectu
r
e
en
a
b
les
th
e
c
u
s
to
m
izatio
n
o
f
n
etwo
r
k
s
f
o
r
s
p
ec
if
ic
task
s
,
lev
er
ag
in
g
d
ee
p
lear
n
in
g
m
o
d
el
s
lik
e
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
e
two
r
k
s
(
C
NN)
an
d
lo
n
g
s
h
o
r
t
-
te
r
m
m
em
o
r
y
(
L
S
T
M)
.
R
ec
en
t
ad
v
an
ce
m
e
n
ts
in
STS
task
s
u
s
in
g
DL
h
av
e
b
e
en
ac
h
iev
ed
th
r
o
u
g
h
C
NN
an
d
L
STM
m
o
d
els,
wh
ich
an
aly
ze
wo
r
d
s
an
d
s
en
ten
c
es
to
ca
p
tu
r
e
b
o
th
m
ea
n
in
g
an
d
s
tr
u
ctu
r
al
asp
ec
ts
f
o
r
STS
ca
lcu
latio
n
[
1
9
]
.
Sev
e
r
al
s
tu
d
ies
h
av
e
ex
p
lo
r
ed
th
e
i
m
p
lem
en
tatio
n
o
f
Siam
ese
Ma
n
h
attan
ar
c
h
itectu
r
e
in
DL
m
o
d
els,
s
u
ch
as
th
e
r
es
ea
r
ch
b
y
R
an
asin
g
h
e
et
a
l.
[
1
5
]
,
wh
ich
p
r
o
p
o
s
es
th
e
im
p
lem
en
tatio
n
o
f
Siam
ese
Ma
n
h
attan
o
n
s
ev
er
al
r
ec
u
r
r
e
n
t
n
eu
r
o
n
n
etwo
r
k
(
R
NN)
-
b
a
s
ed
m
o
d
els,
in
clu
d
in
g
L
STM
to
ca
lcu
late
STS
b
etwe
en
tex
t
p
air
s
i
n
Sem
E
v
a
l
d
ata.
T
h
eir
r
esu
lts
s
h
o
wed
th
at
th
eir
m
o
d
el
p
er
f
o
r
m
e
d
b
etter
th
an
th
e
Siam
ese
n
eu
r
al
n
etwo
r
k
m
o
d
el
f
ir
s
t p
r
o
p
o
s
ed
b
y
Mu
eller
an
d
T
h
y
ag
ar
ajan
[
1
4
]
.
Z
h
en
g
et
a
l.
[
2
0
]
in
tr
o
d
u
ce
d
a
C
NN
-
b
ased
m
o
d
el
to
d
etec
t
s
em
an
tic
s
im
ilar
ities
in
m
ed
ica
l
im
ag
in
g
r
ep
o
r
ts
(
e.
g
.
,
u
ltra
s
o
u
n
d
,
MRI
)
an
d
p
ath
o
lo
g
y
.
T
h
ey
en
h
an
ce
d
th
e
m
o
d
el'
s
s
em
an
tic
u
n
d
er
s
tan
d
in
g
th
r
o
u
g
h
em
b
ed
d
in
g
tech
n
iq
u
es,
o
u
tp
e
r
f
o
r
m
in
g
tr
ad
itio
n
al
ap
p
r
o
ac
h
es
lik
e
k
ey
wo
r
d
m
ap
p
in
g
,
L
SA,
laten
t
Dir
ich
let
allo
ca
tio
n
(
L
DA)
,
an
d
Siam
ese
L
STM
.
Similar
ly
,
Sh
i
et
a
l.
[
1
7
]
em
p
lo
y
ed
Siam
ese
C
NN
t
o
ev
alu
ate
s
en
ten
ce
s
im
ilar
ity
in
C
h
in
ese,
co
m
p
ar
in
g
C
o
s
in
e
an
d
Ma
n
h
attan
s
i
m
ilar
ity
m
etr
ics.
T
h
eir
f
in
d
in
g
s
in
d
icate
th
at
th
e
Ma
n
h
attan
s
im
ilar
ity
m
etr
ic
o
u
tp
er
f
o
r
m
s
o
th
er
m
etr
ics,
h
ig
h
lig
h
tin
g
its
ef
f
ec
tiv
en
ess
in
th
i
s
co
n
tex
t.
T
r
an
et
a
l.
[
2
1
]
u
tili
ze
d
a
Siam
ese
n
eu
r
al
n
etwo
r
k
ar
ch
itec
tu
r
e
au
g
m
en
ted
with
s
em
an
ti
c
f
ea
tu
r
es
ex
tr
ac
ted
f
r
o
m
a
k
n
o
wled
g
e
g
r
ap
h
,
ter
m
ed
Siam
ese
KG
-
L
STM
,
in
co
n
ju
n
ctio
n
with
B
io
W
o
r
d
Vec
em
b
ed
d
i
n
g
s
,
to
p
r
ed
ict
s
y
n
o
n
y
m
o
u
s
an
d
n
o
n
-
s
y
n
o
n
y
m
o
u
s
p
air
s
o
f
b
io
m
e
d
ical
ter
m
s
with
in
T
h
e
u
n
if
ie
d
m
ed
ical
la
n
g
u
a
g
e
s
y
s
tem
(
UM
L
S).
T
h
e
UM
L
S,
d
ev
elo
p
e
d
b
y
th
e
U.
S.
Natio
n
a
l
L
ib
r
ar
y
o
f
Me
d
icin
e,
en
h
an
ce
s
co
m
p
u
ter
s
y
s
tem
s
'
b
io
m
ed
ical
an
d
h
ea
lt
h
lan
g
u
a
g
e
co
m
p
r
eh
en
s
io
n
.
T
h
eir
s
tu
d
y
d
em
o
n
s
tr
ated
im
p
r
ess
iv
e
p
e
r
f
o
r
m
a
n
ce
m
etr
ics:
9
8
.
2
3
%
ac
cu
r
ac
y
,
9
8
.
3
7
%
r
ec
a
ll,
9
7
.
4
0
%
p
r
ec
is
io
n
,
a
n
d
an
F
1
-
s
co
r
e
o
f
9
6
.
4
1
%
f
o
r
s
y
n
o
n
y
m
an
d
n
o
n
-
s
y
n
o
n
y
m
p
r
ed
ictio
n
.
L
i
an
d
He
[
2
2
]
c
o
n
s
tr
u
cted
an
R
NN
m
o
d
el
u
s
in
g
a
Siam
ese
n
eu
r
al
n
etwo
r
k
ar
ch
itectu
r
e
with
W
o
r
d
2
Vec
f
o
r
p
r
o
ce
s
s
in
g
wo
r
d
v
ec
to
r
s
.
T
h
e
y
ev
alu
ated
s
e
m
an
tic
s
im
ilar
ity
o
n
a
d
ataset
o
f
2
2
,
6
5
5
p
air
s
o
f
eth
n
ic
m
ed
ical
q
u
esti
o
n
s
u
s
i
n
g
E
u
clid
ea
n
,
C
o
s
in
e,
an
d
Ma
n
h
attan
d
is
tan
ce
s
.
T
h
e
s
tu
d
y
f
o
u
n
d
th
at
th
e
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
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f d
ee
p
S
ia
mese
mo
d
els fo
r
med
ica
l rep
o
r
ts
text
s
imila
r
ity
(
Dia
n
K
u
r
n
ia
s
a
r
i)
6971
Ma
n
h
attan
d
is
tan
ce
ac
h
iev
e
d
th
e
h
ig
h
est
s
im
ilar
ity
(
F
1
-
s
co
r
e
o
f
9
7
.
3
4
%),
f
o
llo
wed
b
y
C
o
s
in
e
(
9
6
.
9
3
%)
an
d
E
u
clid
ea
n
d
is
tan
ce
s
(
9
4
.
1
3
%).
A
r
ec
en
t
s
tu
d
y
b
y
Ya
n
g
et
a
l.
[
2
3
]
em
p
lo
y
e
d
a
Siam
ese
n
e
u
r
al
n
etwo
r
k
–
C
NN
to
p
r
ed
ict
d
r
u
g
-
d
r
u
g
in
ter
ac
tio
n
(
DDI
)
e
v
en
ts
b
y
tr
ea
tin
g
ea
ch
d
r
u
g
s
ep
a
r
ately
w
ith
two
C
NN
s
u
b
-
n
etwo
r
k
s
s
h
ar
in
g
p
ar
a
m
eter
s
to
lear
n
m
u
ltimo
d
al
d
r
u
g
in
f
o
r
m
atio
n
.
Me
r
g
e
d
f
ea
tu
r
e
r
ep
r
e
s
en
tatio
n
s
wer
e
f
ed
in
to
a
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
,
en
ab
lin
g
ac
c
u
r
ate
ca
teg
o
r
izatio
n
o
f
DDI
ev
en
ts
f
r
o
m
m
u
lti
m
o
d
al
d
ata.
Pre
v
io
u
s
s
tu
d
ies o
n
th
e
Siam
ese
Ma
n
h
attan
m
o
d
el
h
av
e
f
o
cu
s
ed
o
n
its
im
p
lem
en
tatio
n
with
in
s
in
g
le
-
m
o
d
el
ar
ch
itectu
r
es
s
u
ch
as
C
NN
o
r
L
STM
.
I
n
co
n
tr
ast,
o
u
r
r
esear
ch
ex
p
lo
r
es
its
ap
p
licati
o
n
in
h
y
b
r
i
d
C
NN
-
L
STM
an
d
L
STM
-
C
NN
m
o
d
els.
T
h
is
in
teg
r
atio
n
r
ep
r
esen
ts
a
n
o
v
el
co
n
tr
ib
u
tio
n
to
o
u
r
s
tu
d
y
.
Fu
r
th
er
m
o
r
e,
we
in
tr
o
d
u
ce
t
h
e
ad
ap
tatio
n
o
f
th
e
Siam
ese
Ma
n
h
attan
ar
c
h
itectu
r
e
with
B
io
W
o
r
d
Vec
a
n
d
a
d
u
al
lab
ellin
g
p
r
o
ce
s
s
u
s
in
g
C
S
an
d
W
MD
,
s
p
ec
if
ically
f
o
r
th
e
b
io
m
e
d
i
ca
l
d
o
m
ain
,
wh
ich
ad
d
r
ess
es
th
e
ch
allen
g
es
o
f
co
n
t
ex
tu
al
m
ea
n
in
g
a
n
d
wo
r
d
o
r
d
er
i
n
b
io
m
e
d
ical
tex
ts
.
3.
M
E
T
H
O
D
T
h
is
s
tu
d
y
i
n
t
r
o
d
u
c
es
C
NN
a
n
d
L
S
T
M
m
o
d
e
ls
u
s
i
n
g
t
h
e
Si
a
m
ese
Ma
n
h
att
a
n
a
r
c
h
it
ec
t
u
r
e
.
Fu
r
th
e
r
m
o
r
e
,
we
s
u
g
g
est
c
o
m
b
i
n
i
n
g
t
h
ese
m
o
d
els
C
NN
-
L
ST
M
a
n
d
L
S
T
M
-
C
NN
b
y
i
n
t
e
g
r
ati
n
g
t
h
e
Siam
ese
Ma
n
h
a
tta
n
ap
p
r
o
ac
h
t
o
m
o
d
el
S
T
S
in
m
e
d
ic
al
r
e
p
o
r
ts
.
Se
v
e
r
al
s
t
u
d
ies
[
2
4
]
–
[
2
6
]
s
h
o
w
t
h
a
t
th
is
h
y
b
r
id
a
p
p
r
o
ac
h
h
ar
n
ess
e
s
C
NN
a
n
d
L
ST
M
s
tr
en
g
t
h
s
t
o
e
n
h
a
n
c
e
S
T
S
ac
cu
r
a
c
y
an
d
p
r
ec
is
io
n
.
T
h
e
d
ataset
u
s
ed
in
th
is
r
e
s
ea
r
ch
is
th
e
GE
NI
A
B
io
m
ed
ical
ev
en
t
tr
ain
d
ata,
wh
ich
co
m
p
r
is
es
u
n
lab
eled
m
ed
ical
r
ep
o
r
ts
d
etailin
g
th
e
ef
f
ec
ts
o
f
d
r
u
g
u
s
e
o
n
in
d
iv
id
u
als
aim
ed
at
id
en
tify
i
n
g
s
p
ec
if
ic
m
ed
ical
co
n
d
itio
n
s
.
T
h
is
d
ataset
in
clu
d
es 4
,
9
5
7
r
ec
o
r
d
s
wit
h
f
o
u
r
v
ar
iab
les:
s
en
ten
ce
,
T
r
ig
g
er
W
o
r
d
,
T
r
ig
g
er
W
o
r
d
L
o
c,
an
d
E
v
e
n
tTy
p
e.
Ho
wev
er
,
t
h
is
s
tu
d
y
f
o
cu
s
es
o
n
t
h
r
ee
v
a
r
iab
les:
s
en
ten
ce
(
b
io
m
ed
ical
t
ex
t)
,
T
r
ig
g
er
W
o
r
d
(
tr
ig
g
er
wo
r
d
in
a
s
en
ten
ce
)
,
a
n
d
E
v
en
tTy
p
e
(
ty
p
e
o
f
b
i
o
m
ed
ical
ev
en
t
r
elate
d
to
th
e
T
r
ig
g
er
W
o
r
d
)
.
E
ac
h
p
air
:
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
,
s
en
ten
ce
-
E
v
en
tT
y
p
e,
an
d
T
r
ig
g
er
W
o
r
d
-
E
v
en
tT
y
p
e
,
will
b
e
an
aly
ze
d
f
o
r
s
em
an
ti
c
s
im
ilar
ity
,
r
esu
ltin
g
in
4
,
9
5
7
p
air
s
f
o
r
ea
ch
c
o
m
b
in
atio
n
.
Fig
u
r
e
1
illu
s
tr
ates th
e
r
esear
ch
m
eth
o
d
o
lo
g
y
.
Fig
u
r
e
1
.
R
esear
ch
m
eth
o
d
o
l
o
g
y
3
.
1
.
P
re
pro
ce
s
s
ing
A
m
ed
ical
r
ep
o
r
t
is
an
o
f
f
icial
d
o
cu
m
en
t
a
u
th
o
r
e
d
b
y
m
ed
ical
p
r
o
f
ess
io
n
als
th
at
co
n
tain
s
co
m
p
r
eh
e
n
s
iv
e
d
etails
ab
o
u
t
a
p
atien
t'
s
d
iag
n
o
s
is
,
tr
ea
tm
en
t,
an
d
th
er
ap
y
.
Ho
wev
er
,
m
an
y
m
ed
ical
r
ep
o
r
ts
s
u
f
f
er
f
r
o
m
p
o
o
r
s
en
ten
ce
g
r
am
m
ar
[
2
7
]
,
u
n
s
tr
u
ct
u
r
ed
f
o
r
m
attin
g
[
2
8
]
,
an
d
lar
g
e
f
ile
s
izes.
C
o
n
s
eq
u
en
tly
,
p
r
ep
r
o
ce
s
s
in
g
is
e
s
s
en
tial
to
f
ac
ilit
ate
m
o
r
e
ef
f
icien
t
an
aly
s
is
.
T
h
e
tex
t
p
r
ep
r
o
ce
s
s
in
g
s
te
p
s
ap
p
lied
to
th
ese
m
ed
ical
r
ep
o
r
ts
in
clu
d
e:
a.
L
o
wer
ca
s
in
g
:
C
o
n
v
er
tin
g
all
u
p
p
er
ca
s
e
letter
s
to
lo
wer
ca
s
e
is
cr
u
cial
b
ec
au
s
e
co
m
p
u
ter
p
r
o
g
r
am
s
in
ter
p
r
et
v
ar
iatio
n
s
in
ca
p
italizatio
n
(
e.
g
.
,
L
eu
k
em
ia
v
s
leu
k
e
m
ia
)
as
d
i
s
tin
ct
wo
r
d
v
ec
to
r
s
,
lea
d
in
g
to
d
if
f
er
en
t
r
esu
lts
[
2
9
]
.
b.
R
em
o
v
e
p
u
n
ctu
atio
n
:
Pu
n
ctu
at
io
n
m
ar
k
s
s
u
ch
as
p
er
io
d
s
,
co
m
m
as,
ex
clam
atio
n
m
ar
k
s
,
an
d
q
u
esti
o
n
m
ar
k
s
ar
e
r
em
o
v
e
d
b
ec
au
s
e
co
m
p
u
te
r
p
r
o
g
r
am
s
d
o
n
o
t
co
m
p
r
eh
e
n
d
p
u
n
ct
u
atio
n
,
an
d
th
eir
p
r
esen
ce
in
th
e
tex
t
is
tr
ea
ted
as n
o
is
e.
c.
Sto
p
wo
r
d
r
e
m
o
v
al
:
T
h
is
s
tep
elim
in
ates
m
ea
n
in
g
less
wo
r
d
s
,
b
io
m
e
d
ical
d
o
m
ai
n
-
s
p
e
cif
ic
wo
r
d
s
,
o
r
co
m
m
o
n
p
lace
wo
r
d
s
in
th
e
tex
t.
Ad
d
itio
n
ally
,
r
ar
e
wo
r
d
s
ar
e
r
em
o
v
ed
as
th
e
y
ar
e
to
o
n
u
m
er
o
u
s
an
d
d
o
n
o
t
co
n
tr
ib
u
te
s
ig
n
if
ica
n
tly
to
ST
S task
s
.
d.
L
em
m
atiza
tio
n
:
T
h
is
c
r
u
cial
s
tep
tr
an
s
f
o
r
m
s
wo
r
d
s
in
to
th
eir
b
ase
m
o
r
p
h
o
lo
g
ical
f
o
r
m
s
(
lem
m
as),
en
h
an
cin
g
th
e
co
n
s
is
ten
cy
an
d
ac
cu
r
ac
y
o
f
tex
t a
n
aly
s
is
[
3
0
]
.
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.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
9
6
9
-
6
9
8
0
6972
Af
ter
u
n
d
er
g
o
in
g
v
ar
i
o
u
s
s
tep
s
an
d
tr
an
s
f
o
r
m
atio
n
s
to
en
s
u
r
e
its
clea
n
lin
ess
,
s
tr
u
ctu
r
e,
an
d
r
ea
d
in
ess
f
o
r
f
u
r
t
h
er
an
aly
s
is
,
th
e
in
itially
4
,
9
5
7
s
en
ten
ce
p
air
s
o
f
te
x
tu
al
d
ata
h
a
v
e
n
o
w
b
ee
n
r
ed
u
ce
d
to
o
n
ly
1
,
4
0
9
p
air
s
f
o
r
ea
c
h
co
m
b
in
atio
n
.
3
.
2
.
L
a
bellin
g
STS
task
s
n
ec
es
s
itate
lab
elle
d
d
ata,
th
er
e
b
y
r
e
q
u
ir
in
g
th
e
an
n
o
tatio
n
o
f
th
is
d
ata.
T
h
e
an
n
o
tatio
n
p
r
o
ce
s
s
in
v
o
lv
es
u
tili
zin
g
C
S
an
d
W
MD
m
eth
o
d
s
.
C
S
ca
lcu
lates
th
e
s
im
ilar
ity
b
et
wee
n
two
o
b
jects,
r
ep
r
esen
ted
b
y
v
ec
to
r
s
o
f
d
o
cu
m
en
t
k
ey
wo
r
d
s
.
T
h
e
s
im
ilar
ity
s
co
r
e
in
C
S
r
an
g
es
f
r
o
m
0
to
1
:
0
s
ig
n
if
ies
co
m
p
lete
d
is
s
im
ilar
ity
b
etwe
en
th
e
o
b
jects,
wh
er
ea
s
1
i
n
d
ica
tes
th
eir
id
en
tity
o
r
ex
ac
t
s
im
il
ar
ity
[
3
1
]
.
As
s
h
o
wn
in
(
1
)
is
em
p
l
o
y
ed
t
o
co
m
p
u
te
th
e
s
im
ilar
ity
s
co
r
e
u
s
in
g
C
S.
(
,
)
=
(
.
)
‖
‖
.
‖
‖
(
1
)
wh
er
e
‖
‖
is
len
g
th
o
f
v
ec
to
r
an
d
‖
‖
is
len
g
th
o
f
v
ec
to
r
I
n
co
n
tr
ast,
W
MD
is
a
m
et
r
ic
f
o
r
q
u
an
tify
i
n
g
t
h
e
d
is
s
im
ilar
it
y
b
etwe
en
two
d
o
cu
m
en
ts
b
y
co
m
p
u
tin
g
th
e
m
in
im
u
m
d
is
tan
ce
r
eq
u
i
r
ed
to
co
n
v
er
t
o
n
e
d
o
cu
m
e
n
t'
s
v
o
ca
b
u
lar
y
in
to
an
o
th
er
[
3
2
]
.
T
h
e
s
p
ec
if
ic
f
o
r
m
u
latio
n
o
f
W
MD
is
d
etailed
in
(
2
)
.
(
,
)
=
‖
−
‖
2
(
2
)
w
h
er
e
is
wo
r
d
m
o
v
er
'
s
d
is
tan
ce
an
d
is
d
o
cu
m
en
t
weig
h
t
.
T
h
e
C
S
an
d
W
MD
m
eth
o
d
s
s
u
cc
ess
f
u
lly
lab
elled
1
,
4
0
9
s
en
te
n
ce
-
T
r
ig
g
e
r
W
o
r
d
p
air
s
,
1
,
409
s
en
ten
ce
-
E
v
en
tTy
p
e
p
air
s
,
an
d
1
,
4
0
9
T
r
ig
g
er
W
o
r
d
-
E
v
en
tT
y
p
e
p
air
s
.
An
ex
am
p
le
d
em
o
n
s
tr
atin
g
th
e
r
esu
lts
o
f
th
is
lab
ellin
g
p
r
o
ce
s
s
,
ex
p
licitly
u
s
in
g
th
e
C
S
m
eth
o
d
,
ca
n
b
e
f
o
u
n
d
in
T
a
b
le
1
.
T
h
is
tab
le
s
h
o
wca
s
es
h
o
w
th
e
C
S
m
eth
o
d
ca
teg
o
r
izes
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
p
air
s
,
h
ig
h
lig
h
ti
n
g
its
ef
f
ec
tiv
en
ess
in
lab
ellin
g
.
T
ab
le
1
.
L
a
b
ellin
g
u
s
in
g
C
S
S
e
n
t
e
n
c
e
Tr
i
g
g
e
r
W
o
r
d
La
b
e
l
d
o
w
n
r
e
g
u
l
a
t
i
o
n
i
n
t
e
r
f
e
r
o
n
r
e
g
u
l
a
t
o
r
y
f
a
c
t
o
r
4
g
e
n
e
l
e
u
k
e
m
i
c
c
e
l
l
d
u
e
h
y
p
e
r
met
h
y
l
a
t
i
o
n
C
p
G
m
o
t
i
f
p
r
o
m
o
t
e
r
r
e
g
i
o
n
d
o
w
n
r
e
g
u
l
a
t
i
o
n
1
f
i
r
st
t
r
e
a
t
me
n
t
I
R
F
4
n
e
g
a
t
i
v
e
l
y
mp
h
o
i
d
m
y
e
l
o
i
d
m
o
n
o
c
y
t
i
c
c
e
l
l
l
i
n
e
m
e
t
h
y
l
a
t
i
o
n
i
n
h
i
b
i
t
o
r
d
e
o
x
y
c
y
t
i
d
i
n
e
r
e
su
l
t
t
i
m
e
c
o
n
c
e
n
t
r
a
t
i
o
n
d
e
p
e
n
d
e
n
t
i
n
c
r
e
a
se
I
R
F
4
m
R
N
A
p
r
o
t
e
i
n
l
e
v
e
l
n
e
g
a
t
i
v
e
1
sec
o
n
d
u
se
r
e
st
r
i
c
t
i
o
n
P
C
R
a
s
sa
y
b
i
s
u
l
f
i
t
e
se
q
u
e
n
c
i
n
g
i
d
e
n
t
i
f
y
s
p
e
c
i
f
i
c
a
l
l
y
met
h
y
l
a
t
e
d
C
p
G
si
t
I
R
F
4
n
e
g
a
t
i
v
e
I
R
F
4
p
o
s
i
t
i
v
e
c
e
l
l
n
e
g
a
t
i
v
e
1
t
h
i
r
d
c
l
e
a
r
l
y
d
e
t
e
r
m
i
n
e
p
r
o
m
o
t
e
r
m
e
t
h
y
l
a
t
i
o
n
m
e
c
h
a
n
i
s
m
I
R
F
4
d
o
w
n
r
e
g
u
l
a
t
i
o
n
v
i
a
r
e
p
o
r
t
e
r
g
e
n
e
a
ss
a
y
d
e
t
e
c
t
a
sso
c
i
a
t
i
o
n
m
e
t
h
y
l
a
t
i
o
n
a
l
s
t
a
t
u
s m
R
N
A
DNA
m
e
t
h
y
l
t
r
a
n
sf
e
r
a
ses
M
e
t
h
y
l
-
C
p
G
-
b
i
n
d
i
n
g
p
r
o
t
e
i
n
d
o
w
n
r
e
g
u
l
a
t
i
o
n
1
t
o
g
e
t
h
e
r
d
a
t
a
s
u
g
g
e
st
C
p
G
si
t
e
sp
e
c
i
f
i
c
i
r
f
4
p
r
o
m
o
t
e
r
me
t
h
y
l
a
t
i
o
n
p
u
t
a
t
i
v
e
mec
h
a
n
i
sm
d
o
w
n
r
e
g
u
l
a
t
e
d
I
R
F
4
l
e
u
k
e
mi
a
d
o
w
n
r
e
g
u
l
a
t
e
d
1
3
.
3
.
Da
t
a
s
pli
t
t
ing
At
th
is
s
tag
e,
th
e
d
ata
is
d
iv
id
ed
in
to
tr
ain
in
g
a
n
d
test
in
g
s
ets,
a
p
r
o
ce
s
s
k
n
o
wn
as
d
ata
s
p
litt
in
g
.
T
h
e
tr
ain
in
g
d
ata
is
u
s
ed
to
b
u
ild
lear
n
in
g
m
o
d
els,
wh
ile
th
e
test
in
g
d
ata
is
u
s
ed
to
ev
al
u
ate
m
o
d
el
p
er
f
o
r
m
a
n
ce
[
3
3
]
.
C
o
n
s
eq
u
en
tly
,
th
e
p
r
o
p
o
r
tio
n
o
f
tr
ain
in
g
d
ata
m
u
s
t
b
e
m
o
r
e
s
ig
n
if
ican
t
th
an
th
at
o
f
test
in
g
d
ata
to
p
r
e
v
en
t
o
v
er
f
itti
n
g
.
T
h
e
d
ata
s
p
litt
in
g
s
tag
e
also
in
clu
d
es
th
e
d
ata
d
i
v
is
io
n
in
to
th
e
'
lef
t'
an
d
'
r
ig
h
t'
in
p
u
t
s
id
es
to
alig
n
with
th
e
in
p
u
t r
e
q
u
ir
em
e
n
ts
o
f
th
e
Siam
ese
n
etwo
r
k
.
T
h
is
s
tu
d
y
'
s
d
ata
is
d
iv
id
ed
,
with
7
5
% a
llo
ca
ted
f
o
r
tr
ain
in
g
an
d
2
5
% f
o
r
test
in
g
.
T
h
is
7
5
/2
5
s
p
lit is
a
s
tan
d
ar
d
a
p
p
r
o
ac
h
in
m
ac
h
in
e
lear
n
in
g
.
I
t
is
b
alan
cin
g
th
e
n
ee
d
f
o
r
c
o
m
p
r
eh
en
s
iv
e
tr
ain
in
g
with
e
n
o
u
g
h
d
ata
to
test
an
d
v
alid
ate
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
r
ig
o
r
o
u
s
ly
.
3
.
4
.
B
io
Wo
rdVec
B
io
W
o
r
d
Vec
,
d
ev
elo
p
e
d
b
y
Z
h
an
g
et
a
l.
[
3
4
]
,
is
in
teg
r
ate
d
i
n
to
o
u
r
m
o
d
el
as in
p
u
t e
m
b
ed
d
in
g
lay
e
r
s
to
an
aly
ze
th
e
s
p
ec
if
ic
an
d
lo
ca
l
co
n
tex
t
o
f
wo
r
d
s
with
in
m
ed
ical
r
ep
o
r
ts
.
T
h
is
wo
r
d
em
b
ed
d
in
g
is
tr
ain
ed
u
s
in
g
b
io
m
ed
ical
c
o
r
p
o
r
a
f
r
o
m
b
i
o
m
ed
ical
liter
atu
r
e
an
d
m
e
d
ical
s
u
b
ject
h
ea
d
in
g
s
(
Me
SH)
d
o
m
ain
k
n
o
wled
g
e.
Z
h
an
g
et
a
l.
[
3
4
]
u
tili
ze
d
a
s
u
b
wo
r
d
em
b
ed
d
i
n
g
m
o
d
el
t
o
im
p
r
o
v
e
u
n
d
e
r
s
tan
d
in
g
o
f
tex
t
s
eq
u
en
ce
s
an
d
m
ed
ica
l
ter
m
in
o
lo
g
y
in
Me
SH,
en
h
an
c
in
g
b
io
m
e
d
ical
wo
r
d
r
ep
r
esen
t
atio
n
s
an
d
s
em
an
tic
co
m
p
r
eh
e
n
s
io
n
.
B
io
W
o
r
d
Ve
c
i
s
c
at
eg
o
r
i
ze
d
i
n
to
in
tr
in
s
i
c
an
d
ex
tr
in
s
i
c
ty
p
es
.
I
n
tr
in
s
i
c
B
io
W
o
r
d
V
ec
i
s
co
m
m
o
n
ly
u
s
ed
to
p
r
ed
ic
t
s
em
an
t
ic
s
im
i
lar
ity
am
o
n
g
wo
r
d
s
,
t
er
m
s
,
o
r
s
en
t
en
ce
s
.
C
o
n
v
e
r
s
ely
,
ex
tr
i
n
s
ic
B
io
W
o
r
d
Ve
c
s
er
v
e
s
a
s
f
ea
tu
r
e
in
p
u
t
in
v
ar
io
u
s
N
L
P
ta
s
k
s
,
s
u
ch
a
s
r
el
a
t
io
n
ex
tr
a
ct
io
n
an
d
t
ex
t
cl
a
s
s
i
f
ic
at
io
n
.
T
h
i
s
d
u
a
l
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
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f d
ee
p
S
ia
mese
mo
d
els fo
r
med
ica
l rep
o
r
ts
text
s
imila
r
ity
(
Dia
n
K
u
r
n
ia
s
a
r
i)
6973
f
u
n
ct
io
n
a
l
ity
m
ak
e
s
B
i
o
W
o
r
d
Vec
a
v
er
s
a
t
il
e
to
o
l
f
o
r
b
o
o
s
t
in
g
th
e
p
er
f
o
r
m
an
ce
o
f
NL
P
ap
p
l
ic
at
io
n
s
in
th
e
b
io
m
ed
ic
al
f
i
eld
.
3
.
5
.
Sem
a
ntic
s
im
ila
rit
y
m
o
dels
3
.
5
.
1
.
C
o
nv
o
lutio
na
l neura
l
net
wo
rk
s
(
CNN)
a
nd
lo
ng
s
ho
rt
-
t
er
m
m
e
m
o
r
y
(
L
ST
M
)
A
C
NN
DL
alg
o
r
ith
m
id
en
tifi
es
s
en
ten
ce
p
atter
n
s
an
d
s
em
an
tic
r
elatio
n
s
h
ip
s
th
r
o
u
g
h
a
c
o
n
v
o
lu
ti
o
n
al
lay
er
lin
k
ed
to
lo
ca
l
f
ea
tu
r
es.
T
y
p
ically
,
a
C
NN
co
n
s
is
ts
o
f
m
u
ltip
le
lay
er
s
:
in
p
u
t,
co
n
v
o
lu
tio
n
al,
p
o
o
lin
g
,
f
u
lly
co
n
n
ec
ted
,
an
d
o
u
t
p
u
t
lay
e
r
s
[
3
5
]
.
C
NN
h
as
s
u
cc
ee
d
e
d
in
v
a
r
io
u
s
NL
P
task
s
,
s
u
ch
as
s
en
te
n
ce
r
ep
r
esen
tatio
n
,
s
ea
r
ch
q
u
er
y
r
etr
iev
al,
a
n
d
s
e
m
an
tic
p
ar
s
in
g
.
T
h
ey
ar
e
p
ar
ti
cu
lar
ly
ef
f
ec
tiv
e
i
n
id
en
tify
i
n
g
s
em
an
tic
s
im
ilar
itie
s
b
etwe
en
tex
t
p
air
s
,
ex
ce
llin
g
in
task
s
r
elate
d
to
STS
[
2
0
]
.
Ho
wev
er
,
C
NNs
h
av
e
a
lim
itati
o
n
:
th
eir
ar
ch
itectu
r
e
d
o
es
n
o
t
ac
co
u
n
t
f
o
r
wo
r
d
o
r
d
er
r
elatio
n
s
h
ip
s
in
s
en
ten
ce
s
,
wh
ich
m
ea
n
s
th
ey
ca
n
n
o
t
an
al
y
ze
th
e
s
eq
u
en
ce
o
f
wo
r
d
s
.
3
.
5
.
2
.
L
o
ng
s
ho
rt
-
t
er
m m
e
mo
ry
(
L
ST
M
)
Un
lik
e
C
NN,
th
e
L
STM
ar
ch
itectu
r
e
is
s
p
ec
if
ically
d
esig
n
ed
f
o
r
p
r
o
ce
s
s
in
g
s
eq
u
en
ce
d
ata.
L
STM
ef
f
icien
tly
d
is
ca
r
d
ir
r
elev
an
t
i
n
f
o
r
m
atio
n
an
d
r
etain
s
o
n
ly
ess
en
tial
in
f
o
r
m
atio
n
,
s
eq
u
en
t
ially
ca
p
tu
r
in
g
th
e
ess
en
ce
o
f
s
en
ten
ce
s
th
r
o
u
g
h
ce
ll
s
tates
an
d
g
ates
[
3
6
]
.
T
h
is
ca
p
ab
ilit
y
m
ak
es
L
STM
p
ar
ticu
lar
ly
well
-
s
u
ited
f
o
r
task
s
in
v
o
lv
in
g
s
em
an
tic
s
im
ilar
ity
,
as th
ey
ca
n
u
n
d
er
s
tan
d
th
e
co
n
tex
t a
n
d
m
ea
n
i
n
g
o
f
wo
r
d
s
in
s
en
ten
ce
s
s
eq
u
en
tially
.
3
.
5
.
3
.
H
y
brid
m
o
del
T
h
e
h
y
b
r
id
m
o
d
el
a
n
aly
ze
s
w
o
r
d
s
f
r
o
m
b
o
th
g
e
n
er
al
a
n
d
lo
ca
l
co
n
tex
ts
.
T
er
m
s
ar
e
in
ter
p
r
eted
u
s
in
g
wo
r
d
em
b
ed
d
in
g
f
o
r
g
e
n
er
al
c
o
n
tex
t
an
d
s
p
ec
if
ic
s
em
an
tic
a
n
d
s
y
n
tactic
f
ea
t
u
r
es
f
o
r
lo
ca
l
co
n
tex
t.
T
h
er
e
ar
e
two
h
y
b
r
id
m
o
d
els:
C
NN
-
L
STM
an
d
L
STM
-
C
NN.
Hy
b
r
i
d
C
NN
-
L
STM
an
d
L
STM
-
C
NN
m
o
d
els
p
r
o
ce
s
s
d
ata
in
d
if
f
er
e
n
t
s
eq
u
en
ce
s
.
I
n
th
e
h
y
b
r
id
C
NN
-
L
STM
m
o
d
el,
C
NN
f
ir
s
t
ex
tr
ac
ts
th
e
lo
ca
l
co
n
tex
t
o
f
ea
ch
w
o
r
d
i
n
a
s
en
ten
ce
,
f
o
llo
wed
b
y
L
ST
M
to
c
h
ec
k
wo
r
d
o
r
d
er
.
C
o
n
v
er
s
ely
,
in
th
e
h
y
b
r
id
L
STM
-
C
NN
m
o
d
el,
L
STM
in
itially
ch
ec
k
s
th
e
wo
r
d
o
r
d
e
r
,
an
d
th
e
n
C
NN
ex
tr
ac
ts
th
e
lo
ca
l c
o
n
tex
t
[
1
9
]
.
3
.
5
.
4
.
H
idd
en
la
y
er
Hid
d
en
lay
er
s
a
r
e
cr
u
cial
f
o
r
i
m
p
r
o
v
i
n
g
ac
c
u
r
ac
y
an
d
m
an
a
g
in
g
tim
e
c
o
m
p
lex
ity
in
lear
n
i
n
g
m
o
d
els.
R
esear
ch
b
y
Uza
ir
an
d
J
am
il
[
3
7
]
in
d
icate
s
th
at
th
r
ee
h
id
d
en
lay
er
s
o
f
f
er
o
p
tim
al
p
er
f
o
r
m
an
ce
.
Ho
wev
er
,
u
s
in
g
m
o
r
e
th
an
th
r
ee
h
id
d
en
lay
er
s
d
ir
ec
tly
af
f
ec
ts
th
e
m
o
d
el'
s
ac
cu
r
ac
y
n
e
g
ativ
ely
.
3
.
6
.
E
v
a
lua
t
io
n
Mo
d
el
p
e
r
f
o
r
m
an
ce
is
e
v
alu
at
ed
u
s
in
g
th
e
c
o
n
f
u
s
io
n
m
at
r
ix
[
3
8
]
,
wh
e
r
e
ac
c
u
r
ate
STS
ca
lcu
latio
n
s
ap
p
ea
r
al
o
n
g
th
e
d
iag
o
n
al.
E
r
r
o
r
s
in
STS
ca
lcu
latio
n
s
ar
e
r
e
p
r
esen
ted
o
u
ts
id
e
th
is
d
ia
g
o
n
a
l
lin
e.
T
h
is
m
eth
o
d
p
r
o
v
id
es a
clea
r
an
d
s
tr
u
ct
u
r
ed
ev
alu
atio
n
o
f
h
o
w
well
th
e
m
o
d
el
p
r
e
d
icts
tex
tu
al
s
im
ilar
iti
es.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
p
r
esen
ts
th
e
r
esu
lts
ac
h
iev
ed
th
r
o
u
g
h
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
Var
io
u
s
p
air
s
o
f
s
en
ten
ce
s
wer
e
ev
alu
ated
f
o
r
STS,
in
clu
d
in
g
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
,
s
en
ten
ce
-
E
v
e
n
tTy
p
e,
an
d
T
r
ig
g
er
W
o
r
d
-
E
v
e
n
tTy
p
e.
T
h
e
co
m
p
a
r
is
o
n
o
f
ea
ch
m
o
d
e
l's
r
u
n
tim
e
is
illu
s
tr
ated
in
Fig
u
r
e
s
2
(
a)
to
2
(
c)
r
esp
ec
tiv
ely
.
Ap
p
ly
in
g
th
e
Siam
ese
Ma
n
h
at
tan
ar
ch
itectu
r
e
to
f
o
u
r
d
if
f
er
e
n
t
DL
m
eth
o
d
s
y
ield
ed
id
e
n
tical
r
esu
lts
with
1
0
0
%
ac
c
u
r
ac
y
an
d
v
al
id
atio
n
ac
cu
r
ac
y
.
Ho
wev
er
,
e
ac
h
m
o
d
el
v
a
r
ies
in
th
e
tim
e
r
eq
u
ir
e
d
f
o
r
STS
ca
lcu
latio
n
s
.
C
NN
em
er
g
es
as
th
e
m
o
s
t
tim
e
-
ef
f
icien
t
m
o
d
e
l
f
o
r
tr
ain
in
g
c
o
m
p
ar
e
d
to
o
th
er
s
.
Mo
d
els
lik
e
C
NN
-
L
STM
an
d
L
STM
-
C
NN
h
y
b
r
i
d
s
also
ex
h
ib
it g
o
o
d
ef
f
icien
c
y
.
I
n
co
n
tr
ast,
th
e
s
in
g
le
L
STM
m
o
d
el
p
r
o
v
es
less
ef
f
icien
t
d
u
e
to
lo
n
g
er
p
r
o
ce
s
s
in
g
tim
es.
Dete
r
m
in
in
g
th
e
o
p
tim
al
m
o
d
el
is
ch
allen
g
in
g
wh
en
co
n
s
id
er
in
g
ac
cu
r
ac
y
,
v
alid
atio
n
ac
c
u
r
ac
y
,
an
d
tr
ain
in
g
d
u
r
atio
n
al
o
n
e.
An
alter
n
ativ
e
m
eth
o
d
to
ev
al
u
ate
m
o
d
el
p
er
f
o
r
m
an
ce
in
v
o
l
v
es
an
aly
zin
g
th
e
lo
s
s
g
r
ap
h
.
L
o
s
s
g
r
ap
h
s
in
d
icate
wh
eth
er
a
m
o
d
el
is
u
n
d
er
f
itti
n
g
,
well
-
f
itted
,
o
r
o
v
e
r
f
itti
n
g
.
Un
d
er
f
itti
n
g
s
ig
n
if
ies
p
o
o
r
p
er
f
o
r
m
an
ce
d
u
r
in
g
tr
ain
i
n
g
o
r
test
in
g
.
A
well
-
f
itted
m
o
d
e
l
d
is
p
lay
s
co
n
s
is
ten
t
p
er
f
o
r
m
an
ce
ac
r
o
s
s
tr
ain
in
g
an
d
test
in
g
p
h
ases
,
wh
er
ea
s
o
v
er
f
itti
n
g
m
an
if
ests
as
a
s
ig
n
if
ican
t
g
a
p
b
etwe
en
tr
ain
in
g
an
d
test
in
g
r
esu
lts
.
T
h
e
in
itial
m
o
d
el
ex
am
i
n
ed
th
r
o
u
g
h
lo
s
s
g
r
ap
h
s
was
th
e
C
NN
m
o
d
el
ap
p
lied
to
s
en
te
n
ce
-
T
r
ig
g
er
W
o
r
d
p
air
s
.
Deta
iled
co
m
p
ar
is
o
n
s
o
f
th
ese
g
r
ap
h
s
a
r
e
d
ep
icted
in
Fig
u
r
es 3
(
a
)
to
3
(
d
)
.
Ov
er
all,
am
o
n
g
C
NN
m
o
d
els
ev
alu
ated
f
o
r
STS
in
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
p
air
s
,
th
e
m
o
d
el
u
s
in
g
th
r
ee
h
id
d
en
lay
er
s
with
th
e
W
MD
lab
ellin
g
m
eth
o
d
s
tan
d
s
o
u
t
as th
e
m
o
s
t
s
u
itab
le.
T
h
is
c
o
n
clu
s
io
n
is
d
r
awn
f
r
o
m
th
e
lo
s
s
g
r
ap
h
an
aly
s
is
,
wh
er
e
b
o
th
lo
s
s
an
d
v
alid
atio
n
lo
s
s
v
alu
es
co
n
v
er
g
e
ef
f
ec
tiv
ely
.
I
n
co
n
tr
ast,
o
th
e
r
C
NN
m
o
d
els,
as
d
ep
icted
in
Fig
u
r
e
3
,
s
h
o
w
s
ig
n
s
o
f
u
n
d
er
f
itti
n
g
as
th
ey
f
ail
to
co
n
v
er
g
e.
I
n
te
g
r
atin
g
th
e
W
MD
m
eth
o
d
with
th
r
ee
h
id
d
en
lay
e
r
s
p
r
o
v
es b
e
n
ef
icial,
en
h
a
n
cin
g
o
v
er
all
m
o
d
el
p
e
r
f
o
r
m
an
ce
s
i
g
n
if
ican
tly
.
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.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
9
6
9
-
6
9
8
0
6974
T
h
e
n
ex
t
m
o
d
el
u
n
d
er
ev
alu
a
tio
n
in
th
e
lo
s
s
g
r
ap
h
is
th
e
C
NN
-
L
STM
h
y
b
r
id
m
o
d
el,
il
lu
s
tr
ated
in
Fig
u
r
es
4
(
a)
to
4
(
d
)
.
T
h
e
lo
s
s
g
r
ap
h
s
in
Fig
u
r
es
4
(
a
)
an
d
4
(
b
)
illu
s
tr
ate
th
at
th
e
C
NN
-
L
STM
h
y
b
r
id
m
o
d
el,
u
tili
zin
g
two
h
i
d
d
en
lay
er
s
wit
h
d
if
f
e
r
en
t la
b
ellin
g
m
et
h
o
d
s
,
ex
p
er
ien
ce
s
a
d
ec
r
ea
s
in
g
lo
s
s
v
alu
e
ac
co
m
p
an
ied
b
y
a
n
in
c
r
ea
s
in
g
v
alid
atio
n
lo
s
s
v
alu
e,
i
n
d
icatin
g
o
v
e
r
f
itti
n
g
.
C
o
n
s
eq
u
en
tly
,
t
h
is
m
o
d
el
m
ay
n
o
t
g
e
n
er
alize
well
to
n
ew
d
ata
f
o
r
d
eter
m
i
n
in
g
STS.
C
o
n
v
er
s
ely
,
th
e
C
NN
-
L
STM
h
y
b
r
id
m
o
d
els
d
ep
icted
in
Fig
u
r
es
4
(
c)
an
d
4
(
d
)
ex
h
ib
it
m
o
r
e
b
ala
n
ce
d
lo
s
s
g
r
a
p
h
s
,
wh
er
e
th
e
lo
s
s
an
d
v
alid
at
io
n
lo
s
s
v
alu
es
ar
e
clo
s
ely
alig
n
ed
.
H
o
w
e
v
e
r
,
e
v
e
n
w
i
t
h
t
h
es
e
i
m
p
r
o
v
e
m
e
n
ts
,
t
h
ese
m
o
d
e
l
s
s
t
il
l
f
a
il
t
o
d
e
te
r
m
i
n
e
S
T
S
ac
c
u
r
a
t
e
l
y
.
T
h
e
l
as
t
m
o
d
e
l
a
n
a
l
y
z
e
d
t
h
r
o
u
g
h
t
h
e
l
o
s
s
g
r
a
p
h
i
s
t
h
e
L
S
T
M
-
C
N
N
h
y
b
r
i
d
m
o
d
e
l
,
d
i
s
p
l
a
y
e
d
i
n
F
i
g
u
r
e
s
5
(
a
)
t
o
5
(
d
)
.
B
as
e
d
o
n
F
i
g
u
r
e
s
5
(
a
)
t
o
5
(
d
)
,
t
h
e
l
o
s
s
a
n
d
v
a
li
d
a
t
i
o
n
l
o
s
s
v
a
l
u
e
s
o
f
t
h
e
L
S
T
M
-
C
NN
h
y
b
r
i
d
m
o
d
e
l
c
o
n
v
e
r
g
e
c
o
n
s
is
t
e
n
tl
y
a
c
r
o
s
s
v
a
r
i
o
u
s
l
a
b
el
l
i
n
g
m
e
t
h
o
d
s
a
n
d
h
i
d
d
e
n
l
a
y
e
r
c
o
n
f
i
g
u
r
a
t
i
o
n
s
.
T
h
is
co
n
v
er
g
e
n
c
e
s
ig
n
if
ies
th
at
th
e
m
o
d
el
d
em
o
n
s
tr
ates
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
an
d
g
e
n
er
aliza
b
il
ity
d
u
r
in
g
tr
ain
in
g
an
d
wh
en
a
p
p
lied
to
n
ew
d
ata.
(
a)
(
b
)
(
c)
Fig
u
r
e
2
.
C
o
m
p
a
r
is
o
n
o
f
r
u
n
n
i
n
g
tim
e
s
en
ten
ce
p
ai
r
s
o
n
(
a
)
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
,
(
b
)
s
en
t
en
ce
-
E
v
en
t
T
y
p
e,
an
d
(
c)
T
r
ig
g
e
r
W
o
r
d
-
E
v
en
t
T
y
p
e
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
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f d
ee
p
S
ia
mese
mo
d
els fo
r
med
ica
l rep
o
r
ts
text
s
imila
r
ity
(
Dia
n
K
u
r
n
ia
s
a
r
i)
6975
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
3
.
C
NN
m
o
d
el
lo
s
s
g
r
a
p
h
o
n
s
en
ten
ce
-
T
r
i
g
g
er
W
o
r
d
u
s
in
g
(
a)
2
L
ay
er
+CS
,
(
b
)
2
L
ay
er
+WMD,
(
c)
3
L
a
y
er
+CS
,
an
d
(
d
)
3
L
ay
er
+WMD
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
4
.
C
NN
-
L
STM
h
y
b
r
id
m
o
d
el
lo
s
s
g
r
ap
h
o
n
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
u
s
in
g
(
a
)
2
L
a
y
e
r
+CS
,
(
b
)
2
L
ay
er
+WMD,
(
c)
3
L
ay
e
r
+CS
,
an
d
(
d
)
3
L
ay
e
r
+WMD
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.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
9
6
9
-
6
9
8
0
6976
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
5
.
L
STM
-
C
NN
h
y
b
r
id
m
o
d
el
lo
s
s
g
r
ap
h
o
n
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
u
s
in
g
(
a
)
2
L
a
y
e
r
+CS
,
(
b
)
2
L
ay
er
+WMD,
(
c)
3
L
ay
e
r
+CS
,
an
d
(
d
)
3
L
ay
e
r
+WMD
B
ased
o
n
th
e
an
al
y
s
is
o
f
t
h
e
l
o
s
s
g
r
ap
h
,
it
is
ev
id
en
t
th
at
m
o
r
e
c
o
m
p
lex
m
o
d
els
with
m
o
r
e
th
an
two
h
id
d
en
lay
e
r
s
o
u
tp
er
f
o
r
m
th
o
s
e
with
o
n
ly
two
lay
er
s
.
Fu
r
th
er
m
o
r
e,
th
e
ch
o
ice
o
f
th
e
lab
ellin
g
m
eth
o
d
s
ig
n
if
ican
tly
im
p
ac
ts
th
e
tr
ain
in
g
tim
e,
with
th
e
W
MD
m
e
th
o
d
p
r
o
v
in
g
s
u
p
er
io
r
to
C
S
d
u
e
to
its
f
aster
co
n
v
er
g
en
ce
.
I
n
th
e
L
STM
-
C
NN
h
y
b
r
id
m
o
d
el,
th
e
lo
s
s
g
r
a
p
h
r
em
ain
s
s
tab
le
ac
r
o
s
s
d
if
f
er
en
t
lab
ellin
g
m
eth
o
d
s
an
d
h
id
d
en
la
y
er
co
n
f
ig
u
r
atio
n
s
.
T
h
u
s
,
t
h
e
L
STM
-
C
NN
h
y
b
r
id
m
o
d
el
is
o
p
tim
al
f
o
r
d
eter
m
in
in
g
STS
in
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
,
s
en
ten
ce
-
E
v
en
tTy
p
e,
an
d
E
v
e
n
tTy
p
e
-
T
r
ig
g
er
W
o
r
d
p
air
s
.
T
h
ese
f
in
d
in
g
s
u
n
d
e
r
s
co
r
e
th
e
m
o
d
el'
s
co
n
s
is
ten
cy
an
d
ef
f
ec
tiv
en
ess
ac
r
o
s
s
v
ar
ied
s
en
te
n
ce
p
air
in
g
s
.
T
h
e
STS
ca
lcu
latio
n
r
esu
lts
f
r
o
m
th
e
L
STM
-
C
NN
h
y
b
r
id
m
o
d
el,
id
en
tifie
d
as
th
e
m
o
s
t
ef
f
ec
t
iv
e
m
o
d
el
f
o
r
d
eter
m
i
n
in
g
s
em
an
tic
s
im
ilar
ities
ac
r
o
s
s
th
e
th
r
ee
s
en
ten
ce
p
air
s
,
ar
e
p
r
esen
ted
in
th
e
co
n
f
u
s
io
n
m
at
r
ices
s
h
o
wn
in
Fig
u
r
es
6
(
a)
to
6
(
c)
.
T
h
ese
m
atr
ices
illu
s
tr
ate
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
L
STM
-
C
NN
h
y
b
r
id
m
o
d
el,
wh
ich
f
ea
tu
r
es
a
s
o
p
h
is
ticated
ar
ch
itectu
r
e
an
d
u
tili
ze
s
th
e
o
p
tim
al
lab
ellin
g
m
eth
o
d
,
W
MD
.
Fig
u
r
e
6
(
a)
illu
s
tr
ates
th
at
th
e
s
en
ten
ce
-
T
r
ig
g
er
W
o
r
d
p
air
ac
h
iev
es
a
s
i
m
ilar
ity
s
co
r
e
o
f
1
,
in
d
icatin
g
id
e
n
tical
s
em
an
tic
m
ea
n
in
g
b
etwe
en
t
h
e
s
en
ten
ce
an
d
T
r
ig
g
e
r
W
o
r
d
v
ar
iab
les.
T
h
e
s
en
ten
ce
v
ar
iab
le
co
m
p
r
is
es
s
en
ten
ce
s
r
elate
d
to
leu
k
em
ia
,
en
c
o
m
p
ass
in
g
g
en
etic
r
eg
u
latio
n
,
in
ter
-
p
r
o
tei
n
in
ter
ac
tio
n
s
,
g
e
n
es
o
r
p
r
o
te
in
s
,
an
d
ass
o
ciate
d
b
io
lo
g
ical
p
r
o
ce
s
s
es.
I
n
co
n
tr
a
s
t,
th
e
T
r
ig
g
er
W
o
r
d
v
a
r
iab
le
c
o
n
tain
s
d
escr
ip
tiv
e
ter
m
s
as
in
d
icato
r
s
o
r
k
ey
wo
r
d
s
to
id
en
tif
y
s
p
ec
if
ic
b
io
m
e
d
ica
l
ev
en
ts
with
in
th
e
s
en
ten
ce
a
n
d
esti
m
ate
th
ei
r
tim
in
g
.
T
h
er
ef
o
r
e,
b
ased
o
n
th
e
STS
ca
lcu
latio
n
s
f
o
r
th
e
s
en
ten
ce
-
T
r
ig
g
e
r
W
o
r
d
p
air
,
th
e
m
o
d
el
ef
f
ec
tiv
el
y
ex
tr
ac
ts
m
e
an
in
g
f
u
l
i
n
f
o
r
m
atio
n
an
d
cr
u
cial
elem
e
n
ts
f
r
o
m
b
io
m
ed
ical
tex
ts
p
er
tin
en
t to
le
u
k
em
ia
r
esear
ch
.
Similar
ly
,
Fig
u
r
e
6
(
b
)
r
ev
ea
l
s
th
at
th
e
s
en
ten
ce
-
E
v
en
tTy
p
e
p
air
ac
h
ie
v
es
a
s
im
ilar
ity
s
co
r
e
o
f
1
,
in
d
icatin
g
id
e
n
tical
s
em
an
tic
m
ea
n
in
g
b
etwe
en
t
h
ese
v
ar
iab
les.
Alo
n
g
s
id
e
m
atch
in
g
th
e
T
r
ig
g
e
r
W
o
r
d
'
s
s
em
an
tic
m
ea
n
in
g
,
th
e
s
en
ten
ce
v
ar
iab
le
alig
n
s
clo
s
ely
with
th
e
E
v
en
tT
y
p
e
v
ar
iab
le,
wh
ich
d
en
o
tes
s
p
ec
if
ic
b
io
lo
g
ical
ac
tiv
ities
with
in
th
e
co
n
tex
t o
f
leu
k
em
ia
.
STS c
alcu
latio
n
s
in
th
ese
p
air
s
f
ac
ilit
ate
th
e
id
en
tific
atio
n
o
f
co
r
r
elatio
n
s
b
etwe
en
v
ar
i
o
u
s
b
io
m
ed
ical
ev
en
ts
an
d
leu
k
e
m
ia
,
aid
in
g
r
esear
ch
er
s
an
d
m
ed
ical
p
r
o
f
ess
io
n
als.
Fu
r
th
er
m
o
r
e
,
an
aly
zin
g
s
em
an
tic
r
elatio
n
s
h
ip
s
b
etwe
en
th
ese
b
io
m
ed
ical
ev
en
ts
a
n
d
tex
ts
en
h
an
ce
s
u
n
d
er
s
tan
d
i
n
g
o
f
h
o
w
leu
k
e
m
ia
tr
ea
tm
en
ts
in
f
l
u
en
ce
s
p
ec
if
ic
b
io
lo
g
ical
p
ath
way
s
an
d
r
elate
d
b
io
m
ed
ica
l
p
h
en
o
m
en
a.
T
h
is
ap
p
r
o
ac
h
co
n
tr
ib
u
tes
d
e
ep
er
in
s
ig
h
ts
i
n
to
leu
k
e
m
ia
th
er
a
p
y
an
d
its
im
p
licatio
n
s
f
o
r
b
io
m
ed
ical
r
esear
ch
a
n
d
tr
ea
t
m
en
t
s
tr
ateg
ies.
L
astl
y
,
b
ased
o
n
Fig
u
r
e
6
(
c)
,
th
e
T
r
ig
g
er
W
o
r
d
-
E
v
en
tTy
p
e
p
air
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
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f d
ee
p
S
ia
mese
mo
d
els fo
r
med
ica
l rep
o
r
ts
text
s
imila
r
ity
(
Dia
n
K
u
r
n
ia
s
a
r
i)
6977
ac
h
iev
es
a
s
im
ilar
ity
s
co
r
e
1
,
in
d
icatin
g
a
s
o
lid
s
em
an
tic
r
elatio
n
s
h
ip
b
etwe
en
all
tr
i
g
g
er
wo
r
d
s
in
th
e
T
r
ig
g
er
W
o
r
d
v
ar
iab
le
a
n
d
th
e
b
io
m
ed
ical
ev
e
n
t
ty
p
es
in
th
e
E
v
en
tT
y
p
e
v
a
r
iab
le.
T
h
is
an
aly
s
is
r
ev
ea
ls
s
ig
n
if
ican
t in
s
ig
h
ts
in
to
th
e
b
i
o
lo
g
ical
m
ec
h
a
n
is
m
s
ass
o
ciate
d
with
leu
k
em
ia
d
is
ea
s
e.
(
a)
(
b
)
(
c)
Fig
u
r
e
6
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
n
(
a)
s
en
ten
ce
–
T
r
ig
g
e
r
W
o
r
d
,
(
b
)
s
en
ten
ce
–
E
v
en
tT
y
p
e,
a
n
d
(
c)
T
r
ig
g
er
W
o
r
d
–
E
v
en
tT
y
p
e
T
h
e
to
p
-
p
er
f
o
r
m
in
g
m
o
d
el
i
n
th
is
r
esear
ch
s
tu
d
y
is
b
en
c
h
m
ar
k
ed
a
g
ain
s
t
C
NN,
L
STM
,
a
n
d
h
y
b
r
i
d
C
NN
-
L
STM
m
o
d
els
an
d
m
o
d
els
p
r
o
p
o
s
ed
in
p
r
ev
io
u
s
s
tu
d
i
es.
Ho
wev
er
,
it
is
wo
r
th
n
o
tin
g
th
at
p
r
io
r
s
tu
d
ies
o
f
ten
d
id
n
o
t
ex
clu
s
iv
ely
f
o
c
u
s
o
n
th
e
b
io
m
e
d
ical
d
o
m
ai
n
d
u
e
to
its
r
elativ
ely
lim
ited
r
esear
ch
v
o
l
u
m
e
co
m
p
ar
ed
to
o
th
e
r
f
ield
s
.
T
h
is
b
en
ch
m
ar
k
in
g
is
p
r
esen
ted
in
T
ab
le
2
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
e
x
h
ib
its
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
p
r
ev
io
u
s
m
o
d
els,
as
ev
i
d
en
ce
d
b
y
th
e
co
m
p
ar
ativ
e
r
esu
lts
p
r
ese
n
ted
in
T
ab
le
2
.
Am
o
n
g
t
h
ese,
T
r
an
et
a
l.
[
2
1
]
ac
h
iev
ed
n
o
tab
le
s
u
cc
ess
u
s
in
g
B
io
W
o
r
d
Vec
to
p
r
ed
ict
s
y
n
o
n
y
m
s
an
d
n
o
n
-
s
y
n
o
n
y
m
s
in
th
e
b
io
m
ed
ical
d
o
m
ain
with
an
a
cc
u
r
ac
y
ex
ce
e
d
in
g
9
8
%.
T
h
is
s
tu
d
y
,
al
o
n
g
s
id
e
T
r
an
et
a
l.
[
2
1
]
,
u
n
d
er
s
co
r
es
th
e
ef
f
ec
tiv
en
ess
o
f
em
p
lo
y
i
n
g
p
r
e
-
tr
ain
ed
wo
r
d
em
b
ed
d
in
g
s
tailo
r
ed
to
s
p
ec
if
i
c
r
esear
ch
d
o
m
ain
s
,
s
u
ch
as B
io
W
o
r
d
Vec
.
Sp
ec
if
ically
tr
ain
ed
o
n
d
atasets
f
r
o
m
Pu
b
Me
d
an
d
Me
SH,
B
io
W
o
r
d
Vec
en
h
an
ce
s
th
e
STS
m
o
d
el's
p
er
f
o
r
m
a
n
ce
b
y
p
r
o
v
id
in
g
d
o
m
ai
n
-
s
p
ec
if
ic
em
b
ed
d
in
g
s
r
elev
an
t
to
t
h
e
b
io
m
ed
ical
d
ata
u
s
ed
in
t
h
is
r
esear
ch
.
Mo
r
e
o
v
er
,
th
e
h
ig
h
ac
cu
r
a
cy
ac
h
iev
ed
i
n
th
is
s
tu
d
y
is
also
at
tr
ib
u
tab
le
to
th
e
s
elec
tio
n
o
f
ap
p
r
o
p
r
iate
s
im
ilar
ity
m
etr
ics,
p
ar
ticu
lar
ly
h
ig
h
lig
h
ted
in
T
ab
le
2
,
wh
er
e
th
e
Ma
n
h
attan
m
etr
ic
is
s
h
o
wn
to
p
o
ten
tially
im
p
r
o
v
e
th
e
STS m
o
d
el'
s
ac
cu
r
ac
y
s
ig
n
if
ican
tly
.
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.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
9
6
9
-
6
9
8
0
6978
T
ab
le
2
.
B
en
ch
m
a
r
k
in
g
an
aly
s
is
o
f
STS c
alcu
latio
n
s
Ti
t
l
e
A
u
t
h
o
r
s
M
e
t
h
o
d
s
P
e
r
f
o
r
ma
n
c
e
U
n
l
a
b
e
l
e
d
s
h
o
r
t
t
e
x
t
si
m
i
l
a
r
i
t
y
w
i
t
h
LST
M
e
n
c
o
d
e
r
Y
a
o
e
t
a
l
.
[
3
9
]
a.
D
a
t
a
:
M
S
R
p
a
r
a
p
h
r
a
s
e
a
n
d
Q
u
o
r
a
d
a
t
a
set
(
U
n
l
a
b
e
l
e
d
)
b.
La
b
e
l
l
i
n
g
:
c
o
s
i
n
e
s
i
mi
l
a
r
i
t
y
c.
D
o
ma
i
n
:
g
e
n
e
r
a
l
d.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
W
o
r
d
2
V
e
c
M
o
d
e
l
:
LST
M
e
n
c
o
d
e
r
A
c
c
u
r
a
c
y
M
S
R
:
8
8
,
0
0
%
R
e
c
a
l
l
M
S
R
:
8
7
,
0
0
%
A
c
c
u
r
a
c
y
Q
u
o
r
a
:
8
5
,
0
0
%
R
e
c
a
l
l
Q
u
o
r
a
:
9
2
,
0
0
%
S
TS w
i
t
h
S
i
a
m
e
se
n
e
u
r
a
l
n
e
t
w
o
r
k
s
R
a
n
a
s
i
n
g
h
e
e
t
a
l
.
[
1
5
]
a.
D
a
t
a
:
S
e
m
Ev
a
l
2
0
1
7
d
a
t
a
set
(
l
a
b
e
l
e
d
)
b.
D
o
ma
i
n
:
b
i
o
me
d
i
c
a
l
c.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
c
u
s
t
o
m
-
t
r
a
i
n
e
d
e
m
b
e
d
d
i
n
g
M
o
d
e
l
:
S
i
a
mes
e
M
a
n
h
a
t
t
a
n
–
LS
TM
A
c
c
u
r
a
c
y
:
8
6
,
5
1
%
D
e
t
e
c
t
i
o
n
o
f
m
e
d
i
c
a
l
t
e
x
t
sema
n
t
i
c
s
i
m
i
l
a
r
i
t
y
b
a
s
e
d
o
n
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
Zh
e
n
g
e
t
a
l
.
[
2
0
]
a.
D
a
t
a
:
i
ma
g
i
n
g
a
n
d
p
a
t
h
o
l
o
g
y
r
e
p
o
r
t
-
p
a
i
r
s
(
l
a
b
e
l
e
d
)
b.
D
o
ma
i
n
:
b
i
o
me
d
i
c
a
l
c.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
C
M
ESH
M
o
d
e
l
:
S
i
a
mes
e
n
e
u
r
a
l
n
e
t
w
o
r
k
–
C
N
N
u
si
n
g
LI
M
E
a
l
g
o
r
i
t
h
m f
o
r
s
i
mi
l
a
r
i
t
y
o
u
t
p
u
t
R
e
c
a
l
l
:
9
3
,
7
0
%
P
r
e
c
i
s
i
o
n
:
9
4
,
5
0
%
F1
-
S
c
o
r
e
:
9
4
,
1
0
%
A
S
i
a
m
e
se
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
l
e
a
r
n
i
n
g
C
h
i
n
e
s
e
se
n
t
e
n
c
e
si
mi
l
a
r
i
t
y
S
h
i
e
t
a
l
.
[
1
7
]
a.
D
a
t
a
:
C
h
i
n
e
se
se
n
t
e
n
c
e
p
a
i
r
s (l
a
b
e
l
e
d
)
b.
D
o
ma
i
n
:
g
e
n
e
r
a
l
c.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
c
u
s
t
o
m
-
t
r
a
i
n
e
d
e
m
b
e
d
d
i
n
g
M
o
d
e
l
:
S
i
a
m
e
se
n
e
u
r
a
l
n
e
t
w
o
r
k
–
C
N
N
w
i
t
h
t
w
o
d
i
f
f
e
r
e
n
t
me
t
r
i
c
d
i
s
t
a
n
c
e
s
(
C
o
si
n
e
a
n
d
M
a
n
h
a
t
t
a
n
)
A
c
c
u
r
a
c
y
C
o
s
i
n
e
:
7
7
,
0
5
%
A
c
c
u
r
a
c
y
M
a
n
h
a
t
t
a
n
:
7
7
,
3
1
%
S
i
a
mes
e
K
G
–
LS
TM
:
a
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
f
o
r
e
n
r
i
c
h
i
n
g
U
M
LS
M
e
t
a
t
h
e
sa
u
r
u
s
sy
n
o
n
y
my
Tr
a
n
e
t
a
l
.
[
2
1
]
a.
D
a
t
a
:
t
h
e
U
M
LS
d
a
t
a
se
t
(
l
a
b
e
l
e
d
)
b.
D
o
ma
i
n
:
b
i
o
me
d
i
s
c.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
B
i
o
W
o
r
d
V
e
c
M
o
d
e
l
:
S
i
a
mes
e
M
a
n
h
a
t
t
a
n
-
LSTM
w
i
t
h
k
n
o
w
l
e
d
g
e
g
r
a
p
h
A
c
c
u
r
a
c
y
:
9
8
,
2
3
%
R
e
c
a
l
l
:
9
3
,
8
7
%
P
r
e
c
i
s
i
o
n
:
9
7
,
8
6
%
F1
-
S
c
o
r
e
:
9
6
,
4
1
%
S
i
mi
l
a
r
i
t
y
ma
t
c
h
i
n
g
o
f
me
d
i
c
a
l
q
u
e
st
i
o
n
s
b
a
se
d
o
n
S
i
a
mese
n
e
t
w
o
r
k
Li
a
n
d
H
e
[
2
2
]
a.
D
a
t
a
:
e
t
h
n
i
c
me
d
i
c
a
l
Q
u
e
s
t
i
o
n
d
a
t
a
se
t
(
l
a
b
e
l
e
d
)
b.
D
o
ma
i
n
:
B
i
o
me
d
i
s
c.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
W
o
r
d
2
V
e
c
d.
M
o
d
e
l
:
S
i
a
mes
e
n
e
u
r
a
l
n
e
t
w
o
r
k
–
R
N
N
w
i
t
h
M
a
n
h
a
t
t
a
n
d
i
s
t
a
n
c
e
F1
-
S
c
o
r
e
E
u
c
l
i
d
e
a
n
:
9
4
,
1
3
%
F1
-
S
c
o
r
e
C
o
s
i
n
e
:
9
6
,
9
3
%
F1
-
S
c
o
r
e
M
a
n
h
a
t
t
a
n
:
9
7
,
3
4
%
P
r
o
p
o
se
d
m
e
t
h
o
d
K
u
r
n
i
a
sar
i
e
t
a
l
.
a.
D
a
t
a
:
G
EN
I
A
e
v
e
n
t
d
a
t
a
se
t
(
u
n
l
a
b
e
l
e
d
)
b.
La
b
e
l
i
n
g
:
C
S
a
n
d
W
M
D
c.
W
o
r
d
e
m
b
e
d
d
i
n
g
:
B
i
o
W
o
r
d
V
e
c
M
o
d
e
l
:
S
i
a
mes
e
M
a
n
h
a
t
t
a
n
-
h
y
b
r
i
d
L
S
TM
C
N
N
A
c
c
u
r
a
c
y
:
1
0
0
%
R
e
c
a
l
l
:
1
0
0
%
P
r
e
c
i
s
i
o
n
:
1
0
0
%
F1
-
S
c
o
r
e
:
1
0
0
%
5.
CO
NCLU
SI
O
N
T
h
is
r
esear
ch
id
en
tifie
s
th
e
o
p
tim
al
m
o
d
el
f
o
r
d
etec
tin
g
s
em
an
tic
s
im
ilar
ity
am
o
n
g
leu
k
e
m
ia
-
r
elate
d
b
io
m
ed
ical
ter
m
s
,
a
h
y
b
r
id
L
STM
-
C
NN
m
o
d
el
c
o
m
b
in
in
g
Si
am
ese
Ma
n
h
attan
an
d
L
STM
-
C
NN
ar
ch
itectu
r
es.
R
esu
lts
in
d
icate
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
m
o
d
els
p
r
o
p
o
s
ed
in
p
r
ev
io
u
s
r
esear
ch
d
esp
ite
d
if
f
er
in
g
d
o
m
ain
s
b
etwe
en
s
tu
d
ies.
T
h
e
r
esear
ch
aim
s
to
aid
r
esear
ch
er
s
an
d
m
ed
ical
p
r
o
f
ess
io
n
als
in
u
n
co
v
er
i
n
g
co
r
r
elatio
n
s
b
etwe
en
v
ar
io
u
s
b
io
m
ed
ical
ev
en
ts
an
d
leu
k
e
m
ia
,
u
n
d
er
s
tan
d
i
n
g
s
em
an
tic
co
n
n
ec
tio
n
s
b
etwe
en
tr
ig
g
er
wo
r
d
s
an
d
b
io
m
ed
ical
ev
en
ts
,
an
d
au
to
m
atica
lly
ex
tr
ac
tin
g
cr
itical
in
f
o
r
m
atio
n
f
r
o
m
b
io
m
ed
ical
tex
ts
to
en
h
an
ce
leu
k
em
ia
-
r
elate
d
i
n
f
o
r
m
atio
n
r
etr
iev
al.
Mo
r
eo
v
e
r
,
th
e
s
tu
d
y
s
ee
k
s
to
d
ee
p
en
u
n
d
er
s
tan
d
in
g
o
f
h
o
w
l
eu
k
em
ia
tr
ea
tm
en
ts
im
p
ac
t
s
p
ec
if
ic
b
io
lo
g
ical
p
ath
way
s
an
d
s
u
b
s
eq
u
en
t
b
io
m
ed
ical
p
r
o
ce
s
s
es.
Fu
tu
r
e
r
esear
ch
d
ir
ec
tio
n
s
co
u
ld
ex
p
lo
r
e
f
u
r
th
er
ad
v
an
ce
m
en
ts
in
th
e
STS ta
s
k
b
y
ex
tr
ac
tin
g
r
elatio
n
s
h
ip
s
b
etwe
en
b
i
o
m
ed
ical
en
titi
es,
th
er
eb
y
en
h
an
cin
g
in
s
ig
h
ts
in
to
leu
k
em
ia
a
n
d
its
b
io
m
ed
ical
im
p
licatio
n
s
.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
is
r
esear
ch
r
ec
eiv
ed
f
u
n
d
in
g
f
r
o
m
a
Facu
lty
o
f
Ma
th
em
atic
s
an
d
Natu
r
al
Scien
ce
s
g
r
a
n
t
f
o
r
th
e
2
0
2
3
ac
ad
em
ic
y
ea
r
.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
C
h
a
v
e
s
,
C
.
K
e
s
i
k
u
,
a
n
d
B
.
G
a
r
c
i
a
-
Za
p
i
r
a
i
n
,
“
A
u
t
o
m
a
t
i
c
t
e
x
t
su
mm
a
r
i
z
a
t
i
o
n
o
f
b
i
o
m
e
d
i
c
a
l
t
e
x
t
d
a
t
a
:
A
sy
st
e
ma
t
i
c
r
e
v
i
e
w
,
”
I
n
f
o
rm
a
t
i
o
n
,
v
o
l
.
1
3
,
n
o
.
8
,
A
u
g
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
i
n
f
o
1
3
0
8
0
3
9
3
.
[
2
]
Y
.
W
a
n
g
,
S
.
F
u
,
F
.
S
h
e
n
,
S
.
H
e
n
r
y
,
O
.
U
z
u
n
e
r
,
a
n
d
H
.
Li
u
,
“
T
h
e
2
0
1
9
n
2
c
2
/
O
H
N
LP
t
r
a
c
k
o
n
c
l
i
n
i
c
a
l
sem
a
n
t
i
c
t
e
x
t
u
a
l
si
m
i
l
a
r
i
t
y
:
O
v
e
r
v
i
e
w
,
”
J
MIR
Me
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
8
,
n
o
.
1
1
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
2
1
9
6
/
2
3
3
7
5
.
[
3
]
F
.
A
l
a
m
,
M
.
A
f
z
a
l
,
a
n
d
K
.
M
.
M
a
l
i
k
,
“
C
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
se
ma
n
t
i
c
s
i
mi
l
a
r
i
t
y
t
e
c
h
n
i
q
u
e
s
f
o
r
me
d
i
c
a
l
t
e
x
t
,
”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
f
o
rm
a
t
i
o
n
N
e
t
w
o
r
k
i
n
g
(
I
C
O
I
N
)
,
J
a
n
.
2
0
2
0
,
v
o
l
.
2
0
2
0
-
Ja
n
u
a
,
p
p
.
1
0
6
–
109
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
O
I
N
4
8
6
5
6
.
2
0
2
0
.
9
0
1
6
5
7
4
.
[
4
]
Q
.
C
h
e
n
,
J.
D
u
,
S
.
K
i
m,
W
.
J.
W
i
l
b
u
r
,
a
n
d
Z.
L
u
,
“
D
e
e
p
l
e
a
r
n
i
n
g
w
i
t
h
se
n
t
e
n
c
e
e
m
b
e
d
d
i
n
g
s
p
r
e
-
t
r
a
i
n
e
d
o
n
b
i
o
me
d
i
c
a
l
c
o
r
p
o
r
a
i
mp
r
o
v
e
s
t
h
e
p
e
r
f
o
r
ma
n
c
e
o
f
f
i
n
d
i
n
g
si
mi
l
a
r
s
e
n
t
e
n
c
e
s
i
n
e
l
e
c
t
r
o
n
i
c
me
d
i
c
a
l
r
e
c
o
r
d
s
,
”
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
s
i
o
n
Ma
k
i
n
g
,
v
o
l
.
2
0
,
n
o
.
S
1
,
A
p
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
8
6
/
s1
2
9
1
1
-
0
2
0
-
1
0
4
4
-
0.
Evaluation Warning : The document was created with Spire.PDF for Python.